License

README.md

@@ -1 +1,43 @@
-# VictorianLit
+# VictorianLit
+
+Victorian Lit Dataset for Machine Learning-Based Sentiment Analysis of Victorian Literary Texts | by Hoyeol Kim
+
+---
+### Download: [VictorianLit (Kaggle)](https://www.kaggle.com/elibooklover/victorianlit/download)
+
+##### You can download the VictorainLit dataset directly by using the following URL: 
+```
+https://github.com/elibooklover/VictorianLit/raw/master/VictorianLit.csv
+```
+Here is example code to load the VictorianLit dataset at Google Colab through Google Drive: 
+```
+from google.colab import drive
+drive.mount('/content/drive/')
+```
+```
+df=pd.read_csv('drive/My Drive/Colab Notebooks/VictorianLit.csv')
+df.head()
+```
+---
+### Dataset
+There are two columns: sentences and label. The VictorianLit has five labels based on sentiment: 0 (very negative), 1 (negative), 2 (neutral), 3 (positive), 4 (very positive).
+
+The VictorianLit dataset, which has 53,826 rows and 2 columns, consists of five different novels from the Victorian era: Charles Dickens' *Little Dorrit* and *Oliver Twist*, Elizabeth Gaskell's *North and South*, George Eliot's *Adam Bede*, and Mary Elizabeth Braddon's *Lady Audley's Secret*. The maximum sentence length of the VictorianLit dataset is 372.
+---
+### Test Results
+The VictorianLit dataset was tested with the [BERT-Base](https://github.com/google-research/bert) model released by [Google Research](https://github.com/google-research). The BERT-Base, Uncased model (12-layer, 768-hidden, 12-heads, 100M parameters) was run with the VictorianLit dataset in order to validate the dataset.
+
+I divided the VictorianLit dataset into a training set (80%) and a validation set (20%). For pre-training and fine-tuning BERT on sentiment analysis, the following hyperparameters and training environments were set:
+
+tokenizer: BertTokenizer
+max_sequence_length: 372
+batch_size: 16
+model_name: BERT-base, Uncased (12-layer, 768-hidden, 12-heads, 110M parameters)
+learning_rate: 1e-5
+epochs (for fine-tuning): 4
+GPU: Tesla T4
+
+The accuracy is XX. Learning loss is. If the batch_size was larger, the accuracy would be higher. If your GPU ram is enough to cover the large batch_size, I recommend you set the batch_size 64 or 128.
+---
+### Feedback
+The VictorianLit dataset will be continuously updated, added, and tested. Please feel free to suggest changing sentiment values with supporting statements.