Movie-recommendation-system-content-based-filtering

Description

This project is a Movie Recommendation System built using content-based filtering techniques. It recommends movies based on the similarity of their features (such as genres, keywords, cast, and crew) to a user-selected movie. The system is designed to help users discover movies that match their preferences based on a movie they already like.

The dataset used in this project was sourced from Kaggle, and it contains detailed information about movies such as titles, genres, cast, and more.

Tech Stack

The recommendation engine is implemented using Python and the following technologies:

• Streamlit: For creating an interactive web interface where users can input a movie title and view the recommended movies.
• NLTK(Natural Language Tool Kit): For text processing and feature extraction from movie descriptions.
• Pandas: For data manipulation and analysis.
• NumPy: For numerical operations and handling multi-dimensional arrays.
• Scikit-learn: For implementing content-based filtering and cosine similarity.

Detailed Working

Data preparation: There are two datasets which are grouped together on "title" feature and several features are removed from the dataset, left with (movie_id, title, overview, genres, keywords, cast and crew). Cleaning of the data, and for feature-extraction CountVectorizer() is used to pick 5000 tags which will be used to find the similarity (using cosine_similarity()).

Furthermore, steming is used to remove redundancy from the tags.

Using bag-of-word to get the Term-Frequency.

Using Streamlit to create a web interface in which user can select a movie which he/she has watched in history and he/she can get Recommendations based on this movie. In total, 9 Movies are recommended based on the cosine similarity. The top 9 valued movies fetched.

Demo

Installation

Follow these steps to set up the Movie Recommendation System on your local machine.

1. Clone the Repository

Start by cloning the repository to your local machine:

git clone https://github.com/your-username/movie-recommendation-system.git
cd movie-recommendation-system

2. Creating Virtual environment

It's recommended to create a virtual environment to manage dependencies:

python -m venv env-name

Activating Virtual environment:

Windows

./env-name/Scripts/activate

macOS/Linux

source env-name/bin/activate

3. Install dependencies

Install all the required packages using pip and the requirement.txt file:

pip install -r requirement.txt

This provides a comprehensive guide for users to set up your project correctly on their local machines. Adjust the directory paths and any other specific details based on your actual project setup.

How to use

After cloning the repository,

• Download the datasets from Kaggle and keep them in /inputs/ folder.

  Update the src/config.py

    TRAININGFILE_CREDIT = "Path to credits dataset"
    TRAININGFILE_MOVIE = "Path to movies dataset"
  

  Similarly update the paths where you want to keep the .pkl files generated by the system.

    MOVIES_PKL = "path where you want the **Movies.pkl** file to get created"
    SIMILARITY_PKL = "path where you want the **Similarity.pkl** file to get created"
  

  It is advised to keep the Movies.pkl and Similarity.pkl in the /inputs/ folder.

  API key

  Register on TMDB to generate your API key.

    Go to profile > setting > API
  

  Use the obtained key in place of

    APIREQUEST = "https://api.themoviedb.org/3/movie/{}?api_key=<YOUR-API-KEY>"
  

• Open a terminal in the project folder and activate Virtural-environment using

    ./env-name/Scripts/activate
  

  traverse to the /src directory and run the command

    python train.py       
  

  After the execution of train.py

    streamlit run app.py
  

• Some links will appear in the terminal like

    Local URL: http://localhost:8500
    Network URL: http://10.1.***.***:8500
    External URL: http://14.139.***.***:8500
  

• Upon clicking you will be directed to browser.

Datasets

1. Dataset Source

• Name: tmdb_5000_credits and tmdb_5000_movies
• Source: Kaggle
• Description: This dataset contains metadata for over 5,000 movies, including details such as titles, genres, cast, etc.

2. Dataset Structure

• Movies dataset:

  • budget: The budget of the movie.
  • genres: The genre(s) associated with the movie.
  • homepage: The official website of the movie.
  • id: The unique identifier for the movie.
  • keywords: Keywords describing the main themes or elements of the movie.
  • original_language: The original language in which the movie was released.
  • original_title: The original title of the movie.
  • overview: A brief summary or description of the movie's plot.
  • popularity: A popularity score assigned to the movie.
  • production_companies: The companies involved in the production of the movie.
  • production_countries: The countries where the movie was produced.
  • release_date: The date when the movie was released.
  • revenue: The total revenue generated by the movie.
  • runtime: The duration of the movie in minutes.
  • spoken_languages: The languages spoken in the movie.
  • status: The release status of the movie (e.g., Released, Post Production).
  • tagline: A short promotional phrase or tagline for the movie.
  • title: The title of the movie.
  • vote_average: The average rating the movie received.
  • vote_count: The number of votes the movie received.

• Credits dataset:

  • movie_id: The unique identifier for the movie.
  • title: The title of the movie.
  • cast: A list of the main actors in the movie.
  • crew: A list of key crew members involved in the movie, such as directors and producers.

3. Preprocessing

• Steps Taken:

  • Removed rows with missing values in crucial columns like title and overview.
  • Converted all text data to lowercase for consistency.
  • Extracted tags from the columns using NLTK.

• Tools Used:

  • Pandas: For data manipulation.
  • NLTK: For text processing and keyword extraction.

Project Structure

Explanation

• inputs/: Contains dataset files required for the project.
  • movies_dataset
  • credits_dataset
  • Keep the .pkl files, generated from the model training.
• notebooks/:
  • raw_model.ipynb: Notebook for initial model experiments and development.
• src/: Contains the source code for the project.
  • app.py: The main script for running the application.
  • train.py: Script to train the recommendation model.
  • models.py: Defines the recommendation models and their functions.
  • config.py: Configuration file with settings for the project.
  • model_dispatcher.py: Manages model selection and dispatching.
• requirements.txt: Lists all Python dependencies required for the project.

Features

• Content-based filtering.
• Cosine-similarity.
• Displays movie titles along with thumbnails.
• Recommend Most Related Movies.