Virtual Assistant Summit Presentation – San Francisco - January 2016Access the presentation
The goal of People2Vec is to become a free, easy-to-use and open-source library designed to automatically extract semantic topics from people digital content such as social connectors (Twitter, Facebook, LinkedIn, Slack...), local connectors (emails, documents...) and online connectors (Web Page, RSS, blogs...).Join us, we need all the help we can get!
People2Vec is a natural extension of Word2Vec. Word2vec is an algorithm for constructing vector representations of words, also known as word embeddings. The vector for each word is a semantic description of how that word is used in a context, so two words that are used similarly will get similar vector representations. Once you map words into vector space, you can then use vector to find words that have similar semantics.
The approach here is to build a vector space (Bag-of-Contexts) of representations of people that can be subsequently used in many natural language processing applications and support further research like the similarity between people and any other semantic entity vector representations such as people, company, place, group...
This approach is perfectly designed to be used in Deep Learning.