SPARK

• 

  Home Page

• 

  Movie Recommendation when we select the emotion fear

Inspiration

• The existing movie recommendation system works on the browsing history of an individual, ratings of the movie, genres, popularity and trends.
• Settling on the decision of watching films can't generally be understood by just picking something dependent on genre but can be contingent on one’s feelings.
• There is no movie recommendation based on human feelings.

What it does

• This is a web application which suggests the movie based on the emotion which the user has chosen.
• One special thing about this system is that its recommendations are tailored around users’ emotion at that moment.

How we built it

Finding the emotion of each movie

• Data Preprocessing
  • Create a dataset by using labelled data. This labelled data contains the comments along with the emotions inserted by ourselves which is saved as CSV file.
• Training : Deep learning model – LSTM
• Testing : Efficiency and accuracy of the model
• After that by using YouTube Data API, fetch the comments of the trailers of each movie. These comments are being passed to the model to predict the emotions alluded in each comment.
• The prediction of the model is stored in the database along with the name of the movie.

Web Application

• Here the users have the provisions to choose a movie of a certain emotion which he/she likes and get the list of the movies of the same.
• To give an instance, when a user chooses the emotion happy then the details of those movies having the same emotion are being fetched and listed out using TMDB API.

Challenges we ran into

• Creation of the labelled data.

Built With

• css3
• html5
• javascript
• mongodb
• pandas
• python
• tensorflow
• tmdbapi
• youtubedataapi