LaughterLens - Personalized Joke Recommendation System

📌 Project Description

LaughterLens is a collaborative filtering-based recommendation system designed to predict user preferences for jokes. Utilizing cosine similarity between users' ratings, the system suggests jokes that are most likely to be enjoyed by a particular user. It also provides a baseline comparison using a random recommender system for evaluating prediction accuracy. This project aims to offer a personalized joke recommendation experience to users based on their historical preferences.

📂 Dataset

The project uses the Jester Dataset (2006-15) which includes user ratings for jokes. The key aspects of the dataset are:

Contains user ratings where:

• Rows represent individual users.
• Columns represent jokes (150 jokes in total).
• 99 indicates the joke was not rated by the user.
• The first column represents the number of jokes rated by each user.

Contains the text of the jokes corresponding to the joke indices.

📌 Features

• Collaborative Filtering: Uses cosine similarity to find users with similar preferences and predicts ratings based on their ratings.
• Baseline Comparison: Implements a random recommender system for comparative analysis.
• Top-K Similar Users: Identifies the top 2,000 users most similar to a particular user.
• Joke Recommendations: Provides top-N joke recommendations for a specified user.
• Evaluation Metrics: Uses Mean Absolute Error (MAE) for evaluating prediction performance.

📂 Project Structure

│── LaughterLens
│   │── README.md                  # Project documentation
│   │── FINAL jester 2006-15.csv    # Joke ratings dataset
│   │── Dataset3JokeSet.csv         # Joke text dataset
│   │── RecommendationSystem.ipynb        # Main Python script for recommendation system
│   │── requirements.txt            # Dependencies

🔍 Requirements

• Python 3.x
• pandas
• numpy
• scikit-learn

Install the dependencies via:

pip install -r requirements.txt

🚀 Usage

• Place the dataset files (FINAL jester 2006-15.csv and Dataset3JokeSet.csv) in the same directory as recpmmendationSystem.ipynb.
• Run the joke_recommender.py file to train the model and generate recommendations.

📊 Evaluation

The model is evaluated using Mean Absolute Error (MAE) on the test set. Results are also compared against a baseline random recommender system for performance analysis.

📌 How It Works

Data Preprocessing:

• Replace 99 values with NaN to identify unrated jokes.
• Filter active users who have rated at least 80 jokes.
• Split the data into training and testing sets.

Collaborative Filtering:

• Compute user-user similarity using cosine similarity.
• Predict ratings for jokes based on the weighted average of ratings from top-K similar users.

Baseline System:

-Compare the performance with a random recommender system.

Top-K User Recommendation:

• Identify and display the top 2,000 most similar users for each user.

Top-N Joke Recommendations:

• Recommend top jokes based on predicted ratings.

📌 Future Improvements

• Implement matrix factorization techniques (e.g., SVD, ALS) for improved accuracy.
• Add content-based filtering using joke text data.
• Develop a web-based interface for interactive joke recommendations.

💡 Acknowledgments

This project utilizes the Jester Dataset, a collection of joke ratings gathered from various users.

The Representation of the flow of the system through Diagram:

Followed by the follow below: