Movie Recommendation and Sentiment Analysis

This project combines a movie recommendation system with a sentiment analysis tool, allowing users to discover new movies based on their preferences and analyze their reviews. Below is a detailed guide to the implementation and functionality.

Features

1. Movie Recommendation System

   • Uses TF-IDF vectorization to process movie features (Title, Genre, Director, Cast, Description).
   • Employs cosine similarity to find and suggest movies similar to the selected title.

2. Sentiment Analysis

   • Predicts sentiment of user reviews using a hybrid approach:
     • Short Reviews: Analyzed with VADER (lexicon-based sentiment analysis).
     • Long Reviews: Preprocessed and evaluated using a trained XGBoost model.

3. Interactive User Interface

   • Built with Streamlit.
   • Includes features like movie dropdown selection, review input, and visual recommendations with sentiment analysis.

4. Custom Styling

   • Dynamic word cloud background for the UI.
   • Styled buttons, text boxes, and layouts for a user-friendly experience.

---

Setup Instructions

1. Prerequisites

Ensure the following are installed:

• Python 3.8+
• Required libraries (install via pip install -r requirements.txt):

  streamlit
  pandas
  pickle
  sklearn
  nltk
  matplotlib
  wordcloud
  fuzzywuzzy
  xgboost
  base64
  

2. Files Required

The following files are essential for the application to run:

• movies1.pkl: Contains movie data (Title, Genre, Director, Cast, Description, etc.).
• tfidf_matrix_recommendation.pkl: Precomputed TF-IDF matrix for movie recommendations.
• tfidf_vectorizer_sentiment.pkl: Vectorizer for sentiment analysis.
• best_xgb_model.pkl: Trained XGBoost model for sentiment analysis.

3. NLTK Setup

Run these commands to download necessary NLTK components:

import nltk
nltk.download('stopwords')
nltk.download('vader_lexicon')
nltk.download('punkt')
nltk.download('wordnet')

4. Run the Application

Launch the app using Streamlit:

streamlit run app.py

---

Functionality

1. Preprocessing User Reviews

• Short reviews: Minimal preprocessing (lowercase, remove URLs and symbols).
• Long reviews: Tokenized, stopwords removed, lemmatized, and detokenized.

2. Hybrid Sentiment Prediction

• Short reviews: Sentiment calculated using VADER.
• Long reviews: Sentiment predicted using the XGBoost model with the following classes:
  • Negative (0)
  • Neutral (1)
  • Positive (2)

3. Movie Recommendations

• Fuzzy Matching: Matches user-input movie title to dataset.
• Similarity Scoring: Finds similar movies based on cosine similarity and a similarity threshold.

4. Word Cloud

• Dynamically generated using all movie titles in the dataset.
• Set as the background of the Streamlit application.

---

How to Use

1. Select a movie from the dropdown or type the title.
2. Enter your review for the selected movie.
3. Click Get Recommendations.
4. View:
   • Sentiment analysis of your review.
   • Recommended movies with details (poster, genre, director, cast, description).
   • Sentiment of reviews for recommended movies.

---

Example Workflow

1. Select the movie "Inception".
2. Enter your review: "An absolute masterpiece with brilliant storytelling."
3. Get output:
   • Your Review Sentiment: Positive
   • Recommended Movies:
     • Title: Interstellar
     • Genre: Sci-Fi
     • Review Sentiment: Positive

---

Additional Notes

• Recommendations are filtered by a similarity threshold to ensure relevance.
• All processing is optimized for both short and long reviews.
• Background and UI styling enhance user experience.