We address the Named Entity Disambiguation (NED) prob- lem for short, user-generated texts on the social Web. In such settings, the lack of linguistic features and sparse lex- ical context result in a high degree of ambiguity and sharp performance drops of nearly 50% in the accuracy of conven- tional NED systems. We handle these challenges by develop- ing a model of user-interest with respect to a personal knowl- edge context; and Wikipedia, a particularly well-established and reliable knowledge base, is used to instantiate the proce- dure. We conduct systematic evaluations using individuals’ posts from Twitter, YouTube, and Flickr and demonstrate that our novel technique is able to achieve substantial per- formance gains beyond state-of-the-art NED methods.

Top