1. Beautify Me
- Designed the concept of a beauty profile of a user that gathers their social and health related data by aggregating user profiles across multiple channels in order to ensure highly personalized beauty recommendations.
- Fully Integrated with Google Health API and Facebook Graph API.
2. Political Stance Detection using Deep Learning
- Explored several architectures including a word2vec model, Google’s pre-trained word embeddings and a LSTM based Autoencoder to detect a user’s political orientation in the Indian political scenario.
3. Move-R: Movie Recommender system using Collaborative Filtering
- Developed a user-based movie recommender system by employing Pearsons Correlation Coefficient to estimate similarity amongst users in order to suggest movies to new users.
4. Analysing Tweets and Networks during times of “Political Sensitivity”
- Analyzed user tweets using numpy and pandas to draw inferences about different user group interactions.
- Drew network graphs and studied centrality measures to understand network strengths and weaknesses.