Book Recommender Web App using Content Based Filtering

This web app recommends books based on user input. It uses streamlit and is deployed on streamlit community cloud.

Kaggle Notebook: https://www.kaggle.com/code/bhaskarbordoloi/book-recommendation-system?scriptVersionId=214481829

Dataset Link: https://www.kaggle.com/datasets/abdallahwagih/books-dataset

Snapshot

Approach

The idea is to build a recommender system using content-based filtering technique. This recommendation technique utilizes the item features to generate recommendations based on content similarity.

The system combines 'authors', 'categories' and 'description' features and uses cosine similarity to compute similarities between the books.

Word2Vec embedding technique is used to generate word embeddings and thus calculate simlarity between items.

Prerequisites

• Python 3.10 or above
• Streamlit
• A web browser for accessing the UI.

Installation Guide

1. Clone the repository

       git clone https://github.com/bhaskrr/book-recommender-system.git
       cd book-recommender-system

2. Install Dependencies

       pip install -r requirements.txt

3. Run the app

       streamlit run app.py

   A window will open in your default browser with the app. Otherwise, open your browser and visit http://localhost:8501/.

Usage

1. Open the app in the browser
2. Select a book from the dropdown
3. Click on the Get Recommendations button
4. View the generated recommendations

Known Issues and Limitations

1. Limited novelty and diversity.
2. Heavy reliance on feature engineering.