Query-Recommendation-System

Project Description

This project implements a Query Recommendation System that generates synthetic data and applies collaborative filtering techniques to recommend queries based on user preferences and similarities. The system simulates a dataset of books, users, and user queries, and then makes query recommendations by comparing past user interactions and query patterns.

Features

Synthetic Data Generation:

A)Synthetic Data Generation: Generates datasets for books, users, and user queries, which serve as input for the recommendation engine. Datasets include:

1. Book Information: Publisher, Title, Author, and Year of Publication.
   
2. User Information: A list of user IDs.
   
3. User Queries: Simulated queries that form the basis for recommendation predictions.
   

B)Query-Based Collaborative Filtering: The system employs collaborative filtering methods to recommend queries by: User-User Similarity: Identifies users with similar query patterns. Query-Query Similarity: Recommends new queries based on their similarity to previous user queries.

Data Generation

The Generator.py script creates synthetic data in three main parts:

• Book Dataset: Information such as Publisher, Title, Author, and Year of Publication.
• User Dataset: A list of users, each identified by a unique User ID.
• Query Dataset: Simulated user queries that ask for books based on attributes like publisher, title, and year of publication.

Query Recommendation Engine

The utils.py file contains the key functions for the query recommendation system:

• User Similarity Calculation: Computes a similarity matrix between users based on their queries.
• Query Similarity Calculation: Calculates similarity between different user queries.
• Collaborative Filtering Predictions:
  1)User-Based Recommendations: Predicts new queries for a user by identifying similar users and their query patterns.
  2)Query-Based Recommendations: Recommends queries that are similar to past queries made by the user.

Jupyter Notebooks

part A.ipynb:

Handles data loading, preprocessing, and setting up the basic recommendation system infrastructure. Implements basic query and user similarity calculations to prepare for collaborative filtering.

part A1.ipynb:

Continues from part A by refining the model, running more detailed similarity computations, and possibly fine-tuning model parameters. Includes testing of the recommendation system with different scenarios and metrics like RMSE.

part B.ipynb:

Focuses on evaluating the performance of the query recommendation system using a testing dataset. Implements a final evaluation phase to assess the accuracy of the recommendations, including error metrics (e.g., RMSE). How to Run