renish-charaniya / instagramfashionblogger Goto Github PK

This is an Instagram Fashion Blogger recommendation app built on deep learning and NLP.

The original idea is to make customized recommendations of Instagram fashion bloggers through natural language processing. Whenever the user type in one fashion keyword in his/her mind, such as "party", "paris", or "shoes", this app will search in the blogger posts database for the best matches.

A word2vec deep learning neural network has been trained on a 15k-word casual corpus, and validated by a relational word benchmark. A blogger corpus with 0.25 million words has been established by web scrapping. 1300 bloggers recommended by 500 fashion articles are included in this database.

Basic stats are visualized by d3 using nvd3 package.

Try the app at: https://instagram-blogger.herokuapp.com/