Book Recommender System

• A ML based Book Similarity Recommender for promoting books on digital platforms
• Book recommending systems drive business growth by boosting sales and revenue through personalized suggestions
• By recommending relevant and complementary books, it makes shopping experience more enjoyable and tailored to individuals
• This personalization also facilitates the discovery of new books, leading to greater overall engagement and loyalty

Overview

• Book Recommender is designed to recommend books based on their textual content and vector similarity
• This project leverages the Bag of Words (BoW) model for tokenization and stores book data as vectors in DataStax Astra DB, a cloud - native database service optimized for scalable and efficient data management

Key Features

• Textual Analysis with Bag of Words: Utilizes the Bag of Words model to tokenize and represent book descriptions and categories as vectors
• Vector Storage: Stores book vectors in Astra DB for efficient retrieval and similarity searching
• Similar Book Recommendations: Recommends books similar to a given book based on vector similarity, using Approximate Nearest Neighbors (ANN) search

Technologies Used

• Bag of Words Model: For tokenizing and converting book text data into numerical vectors
• DataStax Astra DB: A cloud-native NoSQL database service for storing and managing vector data with minimal operational overhead
• Spring Boot: Framework for building the RESTful API to interact with the recommendation system
• Apache Cassandra: Underlying database technology for Astra DB, known for its high scalability and performance
• Java: Programming language used for implementing backend logic

How It Works

• Tokenization with Bag of Words: The text data of books, such as titles and descriptions, is tokenized using the Bag of Words model. This converts textual data into numerical vectors representing word frequencies
• Storing Vectors in Astra DB: These vectors are stored in Astra DB, allowing for efficient querying and similarity searches
• Similarity Search: When a user requests recommendations for a specific book, the system retrieves the corresponding vector and performs an approximate nearest neighbor search to find and recommend similar books
• Return Recommendations: The system returns a list of the top 3 most similar books based on vector similarity, excluding the original book from the results

API Endpoint

• GET /promotions/books/{bookName}
• Description: Retrieves book recommendations similar to the one specified by {bookName}
• Response: A list of recommended books or a 404 error if no recommendations are found

Output

Setup and Installation

• Clone the repository
• Configure Astra DB credentials and set up your environment
• Build and run the Spring Boot application to start the recommender service

Contribution

Contributions to enhance the recommender system or add new features are welcome! Please fork the repository, make your changes, and submit a pull request

License

This project is licensed under the MIT License. See the LICENSE file for details