From ContE to Entity Type Embeddings in Natural Language Processing

Author(s): Edward Ma

Originally published on Towards AI.

Photo by twinsfisch on Unsplash

In the previous story, TransE (Border et al., 2013) use a translating mechanism to transform subject (s) by the predicate (p) to object (o). Just like word2vec, we can compute “Queen” by using “King” + “Man” embeddings.

Word2vec Sample: King + Woman ~= Queen (source)

Moon et al. (2017) consider contextual relation to solving graph completion. Authors consider not only the triplet (s, p, o) but also outgoing relation types (from s) and incoming relation types (to o).

Contextual relation for a triple (s, p, o). (Moon et al., 2017)

When two entities (“Alex Guinness” and “Star Wars”) have relations… Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI