Sentiments in movies

Inspiration

• Movies can have a significant impact on people's emotions

• Movies with positive emotions can have a positive effect on people's emotions and vice versa

• We need to know which movies make people feel positive or negative

• Fortunately, we have access to immense databases of film reviews like IMDb

What it does

• Analyse the emotions in movie reviews retrieved from IMDb

• Summary of the emotions in a certain film’s reviews

• Auto clustering similar feeling films (unsupervised learning)

• Amazon Alexa can tell you the movie you want to see today based on your feelings

How we built it

• We used a web crawler to retrieve movie reviews from http://www.imdb.com/

• Cluster movies based on reviews (transformed into customized TF-IDF based vector) with unsupervised learning techniques provided by scikit-learn machine-learning library

• Analyse emotions in movie comments with IBM Watson

• Recommend movies based on users' emotions

Challenges we ran into

• Slow web indexing when we tried to retrieve movie reviews from http://www.imdb.com/

• Large data set and slow I/O speed in Python

• AWS Alexa Skills service of our main account stopped working after midnight

Accomplishments that we're proud of

The Lord of the Rings film series fall in the same cluster!

What we learned

Dr. Strangelove and Casablanca are the same kind of movies!

What's next for our project

Built With

• amazon-alexa
• amazon-web-services
• aws-lambda
• machine-learning
• python
• scikit-learn
• web-crawler