Toward Socially-Infused Information Extraction: Embedding Authors, Mentions, and Entities

We present a novel neural network model for entity linking that exploits distributed representations of users, mentions, and entities. • Our system leverages social network structures by utilizing entity homophily to improve entity disambiguation. • Our neural network model is on par with the tree-based model (Yang and Chang 2015) with surface features, but it is much easier to add additional information in the neural network model. EXPERIMENTS