Friday 14:00–15:30 in Hall 7

Practical Word2vec in Gensim

Lev Konstantinovskiy

Audience level:


A hands-on introduction to the Natural Language Processing open-source library Gensim from its maintainers. This is not a theoretical maths talk - we will walk through the main function calls and discuss which business problems can be solved by word2vec and related techniques like Word Movers Distance.


We will walk through code that solves a practical problem for a restaurant review site - a user wants to see reviews covering all aspects of the restaurant(food, restaurant, price etc). We will load, clean and pre-proces a large amount of Yelp Restaurant reviews and see which reviews are similar even though they have no words in common.

Laptop with gensim and word2vec Google News model is required (2Gb download). More details on github