Content-based movie recommendation system using cosine similarity measure

Movies are the primary entertainment source for people who are struggling with many issues in their everyday life. In choosing the best movie to go they are spending a lot of time checking the rating, reviews, etc. It is better to have a recommendation system that recommends the movies fast by considering all the factors. In this paper, a web application was presented that can recommend with one click by considering all the reviews, ratings, casting, genre, etc. There are so many organizations using the recommendation systems some of them are E-commerce applications such as Amazon, Flipkart, Meesho, etc. OTT Flatforms such as Netflix, Amazon prime, Aha, etc. By implementing the recommendation system these organizations improve their sales and followers by that they are earning more profits. In e-commerce and OTT platforms the recommendations will be done based on the user’s browse history and watched history. But if people want to see a movie in the theatre, they need a better recommendation system that can give a recommendation based on various parameters related to the movie selected by the user. In this movie recommendation system, we have considered the TMDB datasets from the Kaggle. Implementation of this project was done in python and by following natural language processing concepts by using so many python libraries. The TMDB dataset contains two different files, movies.csv, and credits.csv. These two files were merged based on the key attribute title to get one file. Various pre-processing techniques were implemented such as vectorization for feature selection, a bag of words used for finding the frequency of occurrences of various keywords that occurred in the dataset and stemming used to obtain the base form of various words. Finally, a recommendation system was developed based on content using a cosine similarity measure. A web application was developed in which the user can select a movie from the given list, by clicking on the recommendation it generates the images and titles of 5 recommended movies. In the future the web application can be converted into a mobile application, also it can be used in real-time.