jupyter-notebooks

'Real-life' examples

Here we present some examples of solving problems similar to real ones using vector spaces and Vektonn.

Use cases

1. Hotels. Task: save the user's time while searching for hotels. This example uses vectorized user reviews to find hotels.

2. Price Match Guarantee. Task: help the seller establish a competitive price for their product on the marketplace.

3. Books. Task: find similar books by user reviews. This example demonstrates usage of sparse vectors.

Prerequisites

To run notebooks presented in this repository locally you will need to install:

• Docker
• Python >= 3.7
• Jupyter Notebook

To download datasets for provided examples you will need to register account on Kaggle, install Kaggle CLI, and prepare authentication token to use its API.

To run example notebooks

1. Clone this repository into /path/to/vektonn-examples dir:

   git clone https://github.com/vektonn/vektonn-examples.git /path/to/vektonn-examples

2. Run Jupyter Notebook from that directory:

   jupyter notebook --notebook-dir=/path/to/vektonn-examples

3. Select notebook to run (e.g. jupyter-notebooks/hotels/hotels.ipynb).

4. Download necessary dataset for selected notebook:

   /path/to/vektonn-examples/jupyter-notebooks/hotels/download-dataset.sh

5. Run selected notebook using Jupyter Notebook GUI.