Skip to content

Demonstration of vectorization of movies, recommendation using collaborative filtering and classification.

Notifications You must be signed in to change notification settings

iamjagdeesh/Movie-Data-Analysis-Recommendation-Classification

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 

Repository files navigation

Vector space modeling of MovieLens & IMDB movie data

This project demonstrates the following Multimedia and Web Database concepts using MovieLens and IMDB movie data:

  • Multi-dimensional data representation
  • Dimensionality reduction using SVD, LDA and PCA
  • Feature significance using TF and TF-IDF measures
  • Latent semantic analysis using SVD, PCA and LDA
  • Tensor decomposition analysis using CP decomposition
  • Node ranking/significance using Personalized PageRank
  • Movie Recommendation system using SVD, LDA, PCA, Tensor decomposition and PageRank
  • Probabilistic Relevance Feedback
  • Multi-dimensional index structures using LSH
  • Nearest Neighbor search using LSH
  • Nearest Neighbor based relevance feedback using query rewrite technique
  • Movie classification using SVM, Decision Trees and KNN classifiers

Phase 1 - Multi-dimentional data representation, Feature significance using TF and TF-IDF measures. This phase is included in phase 2 as phase 2 is a continuation of it.

Phase 2 - Dimentionality reduction (SVD, LDA and PCA), Latent semantics analysis (SVD, PCA and LDA), Tensor decomposition analysis using CP decomposition, Node ranking/significance using Personalized PageRank

Phase 3 - Movie Recommendation system (SVD, LDA, PCA, Tensor decomposition and PageRank), Probablistic Relevance Feedback, Multi-dimensional index structures using LSH, Nearest neighbor search using LSH, Nearest Neighbor based relevance feedback using query rewrite technique, Movie classification (SVM, Decision Trees and KNN classifiers)

For more details please check the project specification file present under each phase directory.