240521_gpt_jargon_identify_eval.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import openai\n",
    "import numpy as np\n",
    "import json\n",
    "import os\n",
    "from src.responses import completion_to_dict\n",
    "from dotenv import load_dotenv\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "sns.set(style=\"whitegrid\")\n",
    "\n",
    "load_dotenv()\n",
    "\n",
    "# Set the seed for reproducibility\n",
    "SEED = 10\n",
    "\n",
    "\n",
    "# Today's date for the filename\n",
    "from datetime import datetime\n",
    "today = datetime.today().strftime('%Y%m%d')[2:]\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Input"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>arxiv_id</th>\n",
       "      <th>url</th>\n",
       "      <th>title</th>\n",
       "      <th>summary</th>\n",
       "      <th>updated</th>\n",
       "      <th>published</th>\n",
       "      <th>authors</th>\n",
       "      <th>comments</th>\n",
       "      <th>categories</th>\n",
       "      <th>primary_category</th>\n",
       "      <th>doi</th>\n",
       "      <th>journal_ref</th>\n",
       "      <th>peer_reviewed</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2403.16190v1</td>\n",
       "      <td>http://arxiv.org/abs/2403.16190v1</td>\n",
       "      <td>Logic-based Explanations for Linear Support Ve...</td>\n",
       "      <td>Support Vector Classifier (SVC) is a well-know...</td>\n",
       "      <td>1711293284000</td>\n",
       "      <td>1711293284000</td>\n",
       "      <td>[Francisco Mateus Rocha Filho, Thiago Alves Ro...</td>\n",
       "      <td>16 pages, submitted to BRACIS 2023 (Brazilian ...</td>\n",
       "      <td>[cs.AI, cs.LG, cs.LO, I.2.4; I.2.6]</td>\n",
       "      <td>cs.AI</td>\n",
       "      <td>10.1007/978-3-031-45368-7_10</td>\n",
       "      <td>None</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2307.05300v4</td>\n",
       "      <td>http://arxiv.org/abs/2307.05300v4</td>\n",
       "      <td>Unleashing the Emergent Cognitive Synergy in L...</td>\n",
       "      <td>Human intelligence thrives on cognitive synerg...</td>\n",
       "      <td>1711463553000</td>\n",
       "      <td>1689086719000</td>\n",
       "      <td>[Zhenhailong Wang, Shaoguang Mao, Wenshan Wu, ...</td>\n",
       "      <td>Accepted as a main conference paper at NAACL 2024</td>\n",
       "      <td>[cs.AI, cs.CL]</td>\n",
       "      <td>cs.AI</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2403.16750v1</td>\n",
       "      <td>http://arxiv.org/abs/2403.16750v1</td>\n",
       "      <td>All Artificial, Less Intelligence: GenAI throu...</td>\n",
       "      <td>Modern hardware designs have grown increasingl...</td>\n",
       "      <td>1711373004000</td>\n",
       "      <td>1711373004000</td>\n",
       "      <td>[Deepak Narayan Gadde, Aman Kumar, Thomas Nala...</td>\n",
       "      <td>Published in DVCon U.S. 2024</td>\n",
       "      <td>[cs.AI]</td>\n",
       "      <td>cs.AI</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2311.10112v2</td>\n",
       "      <td>http://arxiv.org/abs/2311.10112v2</td>\n",
       "      <td>zrLLM: Zero-Shot Relational Learning on Tempor...</td>\n",
       "      <td>Modeling evolving knowledge over temporal know...</td>\n",
       "      <td>1710517087000</td>\n",
       "      <td>1700083515000</td>\n",
       "      <td>[Zifeng Ding, Heling Cai, Jingpei Wu, Yunpu Ma...</td>\n",
       "      <td>Accepted to NAACL 2024 main conference</td>\n",
       "      <td>[cs.AI, cs.CL, cs.LG]</td>\n",
       "      <td>cs.AI</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2310.08992v3</td>\n",
       "      <td>http://arxiv.org/abs/2310.08992v3</td>\n",
       "      <td>CodeChain: Towards Modular Code Generation Thr...</td>\n",
       "      <td>Large Language Models (LLMs) have already beco...</td>\n",
       "      <td>1710386949000</td>\n",
       "      <td>1697192268000</td>\n",
       "      <td>[Hung Le, Hailin Chen, Amrita Saha, Akash Goku...</td>\n",
       "      <td>Accepted to ICLR 2024</td>\n",
       "      <td>[cs.AI, cs.CL, cs.PL]</td>\n",
       "      <td>cs.AI</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       arxiv_id                                url  \\\n",
       "0  2403.16190v1  http://arxiv.org/abs/2403.16190v1   \n",
       "1  2307.05300v4  http://arxiv.org/abs/2307.05300v4   \n",
       "2  2403.16750v1  http://arxiv.org/abs/2403.16750v1   \n",
       "3  2311.10112v2  http://arxiv.org/abs/2311.10112v2   \n",
       "4  2310.08992v3  http://arxiv.org/abs/2310.08992v3   \n",
       "\n",
       "                                               title  \\\n",
       "0  Logic-based Explanations for Linear Support Ve...   \n",
       "1  Unleashing the Emergent Cognitive Synergy in L...   \n",
       "2  All Artificial, Less Intelligence: GenAI throu...   \n",
       "3  zrLLM: Zero-Shot Relational Learning on Tempor...   \n",
       "4  CodeChain: Towards Modular Code Generation Thr...   \n",
       "\n",
       "                                             summary        updated  \\\n",
       "0  Support Vector Classifier (SVC) is a well-know...  1711293284000   \n",
       "1  Human intelligence thrives on cognitive synerg...  1711463553000   \n",
       "2  Modern hardware designs have grown increasingl...  1711373004000   \n",
       "3  Modeling evolving knowledge over temporal know...  1710517087000   \n",
       "4  Large Language Models (LLMs) have already beco...  1710386949000   \n",
       "\n",
       "       published                                            authors  \\\n",
       "0  1711293284000  [Francisco Mateus Rocha Filho, Thiago Alves Ro...   \n",
       "1  1689086719000  [Zhenhailong Wang, Shaoguang Mao, Wenshan Wu, ...   \n",
       "2  1711373004000  [Deepak Narayan Gadde, Aman Kumar, Thomas Nala...   \n",
       "3  1700083515000  [Zifeng Ding, Heling Cai, Jingpei Wu, Yunpu Ma...   \n",
       "4  1697192268000  [Hung Le, Hailin Chen, Amrita Saha, Akash Goku...   \n",
       "\n",
       "                                            comments  \\\n",
       "0  16 pages, submitted to BRACIS 2023 (Brazilian ...   \n",
       "1  Accepted as a main conference paper at NAACL 2024   \n",
       "2                       Published in DVCon U.S. 2024   \n",
       "3             Accepted to NAACL 2024 main conference   \n",
       "4                              Accepted to ICLR 2024   \n",
       "\n",
       "                            categories primary_category  \\\n",
       "0  [cs.AI, cs.LG, cs.LO, I.2.4; I.2.6]            cs.AI   \n",
       "1                       [cs.AI, cs.CL]            cs.AI   \n",
       "2                              [cs.AI]            cs.AI   \n",
       "3                [cs.AI, cs.CL, cs.LG]            cs.AI   \n",
       "4                [cs.AI, cs.CL, cs.PL]            cs.AI   \n",
       "\n",
       "                            doi journal_ref  peer_reviewed  \n",
       "0  10.1007/978-3-031-45368-7_10        None           True  \n",
       "1                          None        None           True  \n",
       "2                          None        None           True  \n",
       "3                          None        None           True  \n",
       "4                          None        None           True  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Read in the arXiv metadata\n",
    "with open(\"data/arxiv_metadata/filtered/march_2024_ai_hc_cy_peer_reviewed_sampled.json\") as json_data:\n",
    "    metadata = json.load(json_data)\n",
    "    json_data.close()\n",
    "\n",
    "# Convert JSON to DataFrame\n",
    "metadata_df = pd.DataFrame.from_dict(metadata, orient='index')\n",
    "\n",
    "metadata_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Light text cleaning so that it can be proprely fed into the model\n",
    "metadata_df['summary'] = metadata_df['summary'].str.replace('\\n', ' ')\n",
    "metadata_df['summary'] = metadata_df['summary'].str.replace('\\r', ' ')\n",
    "metadata_df['summary'] = metadata_df['summary'].str.replace('\\t', ' ')\n",
    "metadata_df['summary'] = metadata_df['summary'].str.replace('  ', ' ')\n",
    "metadata_df['summary'] = metadata_df['summary'].str.strip()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>arxiv_id</th>\n",
       "      <th>url</th>\n",
       "      <th>title</th>\n",
       "      <th>summary</th>\n",
       "      <th>updated</th>\n",
       "      <th>published</th>\n",
       "      <th>authors</th>\n",
       "      <th>comments</th>\n",
       "      <th>categories</th>\n",
       "      <th>primary_category</th>\n",
       "      <th>doi</th>\n",
       "      <th>journal_ref</th>\n",
       "      <th>peer_reviewed</th>\n",
       "      <th>primary_category_readable</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2403.16190v1</td>\n",
       "      <td>http://arxiv.org/abs/2403.16190v1</td>\n",
       "      <td>Logic-based Explanations for Linear Support Ve...</td>\n",
       "      <td>Support Vector Classifier (SVC) is a well-know...</td>\n",
       "      <td>1711293284000</td>\n",
       "      <td>1711293284000</td>\n",
       "      <td>[Francisco Mateus Rocha Filho, Thiago Alves Ro...</td>\n",
       "      <td>16 pages, submitted to BRACIS 2023 (Brazilian ...</td>\n",
       "      <td>[cs.AI, cs.LG, cs.LO, I.2.4; I.2.6]</td>\n",
       "      <td>cs.AI</td>\n",
       "      <td>10.1007/978-3-031-45368-7_10</td>\n",
       "      <td>None</td>\n",
       "      <td>True</td>\n",
       "      <td>Artificial Intelligence</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2307.05300v4</td>\n",
       "      <td>http://arxiv.org/abs/2307.05300v4</td>\n",
       "      <td>Unleashing the Emergent Cognitive Synergy in L...</td>\n",
       "      <td>Human intelligence thrives on cognitive synerg...</td>\n",
       "      <td>1711463553000</td>\n",
       "      <td>1689086719000</td>\n",
       "      <td>[Zhenhailong Wang, Shaoguang Mao, Wenshan Wu, ...</td>\n",
       "      <td>Accepted as a main conference paper at NAACL 2024</td>\n",
       "      <td>[cs.AI, cs.CL]</td>\n",
       "      <td>cs.AI</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>True</td>\n",
       "      <td>Artificial Intelligence</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2403.16750v1</td>\n",
       "      <td>http://arxiv.org/abs/2403.16750v1</td>\n",
       "      <td>All Artificial, Less Intelligence: GenAI throu...</td>\n",
       "      <td>Modern hardware designs have grown increasingl...</td>\n",
       "      <td>1711373004000</td>\n",
       "      <td>1711373004000</td>\n",
       "      <td>[Deepak Narayan Gadde, Aman Kumar, Thomas Nala...</td>\n",
       "      <td>Published in DVCon U.S. 2024</td>\n",
       "      <td>[cs.AI]</td>\n",
       "      <td>cs.AI</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>True</td>\n",
       "      <td>Artificial Intelligence</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       arxiv_id                                url  \\\n",
       "0  2403.16190v1  http://arxiv.org/abs/2403.16190v1   \n",
       "1  2307.05300v4  http://arxiv.org/abs/2307.05300v4   \n",
       "2  2403.16750v1  http://arxiv.org/abs/2403.16750v1   \n",
       "\n",
       "                                               title  \\\n",
       "0  Logic-based Explanations for Linear Support Ve...   \n",
       "1  Unleashing the Emergent Cognitive Synergy in L...   \n",
       "2  All Artificial, Less Intelligence: GenAI throu...   \n",
       "\n",
       "                                             summary        updated  \\\n",
       "0  Support Vector Classifier (SVC) is a well-know...  1711293284000   \n",
       "1  Human intelligence thrives on cognitive synerg...  1711463553000   \n",
       "2  Modern hardware designs have grown increasingl...  1711373004000   \n",
       "\n",
       "       published                                            authors  \\\n",
       "0  1711293284000  [Francisco Mateus Rocha Filho, Thiago Alves Ro...   \n",
       "1  1689086719000  [Zhenhailong Wang, Shaoguang Mao, Wenshan Wu, ...   \n",
       "2  1711373004000  [Deepak Narayan Gadde, Aman Kumar, Thomas Nala...   \n",
       "\n",
       "                                            comments  \\\n",
       "0  16 pages, submitted to BRACIS 2023 (Brazilian ...   \n",
       "1  Accepted as a main conference paper at NAACL 2024   \n",
       "2                       Published in DVCon U.S. 2024   \n",
       "\n",
       "                            categories primary_category  \\\n",
       "0  [cs.AI, cs.LG, cs.LO, I.2.4; I.2.6]            cs.AI   \n",
       "1                       [cs.AI, cs.CL]            cs.AI   \n",
       "2                              [cs.AI]            cs.AI   \n",
       "\n",
       "                            doi journal_ref  peer_reviewed  \\\n",
       "0  10.1007/978-3-031-45368-7_10        None           True   \n",
       "1                          None        None           True   \n",
       "2                          None        None           True   \n",
       "\n",
       "  primary_category_readable  \n",
       "0   Artificial Intelligence  \n",
       "1   Artificial Intelligence  \n",
       "2   Artificial Intelligence  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Add in a column mapping the primary category to a more human readable format\n",
    "category_map = {\n",
    "    'cs.AI': 'Artificial Intelligence',\n",
    "    'cs.CY': 'Computers and Society',\n",
    "    'cs.HC': 'Human-Computer Interaction'\n",
    "}\n",
    "\n",
    "metadata_df['primary_category_readable'] = metadata_df['primary_category'].map(category_map)\n",
    "\n",
    "metadata_df.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read in the prompt templates for this task. These will need to be converted to f-strings.\n",
    "\n",
    "with open(\"data/llm_prompts/jargon_identifyv2.json\") as json_data:\n",
    "    jargon_prompts = json.load(json_data)\n",
    "    json_data.close()\n",
    "\n",
    "# Preview\n",
    "jargon_prompts\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read in reader profiles\n",
    "with open('data/llm_prompts/jargon_readersv2.json', 'r') as file:\n",
    "    readers = json.load(file)\n",
    "\n",
    "# Output the data\n",
    "for reader in readers:\n",
    "    print(f\"Description: {reader['reader_description']}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# GPT4 Setup\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from openai import OpenAI\n",
    "\n",
    "# Setup client\n",
    "client = OpenAI(\n",
    "    api_key = os.getenv(\"OPENAI_KEY\"),\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Model and decoding params\n",
    "GPT_MODEL = \"gpt-4-turbo\"\n",
    "MAX_TOKENS = 512\n",
    "TEMPERATURE = 1.0\n",
    "\n",
    "# Extract prompts from the dict\n",
    "system_prompt = jargon_prompts['system_prompt']\n",
    "query_prompt = jargon_prompts['query_prompt']\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "jp-MarkdownHeadingCollapsed": true
   },
   "source": [
    "# Calling the API"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(system_prompt)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%time\n",
    "# Iterate over the abstracts in thr DF + the reader profiles to generate the prompts and responses\n",
    "\n",
    "# Initialize a list to store the results\n",
    "results = []\n",
    "\n",
    "# Iterate over the metadata -- try two rows first\n",
    "for idx, row in metadata_df.iterrows():\n",
    "\n",
    "    # Extract abstract metadata\n",
    "    arxiv_id = row['arxiv_id']\n",
    "    abstract_text = row['summary']\n",
    "    primary_category_readable = row['primary_category_readable']\n",
    "\n",
    "    # Iterate over the reader profiles\n",
    "    for reader in readers:\n",
    "\n",
    "        # Extract reader ID\n",
    "        reader_id = reader['reader_id']\n",
    "\n",
    "        # Print everything extracted so far\n",
    "        # print(f\"arXiv ID: {arxiv_id}\")\n",
    "        # print(f\"Abstract: {abstract_text}\")\n",
    "        # print(f\"Primary Category: {primary_category_readable}\")\n",
    "        # print(f\"Reader ID: {reader_id}\")\n",
    "\n",
    "        # Edit the system prompt once more\n",
    "        filled_system_prompt = system_prompt.format(\n",
    "            reader_description=reader['reader_description'],\n",
    "            # reader_expertise=reader['reader_expertise'], \n",
    "            primary_category_readable=primary_category_readable\n",
    "        )\n",
    "\n",
    "        # Fill in the query prompt\n",
    "        filled_query_prompt = query_prompt.format(\n",
    "            abstract_text=abstract_text\n",
    "        )\n",
    "\n",
    "        # Create the messages\n",
    "        messages = [\n",
    "            {\"role\": \"system\", \"content\": filled_system_prompt},\n",
    "            {\"role\": \"user\", \"content\": filled_query_prompt},\n",
    "        ]\n",
    "\n",
    "        # Generate the completion\n",
    "        completion = client.chat.completions.create(\n",
    "            model=GPT_MODEL,\n",
    "            seed=SEED,\n",
    "            max_tokens=MAX_TOKENS,\n",
    "            temperature=TEMPERATURE,\n",
    "            messages=messages\n",
    "        )\n",
    "\n",
    "        # Convert the response to a dictionary\n",
    "        completion_dict = completion_to_dict(completion)\n",
    "        \n",
    "        # Store the result\n",
    "        results.append({\n",
    "            'arxiv_id': arxiv_id,\n",
    "            'reader_id': reader_id,\n",
    "            'messages': messages,\n",
    "            'completion': completion_dict\n",
    "        })\n",
    "        \n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Today's date for the filename\n",
    "from datetime import datetime\n",
    "today = datetime.today().strftime('%Y%m%d')[2:]\n",
    "\n",
    "# Save to JSON with indenting\n",
    "with open(f'data/llm_outputs/{today}_march_2024_ai_hc_cy_peer_reviewed_sampled_jargon_list.json', 'w') as file:\n",
    "    json.dump(results, file, indent=4)\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Evaluation of the Jargon Identification\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Some basic data processing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read in the results\n",
    "with open(\"data/llm_outputs/240525_march_2024_ai_hc_cy_peer_reviewed_sampled_jargon_list.json\", \"r\") as file:\n",
    "    results = json.load(file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>arxiv_id</th>\n",
       "      <th>reader_id</th>\n",
       "      <th>gpt4_jargon_list</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2403.16190v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>reject option strategy, formal guarantees, min...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2403.16190v1</td>\n",
       "      <td>rid1</td>\n",
       "      <td>linear classification problems, reject option ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2307.05300v4</td>\n",
       "      <td>rid0</td>\n",
       "      <td>cognitive synergist, fine-grained personas, fa...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2307.05300v4</td>\n",
       "      <td>rid1</td>\n",
       "      <td>cognitive synergy, cognitive synergist, multi-...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2403.16750v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>Common Weakness Enumerations (CWEs), SystemVer...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       arxiv_id reader_id                                   gpt4_jargon_list\n",
       "0  2403.16190v1      rid0  reject option strategy, formal guarantees, min...\n",
       "1  2403.16190v1      rid1  linear classification problems, reject option ...\n",
       "2  2307.05300v4      rid0  cognitive synergist, fine-grained personas, fa...\n",
       "3  2307.05300v4      rid1  cognitive synergy, cognitive synergist, multi-...\n",
       "4  2403.16750v1      rid0  Common Weakness Enumerations (CWEs), SystemVer..."
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Make result into a df that we like\n",
    "results_df = [] \n",
    "\n",
    "for result in results:\n",
    "\n",
    "    results_df.append(\n",
    "        [result['arxiv_id'], result['reader_id'], result['completion']['content']]\n",
    "    )\n",
    "\n",
    "results_df = pd.DataFrame(results_df, columns=['arxiv_id', 'reader_id', 'gpt4_jargon_list'])\n",
    "results_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>arxiv_id</th>\n",
       "      <th>reader_id</th>\n",
       "      <th>human_jargon_list</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2403.16190v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>correctness,minimality,Anchors</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2307.05300v4</td>\n",
       "      <td>rid0</td>\n",
       "      <td>multi-turn,persona</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2403.16750v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>common weakeness enumerations,SystemVerilog,Re...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2311.10112v2</td>\n",
       "      <td>rid0</td>\n",
       "      <td>temporal knowledge graphs,embedding-based TKGF...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2310.08992v3</td>\n",
       "      <td>rid0</td>\n",
       "      <td>pass@1,APPS,CodeContests</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       arxiv_id reader_id                                  human_jargon_list\n",
       "0  2403.16190v1      rid0                     correctness,minimality,Anchors\n",
       "1  2307.05300v4      rid0                                 multi-turn,persona\n",
       "2  2403.16750v1      rid0  common weakeness enumerations,SystemVerilog,Re...\n",
       "3  2311.10112v2      rid0  temporal knowledge graphs,embedding-based TKGF...\n",
       "4  2310.08992v3      rid0                           pass@1,APPS,CodeContests"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Read in the human annotations\n",
    "rid_0_jargon = pd.read_csv(\"data/human_annotations/march_2024_ai_hc_cy_peer_reviewed_sampled_jargon_list_rid0.csv\")\n",
    "rid_1_jargon = pd.read_csv(\"data/human_annotations/march_2024_ai_hc_cy_peer_reviewed_sampled_jargon_list_rid1.csv\")\n",
    "\n",
    "# Stack these two with only the relevant columns\n",
    "rid_0_jargon['reader_id'] = 'rid0'\n",
    "rid_1_jargon['reader_id'] = 'rid1'\n",
    "human_annotations = pd.concat([rid_0_jargon, rid_1_jargon], axis=0)[['arxiv_id', 'reader_id', 'jargon terms']]\n",
    "human_annotations.rename(columns={'jargon terms': 'human_jargon_list'}, inplace=True)\n",
    "\n",
    "human_annotations.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((128, 3), (128, 3))"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results_df.shape, human_annotations.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Merge the human annotations with the LLM outputs\n",
    "merged_humam_llm_jargon = pd.merge(\n",
    "    results_df, human_annotations, on=['arxiv_id', 'reader_id'], how='inner')\n",
    "merged_humam_llm_jargon.fillna(\"\", inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>arxiv_id</th>\n",
       "      <th>reader_id</th>\n",
       "      <th>gpt4_jargon_list</th>\n",
       "      <th>human_jargon_list</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2403.16190v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>reject option strategy, formal guarantees, min...</td>\n",
       "      <td>correctness,minimality,Anchors</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2403.16190v1</td>\n",
       "      <td>rid1</td>\n",
       "      <td>linear classification problems, reject option ...</td>\n",
       "      <td>reject option strategy,Anchors,heuristic algor...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2307.05300v4</td>\n",
       "      <td>rid0</td>\n",
       "      <td>cognitive synergist, fine-grained personas, fa...</td>\n",
       "      <td>multi-turn,persona</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2307.05300v4</td>\n",
       "      <td>rid1</td>\n",
       "      <td>cognitive synergy, cognitive synergist, multi-...</td>\n",
       "      <td>Solo Performance Prompting,multi-turn,Chain-of...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2403.16750v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>Common Weakness Enumerations (CWEs), SystemVer...</td>\n",
       "      <td>common weakeness enumerations,SystemVerilog,Re...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>123</th>\n",
       "      <td>2305.11927v2</td>\n",
       "      <td>rid1</td>\n",
       "      <td>Computer Vision (CV) models, model opacity, In...</td>\n",
       "      <td>Sprite</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>124</th>\n",
       "      <td>2403.00632v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>affective mindfulness, emotional arc, visual m...</td>\n",
       "      <td>affective interface</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>125</th>\n",
       "      <td>2403.00632v1</td>\n",
       "      <td>rid1</td>\n",
       "      <td>affective interface, emotional arc, metaphoric...</td>\n",
       "      <td>Metamorpheus</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>126</th>\n",
       "      <td>2309.15723v2</td>\n",
       "      <td>rid0</td>\n",
       "      <td>data storytelling tools, storytelling workflow...</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>127</th>\n",
       "      <td>2309.15723v2</td>\n",
       "      <td>rid1</td>\n",
       "      <td>data storytelling tools, human-AI collaboratio...</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>128 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         arxiv_id reader_id  \\\n",
       "0    2403.16190v1      rid0   \n",
       "1    2403.16190v1      rid1   \n",
       "2    2307.05300v4      rid0   \n",
       "3    2307.05300v4      rid1   \n",
       "4    2403.16750v1      rid0   \n",
       "..            ...       ...   \n",
       "123  2305.11927v2      rid1   \n",
       "124  2403.00632v1      rid0   \n",
       "125  2403.00632v1      rid1   \n",
       "126  2309.15723v2      rid0   \n",
       "127  2309.15723v2      rid1   \n",
       "\n",
       "                                      gpt4_jargon_list  \\\n",
       "0    reject option strategy, formal guarantees, min...   \n",
       "1    linear classification problems, reject option ...   \n",
       "2    cognitive synergist, fine-grained personas, fa...   \n",
       "3    cognitive synergy, cognitive synergist, multi-...   \n",
       "4    Common Weakness Enumerations (CWEs), SystemVer...   \n",
       "..                                                 ...   \n",
       "123  Computer Vision (CV) models, model opacity, In...   \n",
       "124  affective mindfulness, emotional arc, visual m...   \n",
       "125  affective interface, emotional arc, metaphoric...   \n",
       "126  data storytelling tools, storytelling workflow...   \n",
       "127  data storytelling tools, human-AI collaboratio...   \n",
       "\n",
       "                                     human_jargon_list  \n",
       "0                       correctness,minimality,Anchors  \n",
       "1    reject option strategy,Anchors,heuristic algor...  \n",
       "2                                   multi-turn,persona  \n",
       "3    Solo Performance Prompting,multi-turn,Chain-of...  \n",
       "4    common weakeness enumerations,SystemVerilog,Re...  \n",
       "..                                                 ...  \n",
       "123                                             Sprite  \n",
       "124                                affective interface  \n",
       "125                                       Metamorpheus  \n",
       "126                                                     \n",
       "127                                                     \n",
       "\n",
       "[128 rows x 4 columns]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "merged_humam_llm_jargon"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save merged_humam_llm_jargon to JSON with indenting\n",
    "merged_humam_llm_jargon.to_json(\n",
    "    \"data/llm_outputs/240525_march_2024_human_llm_jargon_merged.json\", \n",
    "    orient='index', \n",
    "    indent=4\n",
    ")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Actual evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create different functions for jaccard\n",
    "from thefuzz import fuzz"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- We will use token set ratio to evaluate similarity of terms\n",
    "- After it takes out the common strings , it finds out the fuzz ratio for the following pairs and then returns the maximum value amongst the three:\n",
    "    - common string and the common string with the remainder of string one\n",
    "    - common string and the common string with the remainder of string two\n",
    "    - common string with the remainder of one and common string with the remainder of two"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 21.7 ms, sys: 2.62 ms, total: 24.3 ms\n",
      "Wall time: 22.8 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "# Set a threshold for the distance\n",
    "threshold = 90\n",
    "\n",
    "# Empty list of stuff\n",
    "jargon_identify_eval = []\n",
    "\n",
    "# Iterate and operate\n",
    "for idx, row in merged_humam_llm_jargon.iterrows():\n",
    "\n",
    "    # Get both lists\n",
    "    gpt_list = row['gpt4_jargon_list']\n",
    "    human_list = row['human_jargon_list']\n",
    "    # Convert to actual lists\n",
    "    human_list = [i.strip() for i in human_list.split(',') if i.strip()]\n",
    "    gpt_list = [i.strip() for i in gpt_list.split(',')]\n",
    "\n",
    "    # Three things will be calculated -- fuzzy true pos, fuzzy false pos, fuzzy false neg\n",
    "    fuzzy_true_positives = 0\n",
    "    fuzzy_false_positives = 0\n",
    "    fuzzy_false_negatives = 0\n",
    "\n",
    "    # # First iterate over the predicted terms\n",
    "    # for pred in gpt_list:\n",
    "    #     if any(fuzz.token_set_ratio(pred, actual) >= threshold for actual in human_list):\n",
    "    #         fuzzy_true_positives += 1\n",
    "    #     else:\n",
    "    #         fuzzy_false_positives += 1\n",
    "\n",
    "    # for actual in human_list:\n",
    "    #     if not any(fuzz.token_set_ratio(pred, actual) >= threshold for pred in gpt_list):\n",
    "    #         fuzzy_false_negatives += 1\n",
    "\n",
    "    matched_ground_truth = set()  # Set to store matched ground truth terms\n",
    "    for pred in gpt_list:\n",
    "        found_match = False\n",
    "        for actual in human_list:\n",
    "            if fuzz.token_set_ratio(pred, actual) >= threshold:\n",
    "                if actual not in matched_ground_truth:\n",
    "                    fuzzy_true_positives += 1\n",
    "                    matched_ground_truth.add(actual)\n",
    "                    found_match = True\n",
    "                    break  # Break after finding a match to avoid double counting\n",
    "        if not found_match:\n",
    "            fuzzy_false_positives += 1\n",
    "\n",
    "    fuzzy_false_negatives = len(human_list) - len(matched_ground_truth)\n",
    "\n",
    "    precision = fuzzy_true_positives / len(gpt_list) if gpt_list else 0\n",
    "    recall = fuzzy_true_positives / len(human_list) if human_list else 0\n",
    "    f1_score = 2 * (precision * recall) / (precision + recall) if (precision + recall) else 0\n",
    "    f2_score = (5 * precision * recall) / (4 * precision + recall) if (precision + recall) else 0 #  F2 score puts more attention on minimizing false negatives\n",
    "\n",
    "    # Take care of the edge-cases\n",
    "    if not gpt_list and not human_list:\n",
    "        precision, recall, f1_score, f2_score = 1.0, 1.0, 1.0, 1.0  # Perfect score if both are empty\n",
    "    if not gpt_list:\n",
    "        precision, recall, f1_score, f2_score = 0.0, 0.0, 0.0, 0.0  # No predicted terms, precision and F1 are 0\n",
    "    if not human_list:\n",
    "        precision, recall, f1_score, f2_score = 0.0, 0.0, 0.0, 0.0   # No ground truth terms, recall and F1 are 0\n",
    "\n",
    "    # Append stuff\n",
    "    jargon_identify_eval.append(\n",
    "        {\n",
    "            'arxiv_id': row['arxiv_id'], \n",
    "            'reader_id': row['reader_id'], \n",
    "            'len_gpt_jargon': len(gpt_list), \n",
    "            'len_human_jargon': len(human_list), \n",
    "            'precision': precision, \n",
    "            'recall': recall, \n",
    "            'f1_score': f1_score, \n",
    "            'f2_score' : f2_score, \n",
    "            'fuzzy_true_positives': fuzzy_true_positives, \n",
    "            'fuzzy_false_positives': fuzzy_false_positives, \n",
    "            'fuzzy_false_negatives': fuzzy_false_negatives\n",
    "        }\n",
    "    )\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "with open(f'data/llm_outputs/{today}_march_2024_sampled_jargon_identify_eval_metrics.json', 'w') as file:\n",
    "    json.dump(jargon_identify_eval, file, indent=4)\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Analysing the Eval"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>arxiv_id</th>\n",
       "      <th>reader_id</th>\n",
       "      <th>gpt4_jargon_list</th>\n",
       "      <th>human_jargon_list</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>2403.17873v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>Social Transparency, epistemic injustice, unwa...</td>\n",
       "      <td>Social Transparency (ST) framework,W-question</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>2403.17873v1</td>\n",
       "      <td>rid1</td>\n",
       "      <td>Human-centered explainable AI (HCXAI), Social ...</td>\n",
       "      <td>Human-centered explainable AI,Social\\nTranspar...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        arxiv_id reader_id                                   gpt4_jargon_list  \\\n",
       "34  2403.17873v1      rid0  Social Transparency, epistemic injustice, unwa...   \n",
       "35  2403.17873v1      rid1  Human-centered explainable AI (HCXAI), Social ...   \n",
       "\n",
       "                                    human_jargon_list  \n",
       "34      Social Transparency (ST) framework,W-question  \n",
       "35  Human-centered explainable AI,Social\\nTranspar...  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "merged_humam_llm_jargon.loc[\n",
    "    merged_humam_llm_jargon['arxiv_id']==\"2403.17873v1\"\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>arxiv_id</th>\n",
       "      <th>reader_id</th>\n",
       "      <th>len_gpt_jargon</th>\n",
       "      <th>len_human_jargon</th>\n",
       "      <th>precision</th>\n",
       "      <th>recall</th>\n",
       "      <th>f1_score</th>\n",
       "      <th>f2_score</th>\n",
       "      <th>fuzzy_true_positives</th>\n",
       "      <th>fuzzy_false_positives</th>\n",
       "      <th>fuzzy_false_negatives</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2403.16190v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2403.16190v1</td>\n",
       "      <td>rid1</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>0.444444</td>\n",
       "      <td>0.555556</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2307.05300v4</td>\n",
       "      <td>rid0</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2307.05300v4</td>\n",
       "      <td>rid1</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>0.166667</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.222222</td>\n",
       "      <td>0.277778</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2403.16750v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>6</td>\n",
       "      <td>6</td>\n",
       "      <td>0.833333</td>\n",
       "      <td>0.833333</td>\n",
       "      <td>0.833333</td>\n",
       "      <td>0.833333</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       arxiv_id reader_id  len_gpt_jargon  len_human_jargon  precision  \\\n",
       "0  2403.16190v1      rid0               3                 3   0.333333   \n",
       "1  2403.16190v1      rid1               6                 3   0.333333   \n",
       "2  2307.05300v4      rid0               3                 2   0.000000   \n",
       "3  2307.05300v4      rid1               6                 3   0.166667   \n",
       "4  2403.16750v1      rid0               6                 6   0.833333   \n",
       "\n",
       "     recall  f1_score  f2_score  fuzzy_true_positives  fuzzy_false_positives  \\\n",
       "0  0.333333  0.333333  0.333333                     1                      2   \n",
       "1  0.666667  0.444444  0.555556                     2                      4   \n",
       "2  0.000000  0.000000  0.000000                     0                      3   \n",
       "3  0.333333  0.222222  0.277778                     1                      5   \n",
       "4  0.833333  0.833333  0.833333                     5                      1   \n",
       "\n",
       "   fuzzy_false_negatives  \n",
       "0                      2  \n",
       "1                      1  \n",
       "2                      2  \n",
       "3                      2  \n",
       "4                      1  "
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Convert to DF\n",
    "jargon_identify_eval_df = pd.DataFrame(jargon_identify_eval)\n",
    "jargon_identify_eval_df.head()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Counts of Jargon Terms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get the difference of the jargon lengths\n",
    "jargon_identify_eval_df['len_jargon_diff'] = jargon_identify_eval_df['len_gpt_jargon'] - jargon_identify_eval_df['len_human_jargon']\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Sort the DataFrame by len_gpt_jargon and len_human_jargon\n",
    "# df = jargon_identify_eval_df.sort_values(['len_gpt_jargon', 'len_human_jargon'])\n",
    "\n",
    "# plt.figure(figsize=(12, 6))\n",
    "\n",
    "# # Determine the common x and y limits\n",
    "# x_min = -5\n",
    "# x_max = max(df['len_gpt_jargon'].max(), df['len_human_jargon'].max())\n",
    "# y_min = 0\n",
    "# y_max = 0.1\n",
    "\n",
    "# # Set the tick positions and labels\n",
    "# x_ticks = range(int(x_min), int(x_max) + 1, 5)  # Adjust the step size as needed\n",
    "# y_ticks = [0.0, 0.02, 0.04, 0.06, 0.08, 0.1, 0.12]  # Adjust the step size as needed\n",
    "\n",
    "# # Subplot for len_gpt_jargon\n",
    "# plt.subplot(1, 2, 1)\n",
    "# sns.kdeplot(data=df, x='len_gpt_jargon', hue='reader_id', fill=True, alpha=0.7, palette='viridis')\n",
    "# plt.title('Distribution of GPT Jargon Length')\n",
    "# plt.xlabel('Length')\n",
    "# plt.ylabel('Density')\n",
    "# plt.xlim(x_min, x_max)\n",
    "# plt.ylim(y_min, y_max)\n",
    "# plt.xticks(x_ticks)\n",
    "# plt.yticks(y_ticks)\n",
    "\n",
    "# # Subplot for len_human_jargon\n",
    "# plt.subplot(1, 2, 2)\n",
    "# sns.kdeplot(data=df, x='len_human_jargon', hue='reader_id', fill=True, alpha=0.7, palette='viridis')\n",
    "# plt.title('Distribution of Human Jargon Length')\n",
    "# plt.xlabel('Length')\n",
    "# plt.ylabel('Density')\n",
    "# plt.xlim(x_min, x_max)\n",
    "# plt.ylim(y_min, y_max)\n",
    "# plt.xticks(x_ticks)\n",
    "# plt.yticks(y_ticks)\n",
    "\n",
    "# plt.tight_layout()\n",
    "# plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import matplotlib.pyplot as plt\n",
    "# import seaborn as sns\n",
    "# import pandas as pd\n",
    "\n",
    "# # Determine the common x and y limits\n",
    "# x_min = -5\n",
    "# x_max = max(df['len_gpt_jargon'].max(), df['len_human_jargon'].max())\n",
    "# y_min = 0\n",
    "# y_max = 0.1\n",
    "\n",
    "# # Set the tick positions and labels\n",
    "# x_ticks = range(int(x_min), int(x_max) + 1, 5)  # Adjust the step size as needed\n",
    "# y_ticks = [0.0, 0.02, 0.04, 0.06, 0.08, 0.1, 0.12]  # Adjust the step size as needed\n",
    "\n",
    "# # Create subplots for each reader_id\n",
    "# unique_reader_ids = df['reader_id'].unique()\n",
    "# num_reader_ids = len(unique_reader_ids)\n",
    "\n",
    "# plt.figure(figsize=(12, 6))\n",
    "\n",
    "# for i, reader_id in enumerate(unique_reader_ids):\n",
    "#     plt.subplot(1, num_reader_ids, i+1)\n",
    "    \n",
    "#     # Melt the DataFrame to combine 'len_gpt_jargon' and 'len_human_jargon' into a single column\n",
    "#     melted_df = pd.melt(df[df['reader_id'] == reader_id], id_vars=['reader_id'], value_vars=['len_gpt_jargon', 'len_human_jargon'], var_name='jargon_type', value_name='length')\n",
    "    \n",
    "#     sns.kdeplot(data=melted_df, x='length', hue='jargon_type', fill=True, alpha=0.7, palette='viridis')\n",
    "    \n",
    "#     plt.title(f'Distribution of Jargon Length (Reader {reader_id})', fontsize=16, pad=20)\n",
    "#     plt.xlabel('Length', fontsize=14, labelpad=12)\n",
    "#     plt.ylabel('Probability Density', fontsize=14, labelpad=12)\n",
    "    \n",
    "#     plt.xlim(x_min, x_max)\n",
    "#     plt.ylim(y_min, y_max)\n",
    "    \n",
    "#     plt.xticks(x_ticks, fontsize=12)\n",
    "#     plt.yticks(y_ticks, fontsize=12)\n",
    "    \n",
    "#     plt.gca().tick_params(axis='both', which='major', labelsize=12)\n",
    "    \n",
    "#     # Calculate median values\n",
    "#     median_gpt = df[df['reader_id'] == reader_id]['len_gpt_jargon'].median()\n",
    "#     median_human = df[df['reader_id'] == reader_id]['len_human_jargon'].median()\n",
    "    \n",
    "#     # Add median lines\n",
    "#     plt.axvline(x=median_gpt, color='gray', linestyle='--', linewidth=1.5)\n",
    "#     plt.axvline(x=median_human, color='gray', linestyle='--', linewidth=1.5)\n",
    "    \n",
    "#     # Add median information to the legend\n",
    "#     # plt.legend(title='Jargon Source', labels=[f'Human (Median: {median_human:.2f})', f'GPT-4 (Median: {median_gpt:.2f})'], title_fontsize=14, fontsize=12)\n",
    "#     plt.legend(title='Jargon identified by:', labels=['Human', 'GPT-4'], title_fontsize=12, fontsize=12)\n",
    "#     # Add median information to the legend\n",
    "#     # plt.legend(title='Jargon Source', labels=[f'Human (Median: {int(median_human)})', f'GPT-4 (Median: {int(median_gpt)})'], title_fontsize=14, fontsize=12)\n",
    "#     # Add legend entry for median lines\n",
    "#     # plt.legend(handles=[plt.Line2D([], [], color='gray', linestyle='--', linewidth=1.5)], labels=['Median'], loc='upper right', fontsize=12)\n",
    "\n",
    "\n",
    "#     plt.text(median_gpt, y_max * 1.0, f'GPT Median: {median_gpt:.2f}', fontsize=12, ha='center', va='center', bbox=dict(facecolor='white', edgecolor='gray', boxstyle='round'))\n",
    "#     plt.text(median_human, y_max * 0.8, f'Human Median: {median_human:.2f}', fontsize=12, ha='center', va='center', bbox=dict(facecolor='white', edgecolor='gray', boxstyle='round'))\n",
    "\n",
    "\n",
    "# plt.tight_layout(pad=2.0)\n",
    "# plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import matplotlib.pyplot as plt\n",
    "# import seaborn as sns\n",
    "# import pandas as pd\n",
    "\n",
    "# # Determine the common x and y limits\n",
    "# x_min = -5\n",
    "# x_max = max(df['len_gpt_jargon'].max(), df['len_human_jargon'].max())\n",
    "# y_min = 0\n",
    "# y_max = 0.1\n",
    "\n",
    "# # Set the tick positions and labels\n",
    "# x_ticks = range(int(x_min), int(x_max) + 1, 5)  # Adjust the step size as needed\n",
    "# y_ticks = [0.0, 0.02, 0.04, 0.06, 0.08, 0.1, 0.12]  # Adjust the step size as needed\n",
    "\n",
    "# # Create subplots for each reader_id\n",
    "# unique_reader_ids = df['reader_id'].unique()\n",
    "# num_reader_ids = len(unique_reader_ids)\n",
    "\n",
    "# # Define colors for each jargon source\n",
    "# colors = ['#005AB5', '#DC3220']  # red for human, blue for GPT\n",
    "\n",
    "# plt.figure(figsize=(12, 6))\n",
    "\n",
    "# for i, reader_id in enumerate(unique_reader_ids):\n",
    "#     plt.subplot(1, num_reader_ids, i+1)\n",
    "    \n",
    "#     # Melt the DataFrame to combine 'len_gpt_jargon' and 'len_human_jargon' into a single column\n",
    "#     melted_df = pd.melt(df[df['reader_id'] == reader_id], id_vars=['reader_id'], value_vars=['len_gpt_jargon', 'len_human_jargon'], var_name='jargon_type', value_name='length')\n",
    "    \n",
    "#     # Create the KDE plot\n",
    "#     sns.kdeplot(data=melted_df, x='length', hue='jargon_type', fill=True, alpha=0.7, palette=colors)\n",
    "    \n",
    "#     plt.title(f'Distribution of Jargon Count (Reader {reader_id})', fontsize=16, pad=20)\n",
    "#     plt.xlabel('Count of Jargon Terms Identified', fontsize=14, labelpad=12)\n",
    "#     plt.ylabel('Probability Density', fontsize=14, labelpad=12)\n",
    "    \n",
    "#     plt.xlim(x_min, x_max)\n",
    "#     plt.ylim(y_min, y_max)\n",
    "    \n",
    "#     plt.xticks(x_ticks, fontsize=12)\n",
    "#     plt.yticks(y_ticks, fontsize=12)\n",
    "    \n",
    "#     plt.gca().tick_params(axis='both', which='major', labelsize=12)\n",
    "    \n",
    "#     plt.legend(title='Jargon Source', labels=['Human', 'GPT-4'], title_fontsize=14, fontsize=12)\n",
    "    \n",
    "#     # Calculate median values\n",
    "#     median_gpt = df[df['reader_id'] == reader_id]['len_gpt_jargon'].median()\n",
    "#     median_human = df[df['reader_id'] == reader_id]['len_human_jargon'].median()\n",
    "    \n",
    "#     # Add median lines with colors matching the KDE plot\n",
    "#     plt.axvline(x=median_human, color=colors[1], linestyle='--', linewidth=1.5)\n",
    "#     plt.axvline(x=median_gpt, color=colors[0], linestyle='--', linewidth=1.5)\n",
    "    \n",
    "#     # Add text boxes with median values and colors matching the KDE plot\n",
    "#     plt.text(median_human, y_max * 0.9, f'Median: {int(median_human)}', fontsize=12, color=colors[1],\n",
    "#              ha='center', va='center', bbox=dict(facecolor='white', edgecolor=colors[1], boxstyle='round'))\n",
    "#     plt.text(median_gpt, y_max * 0.8, f'Median: {int(median_gpt)}', fontsize=12, color=colors[0],\n",
    "#              ha='center', va='center', bbox=dict(facecolor='white', edgecolor=colors[0], boxstyle='round'))\n",
    "\n",
    "# plt.tight_layout(pad=2.0)\n",
    "# plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import pandas as pd\n",
    "\n",
    "# Determine the common x and y limits\n",
    "x_min = -5\n",
    "x_max = max(df['len_gpt_jargon'].max(), df['len_human_jargon'].max())\n",
    "y_min = 0\n",
    "y_max = 0.25\n",
    "\n",
    "# Set the tick positions and labels\n",
    "x_ticks = range(int(x_min), int(x_max) + 1, 5)  # Adjust the step size as needed\n",
    "# y_ticks = [0.0, 0.02, 0.04, 0.06, 0.08, 0.1, 0.12]  # Adjust the step size as needed\n",
    "\n",
    "# Define colors for each reader\n",
    "# colors = ['#E69F00', '#56B4E9']  # Orange for rid1, Sky Blue for rid0\n",
    "\n",
    "# Define colors for each reader\n",
    "colors = ['#005AB5', '#DC3220']  # blue for rid0, red for rid1\n",
    "\n",
    "plt.figure(figsize=(12, 6))\n",
    "\n",
    "# Subplot for human jargon length\n",
    "plt.subplot(1, 2, 1)\n",
    "for reader_id, color in zip(df['reader_id'].unique(), colors):\n",
    "    sns.kdeplot(data=df[df['reader_id'] == reader_id], x='len_human_jargon', fill=True, alpha=0.7, color=color, label=f'rid{int(reader_id[-1])}')\n",
    "\n",
    "plt.title('Human Annotators', fontsize=20, pad=15)\n",
    "plt.xlabel('Count of Jargon Terms per Abstract', fontsize=16, labelpad=10)\n",
    "plt.ylabel('Probability Density', fontsize=18, labelpad=10)\n",
    "plt.xlim(x_min, x_max)\n",
    "plt.ylim(y_min, y_max)\n",
    "plt.xticks(x_ticks, fontsize=16)\n",
    "# plt.yticks(y_ticks, fontsize=12)\n",
    "plt.gca().tick_params(axis='both', which='major', labelsize=12)\n",
    "plt.legend(title='Annotator ID', title_fontsize=14, fontsize=14)\n",
    "\n",
    "# Calculate median values for human jargon length\n",
    "for i, (reader_id, color) in enumerate(zip(df['reader_id'].unique(), colors)):\n",
    "    median_human = df[df['reader_id'] == reader_id]['len_human_jargon'].median()\n",
    "    plt.axvline(x=median_human, color=color, linestyle='--', linewidth=1.5)\n",
    "    plt.text(median_human, y_max * (0.9 - 0.1 * i), f'Median: {int(median_human)}', fontsize=14, color=color,\n",
    "             ha='center', va='center', bbox=dict(facecolor='white', edgecolor=color, boxstyle='round'))\n",
    "\n",
    "\n",
    "# Subplot for GPT jargon length\n",
    "plt.subplot(1, 2, 2)\n",
    "for reader_id, color in zip(df['reader_id'].unique(), colors):\n",
    "    sns.kdeplot(data=df[df['reader_id'] == reader_id], x='len_gpt_jargon', fill=True, alpha=0.7, color=color, label=f'rid{int(reader_id[-1])}')\n",
    "\n",
    "# plt.title('Jargon Identified by GPT-4', fontsize=20, pad=15)\n",
    "plt.title('GPT-4 Personalization', fontsize=20, pad=15)\n",
    "plt.xlabel('Count of Jargon Terms per Abstract', fontsize=16, labelpad=12)\n",
    "plt.ylabel('Probability Density', fontsize=18, labelpad=12)\n",
    "plt.xlim(x_min, x_max)\n",
    "plt.ylim(y_min, y_max)\n",
    "plt.xticks(x_ticks, fontsize=16)\n",
    "# plt.yticks(y_ticks, fontsize=12)\n",
    "plt.gca().tick_params(axis='both', which='major', labelsize=12)\n",
    "plt.legend(title='Annotator ID', title_fontsize=14, fontsize=14)\n",
    "\n",
    "# Calculate median values for GPT jargon length\n",
    "for i, (reader_id, color) in enumerate(zip(df['reader_id'].unique(), colors)):\n",
    "    median_gpt = df[df['reader_id'] == reader_id]['len_gpt_jargon'].median()\n",
    "    plt.axvline(x=median_gpt, color=color, linestyle='--', linewidth=1.5)\n",
    "    plt.text(median_gpt, y_max * (0.9 - 0.1 * i), f'Median: {int(median_gpt)}', fontsize=14, color=color,\n",
    "             ha='center', va='center', bbox=dict(facecolor='white', edgecolor=color, boxstyle='round'))\n",
    "\n",
    "\n",
    "plt.tight_layout(pad=2.0)\n",
    "plt.savefig('figs/humanvsgpt_kdeplot.png', dpi=300, bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Do a Wilcoxon signed-rank test to see if the sample from population for len_gpt_jargon is greater than len_human_jargon, for each rader. Since this is paired data (for each reader, for each abstract, you have this)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Median differences (len_gpt_jargon - len_human_jargon) for each reader:\n",
      "  reader_id  median_diff\n",
      "0      rid0          4.0\n",
      "1      rid1          5.0\n",
      "\n",
      "Wilcoxon signed-rank test p-values for each reader:\n",
      "  reader_id       p_value\n",
      "0      rid0  1.350370e-09\n",
      "1      rid1  6.817292e-11\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/var/folders/5y/b4h6jbjn4y57lpgqtmtjb5240000gq/T/ipykernel_57123/1390841816.py:5: DeprecationWarning: DataFrameGroupBy.apply operated on the grouping columns. This behavior is deprecated, and in a future version of pandas the grouping columns will be excluded from the operation. Either pass `include_groups=False` to exclude the groupings or explicitly select the grouping columns after groupby to silence this warning.\n",
      "  median_diffs = df.groupby('reader_id').apply(lambda x: x['len_gpt_jargon'].median() - x['len_human_jargon'].median()).reset_index(name='median_diff')\n",
      "/var/folders/5y/b4h6jbjn4y57lpgqtmtjb5240000gq/T/ipykernel_57123/1390841816.py:11: DeprecationWarning: DataFrameGroupBy.apply operated on the grouping columns. This behavior is deprecated, and in a future version of pandas the grouping columns will be excluded from the operation. Either pass `include_groups=False` to exclude the groupings or explicitly select the grouping columns after groupby to silence this warning.\n",
      "  p_values = df.groupby('reader_id').apply(lambda x: wilcoxon(x['len_gpt_jargon'], x['len_human_jargon'])).reset_index(name='wilcoxon_result')\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "from scipy.stats import wilcoxon\n",
    "\n",
    "# Group the data by reader_id and calculate the median differences\n",
    "median_diffs = df.groupby('reader_id').apply(lambda x: x['len_gpt_jargon'].median() - x['len_human_jargon'].median()).reset_index(name='median_diff')\n",
    "\n",
    "print(\"Median differences (len_gpt_jargon - len_human_jargon) for each reader:\")\n",
    "print(median_diffs)\n",
    "\n",
    "# Perform Wilcoxon signed-rank test for each reader\n",
    "p_values = df.groupby('reader_id').apply(lambda x: wilcoxon(x['len_gpt_jargon'], x['len_human_jargon'])).reset_index(name='wilcoxon_result')\n",
    "p_values['p_value'] = p_values['wilcoxon_result'].apply(lambda x: x[1])\n",
    "\n",
    "print(\"\\nWilcoxon signed-rank test p-values for each reader:\")\n",
    "print(p_values[['reader_id', 'p_value']])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import pandas as pd\n",
    "# from scipy.stats import wilcoxon\n",
    "\n",
    "# # Calculate the percentage difference between len_gpt_jargon and len_human_jargon for each row\n",
    "# df['pct_diff'] = (df['len_gpt_jargon'] - df['len_human_jargon']) / df['len_human_jargon'] * 100\n",
    "\n",
    "# # Group the data by reader_id and calculate the median percentage differences\n",
    "# median_pct_diffs = df.groupby('reader_id')['pct_diff'].median().reset_index(name='median_pct_diff')\n",
    "\n",
    "# print(\"Median percentage differences ((len_gpt_jargon - len_human_jargon) / len_human_jargon * 100) for each reader:\")\n",
    "# print(median_pct_diffs)\n",
    "\n",
    "# # Perform Wilcoxon signed-rank test on the percentage differences for each reader\n",
    "# p_values = df.groupby('reader_id')['pct_diff'].apply(lambda x: wilcoxon(x, zero_method='pratt')).reset_index(name='wilcoxon_result')\n",
    "# p_values['p_value'] = p_values['wilcoxon_result'].apply(lambda x: x[1])\n",
    "\n",
    "# print(\"\\nWilcoxon signed-rank test p-values for each reader:\")\n",
    "# print(p_values[['reader_id', 'p_value']])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(28, 12)"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Counts of instances where no jargon was reported\n",
    "jargon_identify_eval_df.loc[\n",
    "    jargon_identify_eval_df['len_human_jargon']==0\n",
    "].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Set the style of seaborn\n",
    "# sns.set(style=\"whitegrid\")\n",
    "\n",
    "# # Create a boxplot for len_gpt_jargon for each reader_id\n",
    "# plt.figure(figsize=(4, 6))\n",
    "# sns.boxplot(\n",
    "#     data=jargon_identify_eval_df, \n",
    "#     y='len_gpt_jargon',\n",
    "#     hue='reader_id', \n",
    "#     palette='colorblind', \n",
    "#     width=0.5, \n",
    "#     gap=0.1\n",
    "# )\n",
    "# plt.xlabel('Reader ID')\n",
    "# plt.ylabel('Length of GPT Jargon')\n",
    "# plt.title('Length of GPT Jargon by Reader ID')\n",
    "\n",
    "# # Manually set integer yticks\n",
    "# plt.yticks(np.arange(0, 25, 2))\n",
    "\n",
    "\n",
    "# plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Set the style of seaborn\n",
    "# sns.set(style=\"whitegrid\")\n",
    "\n",
    "# # Create a boxplot for len_gpt_jargon for each reader_id\n",
    "# plt.figure(figsize=(4, 6))\n",
    "# sns.boxplot(\n",
    "#     data=jargon_identify_eval_df, \n",
    "#     y='len_human_jargon',\n",
    "#     hue='reader_id', \n",
    "#     palette='colorblind', \n",
    "#     width=0.5, \n",
    "#     gap=0.1\n",
    "# )\n",
    "# plt.xlabel('Reader ID')\n",
    "# plt.ylabel('Length of Annotated Jargon')\n",
    "# plt.title('Length of Annotated Jargon by Reader ID')\n",
    "\n",
    "# # Manually set integer yticks\n",
    "# plt.yticks(np.arange(0, 25, 2))\n",
    "\n",
    "\n",
    "# plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Set the style of seaborn\n",
    "# sns.set(style=\"whitegrid\")\n",
    "\n",
    "# # Create a boxplot for len_gpt_jargon for each reader_id\n",
    "# plt.figure(figsize=(4, 6))\n",
    "# sns.boxplot(\n",
    "#     data=jargon_identify_eval_df, \n",
    "#     y='len_jargon_diff',\n",
    "#     hue='reader_id', \n",
    "#     palette='colorblind', \n",
    "#     width=0.5, \n",
    "#     gap=0.1\n",
    "# )\n",
    "# plt.xlabel('Reader ID')\n",
    "# plt.ylabel('Length of GPT Jargon')\n",
    "# plt.title('Length of GPT Jargon by Reader ID')\n",
    "\n",
    "# # Manually set integer yticks\n",
    "# # plt.yticks(np.arange(0, 25, 2))\n",
    "\n",
    "\n",
    "# plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Set the style of seaborn\n",
    "# sns.set(style=\"whitegrid\")\n",
    "\n",
    "# # Create a boxplot for len_gpt_jargon for each reader_id\n",
    "# plt.figure(figsize=(4, 6))\n",
    "# sns.boxplot(\n",
    "#     data=jargon_identify_eval_df, \n",
    "#     y='len_jargon_diff',\n",
    "#     #hue='reader_id', \n",
    "#     #palette='colorblind', \n",
    "#     width=0.5, \n",
    "#     gap=0.1\n",
    "# )\n",
    "# plt.xlabel('Reader ID')\n",
    "# plt.ylabel('Length of GPT Jargon')\n",
    "# plt.title('Length of GPT Jargon by Reader ID')\n",
    "\n",
    "# # Manually set integer yticks\n",
    "# # plt.yticks(np.arange(0, 25, 2))\n",
    "\n",
    "\n",
    "# plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shapiro-Wilk p-value: 0.0034\n"
     ]
    }
   ],
   "source": [
    "# Test if diffs are normal\n",
    "from scipy import stats\n",
    "_, p_value = stats.shapiro(jargon_identify_eval_df['len_jargon_diff'])\n",
    "print(f\"Shapiro-Wilk p-value: {p_value:.4f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "p-value less than 0.05, we reject the null. The distribution is not normal."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wilcoxon signed-rank test p-value: 0.0000\n"
     ]
    }
   ],
   "source": [
    "# Do a Wilcoxon signed-rank test\n",
    "_, p_value = stats.wilcoxon(jargon_identify_eval_df['len_gpt_jargon'], jargon_identify_eval_df['len_human_jargon'])\n",
    "print(f\"Wilcoxon signed-rank test p-value: {p_value:.4f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "p-value less than 0.05, we reject the null. The distributions are significantly different."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Median difference: 4.0000\n"
     ]
    }
   ],
   "source": [
    "differences = [gpt_len - human_len for human_len, gpt_len in zip(jargon_identify_eval_df['len_human_jargon'], jargon_identify_eval_df['len_gpt_jargon'])]\n",
    "median_diff = np.median(differences)\n",
    "print(f\"Median difference: {median_diff:.4f}\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Precision and Recall"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>arxiv_id</th>\n",
       "      <th>reader_id</th>\n",
       "      <th>len_human_jargon</th>\n",
       "      <th>variable</th>\n",
       "      <th>value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2403.16190v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>3</td>\n",
       "      <td>precision</td>\n",
       "      <td>0.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2403.16190v1</td>\n",
       "      <td>rid1</td>\n",
       "      <td>3</td>\n",
       "      <td>precision</td>\n",
       "      <td>0.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2307.05300v4</td>\n",
       "      <td>rid0</td>\n",
       "      <td>2</td>\n",
       "      <td>precision</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2307.05300v4</td>\n",
       "      <td>rid1</td>\n",
       "      <td>3</td>\n",
       "      <td>precision</td>\n",
       "      <td>0.166667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2403.16750v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>6</td>\n",
       "      <td>precision</td>\n",
       "      <td>0.833333</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       arxiv_id reader_id  len_human_jargon   variable     value\n",
       "0  2403.16190v1      rid0                 3  precision  0.333333\n",
       "1  2403.16190v1      rid1                 3  precision  0.333333\n",
       "2  2307.05300v4      rid0                 2  precision  0.000000\n",
       "3  2307.05300v4      rid1                 3  precision  0.166667\n",
       "4  2403.16750v1      rid0                 6  precision  0.833333"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "jargon_identify_eval_df_melted = pd.melt(\n",
    "    jargon_identify_eval_df[['arxiv_id', 'reader_id', 'precision', 'recall', 'f1_score', 'f2_score', 'len_human_jargon']], \n",
    "    id_vars = ['arxiv_id', 'reader_id', 'len_human_jargon'], \n",
    "    value_vars = ['precision', 'recall', 'f1_score', 'f2_score']\n",
    ")\n",
    "jargon_identify_eval_df_melted.head()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 900x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import numpy as np\n",
    "\n",
    "# Define colors for each reader\n",
    "# colors = ['#005AB5', '#DC3220']  # blue for rid0, red for rid1\n",
    "# Create a custom color palette\n",
    "custom_palette = sns.color_palette(['#1E7BDA', '#F5301B'])\n",
    "\n",
    "# Create the grouped bar plot\n",
    "plt.figure(figsize=(9, 6))\n",
    "ax = sns.boxplot(\n",
    "    data=jargon_identify_eval_df_melted.loc[\n",
    "        (jargon_identify_eval_df_melted['len_human_jargon']>0) &\n",
    "        (jargon_identify_eval_df_melted['variable']!='f1_score')\n",
    "    ], \n",
    "    y='value', \n",
    "    x='variable', \n",
    "    hue='reader_id', \n",
    "    width=0.4, \n",
    "    gap=0.1, \n",
    "    dodge=True, \n",
    "    palette=custom_palette,  # Set the color palette\n",
    "    boxprops=dict(alpha=0.9)\n",
    ")\n",
    "plt.xlabel(\"Evaluation Metrics for GPT-4's Predicted Jargon Terms\", fontsize=14, labelpad=12)\n",
    "plt.ylabel('Score', fontsize=14, labelpad=12)\n",
    "# plt.title(\"Evaluating GPT4's Jargon Identification\", fontsize=16, pad=20)\n",
    "\n",
    "# Set the x-tick labels\n",
    "x_labels = ['Precision', 'Recall', 'F2.0']\n",
    "plt.xticks(range(len(x_labels)), x_labels, rotation=0, fontsize=12)\n",
    "\n",
    "plt.yticks(np.arange(0, 1.1, 0.1), fontsize=12)\n",
    "\n",
    "# Modify the legend title\n",
    "plt.legend(title='Annotator ID', title_fontsize=12, fontsize=12, bbox_to_anchor=(1.02, 1), loc='upper left', borderaxespad=0)\n",
    "plt.tight_layout(pad=2.0)\n",
    "plt.savefig('figs/prf2_boxplot.png', dpi=300, bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Median recall for rid0:\n",
      "0.8333333333333334\n"
     ]
    }
   ],
   "source": [
    "print(\"Median recall for rid0:\")\n",
    "print(jargon_identify_eval_df_melted.loc[\n",
    "        (jargon_identify_eval_df_melted['len_human_jargon']>0) &\n",
    "        (jargon_identify_eval_df_melted['variable']==0'recall') &\n",
    "        (jargon_identify_eval_df_melted['reader_id']=='rid0')\n",
    "\n",
    "    ]['value'].median())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Median recall for rid1:\n",
      "0.6666666666666666\n"
     ]
    }
   ],
   "source": [
    "print(\"Median recall for rid1:\")\n",
    "print(jargon_identify_eval_df_melted.loc[\n",
    "        (jargon_identify_eval_df_melted['len_human_jargon']>0) &\n",
    "        (jargon_identify_eval_df_melted['variable']=='recall') &\n",
    "        (jargon_identify_eval_df_melted['reader_id']=='rid1')\n",
    "\n",
    "    ]['value'].median())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Median recall for all:\n",
      "0.6833333333333333\n"
     ]
    }
   ],
   "source": [
    "print(\"Median recall for all:\")\n",
    "print(jargon_identify_eval_df_melted.loc[\n",
    "        (jargon_identify_eval_df_melted['len_human_jargon']>0) &\n",
    "        (jargon_identify_eval_df_melted['variable']=='recall')\n",
    "    ]['value'].median())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Median precision for all:\n",
      "0.3333333333333333\n"
     ]
    }
   ],
   "source": [
    "print(\"Median precision for all:\")\n",
    "print(jargon_identify_eval_df_melted.loc[\n",
    "        (jargon_identify_eval_df_melted['len_human_jargon']>0) &\n",
    "        (jargon_identify_eval_df_melted['variable']=='precision')\n",
    "    ]['value'].median())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Median precision for rid0:\n",
      "0.3333333333333333\n"
     ]
    }
   ],
   "source": [
    "print(\"Median precision for rid0:\")\n",
    "print(jargon_identify_eval_df_melted.loc[\n",
    "        (jargon_identify_eval_df_melted['len_human_jargon']>0) &\n",
    "        (jargon_identify_eval_df_melted['variable']=='precision') &\n",
    "        (jargon_identify_eval_df_melted['reader_id']=='rid0')\n",
    "\n",
    "    ]['value'].median())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Median precision for rid1:\n",
      "0.2761904761904762\n"
     ]
    }
   ],
   "source": [
    "print(\"Median precision for rid1:\")\n",
    "print(jargon_identify_eval_df_melted.loc[\n",
    "        (jargon_identify_eval_df_melted['len_human_jargon']>0) &\n",
    "        (jargon_identify_eval_df_melted['variable']=='precision') &\n",
    "        (jargon_identify_eval_df_melted['reader_id']=='rid1')\n",
    "\n",
    "    ]['value'].median())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Median F2 for all:\n",
      "0.5495169082125604\n",
      "Median F2 for rid0:\n",
      "0.5718954248366013\n",
      "Median F2 for rid1:\n",
      "0.5263157894736843\n"
     ]
    }
   ],
   "source": [
    "print(\"Median F2 for all:\")\n",
    "print(jargon_identify_eval_df_melted.loc[\n",
    "        (jargon_identify_eval_df_melted['len_human_jargon']>0) &\n",
    "        (jargon_identify_eval_df_melted['variable']=='f2_score')\n",
    "    ]['value'].median())\n",
    "\n",
    "print(\"Median F2 for rid0:\")\n",
    "print(jargon_identify_eval_df_melted.loc[\n",
    "        (jargon_identify_eval_df_melted['len_human_jargon']>0) &\n",
    "        (jargon_identify_eval_df_melted['variable']=='f2_score') &\n",
    "        (jargon_identify_eval_df_melted['reader_id']=='rid0')\n",
    "\n",
    "    ]['value'].median())\n",
    "\n",
    "print(\"Median F2 for rid1:\")\n",
    "print(jargon_identify_eval_df_melted.loc[\n",
    "        (jargon_identify_eval_df_melted['len_human_jargon']>0) &\n",
    "        (jargon_identify_eval_df_melted['variable']=='f2_score') &\n",
    "        (jargon_identify_eval_df_melted['reader_id']=='rid1')\n",
    "\n",
    "    ]['value'].median())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Length of Individual Jargon Terms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>arxiv_id</th>\n",
       "      <th>reader_id</th>\n",
       "      <th>jargon_term</th>\n",
       "      <th>len_jargon_term</th>\n",
       "      <th>source_jargon_term</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2403.16190v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>correctness</td>\n",
       "      <td>1</td>\n",
       "      <td>human</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2403.16190v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>minimality</td>\n",
       "      <td>1</td>\n",
       "      <td>human</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2403.16190v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>Anchors</td>\n",
       "      <td>1</td>\n",
       "      <td>human</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2403.16190v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>reject option strategy</td>\n",
       "      <td>3</td>\n",
       "      <td>gpt</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2403.16190v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>formal guarantees</td>\n",
       "      <td>2</td>\n",
       "      <td>gpt</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       arxiv_id reader_id             jargon_term  len_jargon_term  \\\n",
       "0  2403.16190v1      rid0             correctness                1   \n",
       "1  2403.16190v1      rid0              minimality                1   \n",
       "2  2403.16190v1      rid0                 Anchors                1   \n",
       "3  2403.16190v1      rid0  reject option strategy                3   \n",
       "4  2403.16190v1      rid0       formal guarantees                2   \n",
       "\n",
       "  source_jargon_term  \n",
       "0              human  \n",
       "1              human  \n",
       "2              human  \n",
       "3                gpt  \n",
       "4                gpt  "
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Maintain a list of jargon terms individually instead of in a list\n",
    "jargon_terms_individual = []\n",
    "\n",
    "# Iterate and operate\n",
    "for idx, row in merged_humam_llm_jargon.iterrows():\n",
    "\n",
    "    # Get both lists\n",
    "    gpt_list = row['gpt4_jargon_list']\n",
    "    human_list = row['human_jargon_list']\n",
    "    # Convert to actual lists\n",
    "    human_list = [i.strip() for i in human_list.split(',') if i.strip()]\n",
    "    gpt_list = [i.strip() for i in gpt_list.split(',')]\n",
    "\n",
    "    # Begin appending\n",
    "    for term in human_list:\n",
    "        jargon_terms_individual.append(\n",
    "            {\n",
    "                'arxiv_id': row['arxiv_id'],\n",
    "                'reader_id': row['reader_id'],\n",
    "                'jargon_term': term, \n",
    "                'len_jargon_term': len(term.split(' ')),\n",
    "                'source_jargon_term': 'human'\n",
    "            }\n",
    "        \n",
    "        )\n",
    "    for term in gpt_list:\n",
    "        jargon_terms_individual.append(\n",
    "            {\n",
    "                'arxiv_id': row['arxiv_id'],\n",
    "                'reader_id': row['reader_id'],\n",
    "                'jargon_term': term, \n",
    "                'len_jargon_term': len(term.split(' ')),\n",
    "                'source_jargon_term': 'gpt'\n",
    "            }\n",
    "        \n",
    "        )\n",
    "\n",
    "# Convert to DF\n",
    "jargon_terms_individual_df = pd.DataFrame(jargon_terms_individual)\n",
    "jargon_terms_individual_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "plt.figure(figsize=(6, 6))\n",
    "\n",
    "sns.boxplot(\n",
    "    data=jargon_terms_individual_df, \n",
    "    x='source_jargon_term', \n",
    "    y='len_jargon_term', \n",
    "    hue='reader_id', \n",
    "    palette='viridis', \n",
    "    width=0.4, \n",
    "    gap=0.1,\n",
    "    dodge=True,  \n",
    "    boxprops=dict(alpha=0.9)\n",
    ")\n",
    "\n",
    "plt.title('Distribution of Jargon Term Length by Source and Reader', fontsize=16, pad=20)\n",
    "plt.xlabel('Source of Jargon Terms', fontsize=14, labelpad=12)\n",
    "plt.ylabel('Length of Jargon Terms', fontsize=14, labelpad=12)\n",
    "\n",
    "plt.xticks(fontsize=12, rotation=45)\n",
    "plt.yticks(fontsize=12)\n",
    "\n",
    "plt.legend(title='Reader ID', fontsize=12)\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.savefig('boxplot_jargon_term_length_dodged.png', dpi=300, bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>arxiv_id</th>\n",
       "      <th>reader_id</th>\n",
       "      <th>jargon_term</th>\n",
       "      <th>len_jargon_term</th>\n",
       "      <th>source_jargon_term</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>409</th>\n",
       "      <td>2307.03067v2</td>\n",
       "      <td>rid0</td>\n",
       "      <td>Bio-ML track of the Ontology Alignment Evaluat...</td>\n",
       "      <td>9</td>\n",
       "      <td>gpt</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>788</th>\n",
       "      <td>2403.10851v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>wireless earbuds for sensing different eating ...</td>\n",
       "      <td>9</td>\n",
       "      <td>gpt</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>800</th>\n",
       "      <td>2403.10851v1</td>\n",
       "      <td>rid1</td>\n",
       "      <td>wireless earbuds for sensing different eating ...</td>\n",
       "      <td>9</td>\n",
       "      <td>gpt</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>685</th>\n",
       "      <td>2403.08451v1</td>\n",
       "      <td>rid1</td>\n",
       "      <td>European Union and Gulf Cooperation Council (G...</td>\n",
       "      <td>8</td>\n",
       "      <td>human</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>516</th>\n",
       "      <td>2403.16289v1</td>\n",
       "      <td>rid1</td>\n",
       "      <td>prototype of a pipeline of prompts and LLMs</td>\n",
       "      <td>8</td>\n",
       "      <td>gpt</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>549</th>\n",
       "      <td>2402.01786v2</td>\n",
       "      <td>rid1</td>\n",
       "      <td>militarized version of the StarCraft II game</td>\n",
       "      <td>7</td>\n",
       "      <td>gpt</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>175</th>\n",
       "      <td>2403.17358v1</td>\n",
       "      <td>rid0</td>\n",
       "      <td>large constrained partially observable Markov ...</td>\n",
       "      <td>7</td>\n",
       "      <td>human</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1100</th>\n",
       "      <td>2403.08057v1</td>\n",
       "      <td>rid1</td>\n",
       "      <td>personalized XR user interaction and experienc...</td>\n",
       "      <td>7</td>\n",
       "      <td>gpt</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>537</th>\n",
       "      <td>2402.01786v2</td>\n",
       "      <td>rid0</td>\n",
       "      <td>militarized version of the StarCraft II game</td>\n",
       "      <td>7</td>\n",
       "      <td>gpt</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>191</th>\n",
       "      <td>2403.17358v1</td>\n",
       "      <td>rid1</td>\n",
       "      <td>constrained partially observable Markov decisi...</td>\n",
       "      <td>7</td>\n",
       "      <td>gpt</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          arxiv_id reader_id  \\\n",
       "409   2307.03067v2      rid0   \n",
       "788   2403.10851v1      rid0   \n",
       "800   2403.10851v1      rid1   \n",
       "685   2403.08451v1      rid1   \n",
       "516   2403.16289v1      rid1   \n",
       "549   2402.01786v2      rid1   \n",
       "175   2403.17358v1      rid0   \n",
       "1100  2403.08057v1      rid1   \n",
       "537   2402.01786v2      rid0   \n",
       "191   2403.17358v1      rid1   \n",
       "\n",
       "                                            jargon_term  len_jargon_term  \\\n",
       "409   Bio-ML track of the Ontology Alignment Evaluat...                9   \n",
       "788   wireless earbuds for sensing different eating ...                9   \n",
       "800   wireless earbuds for sensing different eating ...                9   \n",
       "685   European Union and Gulf Cooperation Council (G...                8   \n",
       "516         prototype of a pipeline of prompts and LLMs                8   \n",
       "549        militarized version of the StarCraft II game                7   \n",
       "175   large constrained partially observable Markov ...                7   \n",
       "1100  personalized XR user interaction and experienc...                7   \n",
       "537        militarized version of the StarCraft II game                7   \n",
       "191   constrained partially observable Markov decisi...                7   \n",
       "\n",
       "     source_jargon_term  \n",
       "409                 gpt  \n",
       "788                 gpt  \n",
       "800                 gpt  \n",
       "685               human  \n",
       "516                 gpt  \n",
       "549                 gpt  \n",
       "175               human  \n",
       "1100                gpt  \n",
       "537                 gpt  \n",
       "191                 gpt  "
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "jargon_terms_individual_df.sort_values('len_jargon_term', ascending=False).head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mann-Whitney U test:\n",
      "Statistic: 131235.0\n",
      "p-value: 0.0008107229811835362\n",
      "The two distributions are significantly different.\n",
      "On average, values from distribution 2 are 0.0 units greater than values from distribution 1.\n"
     ]
    }
   ],
   "source": [
    "# Do the shapiro wilk on human vs. gpt len_jargon_term for entire dataset\n",
    "from scipy.stats import mannwhitneyu\n",
    "\n",
    "dist1 = jargon_terms_individual_df.loc[\n",
    "    (jargon_terms_individual_df['source_jargon_term']=='human') \n",
    "]['len_jargon_term'].tolist()  # Data for distribution 1\n",
    "dist2 = jargon_terms_individual_df.loc[\n",
    "    (jargon_terms_individual_df['source_jargon_term']=='gpt') \n",
    "]['len_jargon_term'].tolist()    # Data for distribution 2\n",
    "\n",
    "# Perform the Mann-Whitney U test\n",
    "statistic, p_value = mannwhitneyu(dist1, dist2)\n",
    "\n",
    "# Calculate the median difference\n",
    "median_diff = np.median(dist1) - np.median(dist2)\n",
    "\n",
    "# Print the results\n",
    "print(\"Mann-Whitney U test:\")\n",
    "print(f\"Statistic: {statistic}\")\n",
    "print(f\"p-value: {p_value}\")\n",
    "\n",
    "# Interpret the results\n",
    "alpha = 0.05  # Significance level\n",
    "if p_value < alpha:\n",
    "    print(\"The two distributions are significantly different.\")\n",
    "    if median_diff > 0:\n",
    "        print(f\"On average, values from distribution 1 are {median_diff} units greater than values from distribution 2.\")\n",
    "    else:\n",
    "        print(f\"On average, values from distribution 2 are {abs(median_diff)} units greater than values from distribution 1.\")\n",
    "else:\n",
    "    print(\"The two distributions are not significantly different.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mann-Whitney U test:\n",
      "Statistic: 49069.0\n",
      "p-value: 1.548611256487033e-05\n",
      "The two distributions are significantly different.\n",
      "On average, values from distribution 2 are 0.31615476466881187 units greater than values from distribution 1.\n"
     ]
    }
   ],
   "source": [
    "# Do the shapiro wilk on human vs. gpt len_jargon_term for each reader\n",
    "from scipy.stats import mannwhitneyu\n",
    "\n",
    "dist1 = jargon_terms_individual_df.loc[\n",
    "    (jargon_terms_individual_df['source_jargon_term']=='human') &\n",
    "    (jargon_terms_individual_df['reader_id']=='rid1')\n",
    "]['len_jargon_term'].tolist()  # Data for distribution 1\n",
    "dist2 = jargon_terms_individual_df.loc[\n",
    "    (jargon_terms_individual_df['source_jargon_term']=='gpt') &\n",
    "    (jargon_terms_individual_df['reader_id']=='rid1')\n",
    "]['len_jargon_term'].tolist()    # Data for distribution 2\n",
    "\n",
    "# Perform the Mann-Whitney U test\n",
    "statistic, p_value = mannwhitneyu(dist1, dist2)\n",
    "\n",
    "# Calculate the median difference\n",
    "mean_diff = np.mean(dist1) - np.mean(dist2)\n",
    "\n",
    "# Print the results\n",
    "print(\"Mann-Whitney U test:\")\n",
    "print(f\"Statistic: {statistic}\")\n",
    "print(f\"p-value: {p_value}\")\n",
    "\n",
    "# Interpret the results\n",
    "alpha = 0.01  # Significance level\n",
    "if p_value < alpha:\n",
    "    print(\"The two distributions are significantly different.\")\n",
    "    if mean_diff > 0:\n",
    "        print(f\"On average, values from distribution 1 are {mean_diff} units greater than values from distribution 2.\")\n",
    "    else:\n",
    "        print(f\"On average, values from distribution 2 are {abs(mean_diff)} units greater than values from distribution 1.\")\n",
    "else:\n",
    "    print(\"The two distributions are not significantly different.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Exploring the Low Recall Cases"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get the low recall abstracts out\n",
    "low_recall_abstracts = jargon_identify_eval_df.loc[\n",
    "    jargon_identify_eval_df['recall'] < 0.5\n",
    "]\n",
    "\n",
    "print(low_recall_abstracts.shape)\n",
    "\n",
    "# Add abstract metadata\n",
    "low_recall_abstracts = pd.merge(\n",
    "    low_recall_abstracts, metadata_df, on='arxiv_id'\n",
    ")\n",
    "\n",
    "# Add in the merged_humam_llm_jargon\n",
    "low_recall_abstracts = pd.merge(\n",
    "    low_recall_abstracts, merged_humam_llm_jargon, on=['arxiv_id', 'reader_id']\n",
    ")\n",
    "\n",
    "low_recall_abstracts.to_json(\n",
    "    f'data/llm_outputs/{today}_march_2024_sampled_jargon_identify_eval_metrics_low_recall.json', \n",
    "    orient='index'\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "low_recall_abstracts.value_counts('reader_id')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get the low prec abstracts out\n",
    "low_prec_abstracts = jargon_identify_eval_df.loc[\n",
    "    jargon_identify_eval_df['precision'] < 0.25\n",
    "]\n",
    "\n",
    "# Add abstract metadata\n",
    "low_prec_abstracts = pd.merge(\n",
    "    low_prec_abstracts, metadata_df, on='arxiv_id'\n",
    ")\n",
    "\n",
    "# Add in the merged_humam_llm_jargon\n",
    "low_prec_abstracts = pd.merge(\n",
    "    low_prec_abstracts, merged_humam_llm_jargon, on=['arxiv_id', 'reader_id']\n",
    ")\n",
    "\n",
    "low_prec_abstracts.to_json(\n",
    "    f'data/llm_outputs/{today}_march_2024_sampled_jargon_identify_eval_metrics_low_prec.json', \n",
    "    orient='index'\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "low_prec_abstracts.value_counts('reader_id')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Exploring the Low Precision Cases"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(20, 27)\n"
     ]
    }
   ],
   "source": [
    "# Get the low recall abstracts out + Make sure it's stuff w/ at least one human annotated jargon term\n",
    "low_prec_abstracts = jargon_identify_eval_df.loc[\n",
    "    (jargon_identify_eval_df['precision'] <= 0.15) & \n",
    "    (jargon_identify_eval_df['len_human_jargon'] >0 )\n",
    "]\n",
    "\n",
    "# Add abstract metadata\n",
    "low_prec_abstracts = pd.merge(\n",
    "    low_prec_abstracts, metadata_df, on='arxiv_id'\n",
    ")\n",
    "\n",
    "# Add in the merged_humam_llm_jargon\n",
    "low_prec_abstracts = pd.merge(\n",
    "    low_prec_abstracts, merged_humam_llm_jargon, on=['arxiv_id', 'reader_id']\n",
    ")\n",
    "\n",
    "\n",
    "\n",
    "print(low_prec_abstracts.shape)\n",
    "\n",
    "low_prec_abstracts.to_json(\n",
    "    f'data/llm_outputs/{today}_march_2024_sampled_jargon_identify_eval_metrics_low_prec.json', \n",
    "    orient='index'\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "reader_id\n",
       "rid1    12\n",
       "rid0     8\n",
       "Name: count, dtype: int64"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# how many of each annotator\n",
    "low_prec_abstracts.value_counts('reader_id')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "reader_id\n",
       "rid0    1.625000\n",
       "rid1    1.916667\n",
       "Name: len_human_jargon, dtype: float64"
      ]
     },
     "execution_count": 112,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# what is the mean length of human jargon\n",
    "low_prec_abstracts.groupby('reader_id')['len_human_jargon'].mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "reader_id\n",
       "rid0    6.125000\n",
       "rid1    9.833333\n",
       "Name: len_gpt_jargon, dtype: float64"
      ]
     },
     "execution_count": 113,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# what is the mean length of gpt jargon\n",
    "low_prec_abstracts.groupby('reader_id')['len_gpt_jargon'].mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "reader_id\n",
       "rid0    1.5\n",
       "rid1    2.0\n",
       "Name: len_human_jargon, dtype: float64"
      ]
     },
     "execution_count": 111,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# what is the median length of human jargon \n",
    "low_prec_abstracts.groupby('reader_id')['len_human_jargon'].median()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "reader_id\n",
       "rid0    7.0\n",
       "rid1    9.0\n",
       "Name: len_gpt_jargon, dtype: float64"
      ]
     },
     "execution_count": 114,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# what is the median length of gpt jargon \n",
    "low_prec_abstracts.groupby('reader_id')['len_gpt_jargon'].median()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.16"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}