Entity disambiguation in anonymized graphs using graph kernels

Linus Hermansson, Tommi Kerola, Fredrik Johansson, Vinay Jethava, Devdatt Dubhashi
2013 Proceedings of the 22nd ACM international conference on Conference on information & knowledge management - CIKM '13  
In recent years, the explosion of available online information has brought forth new data mining applications into the spotlight, such as automated querying about realworld entities. This requires extraction of identifiers such as names and places from text. The problem, however, is complicated by the non-uniqueness of identifiers. A motivating example is the name Chris Anderson, which could either refer to Chris Anderson, the curator of TED Talks, or Chris Anderson, the former editor-in-chief
more » ... f WIRED Magazine. Both individuals work in overlapping fields, and deciding whom is referred to could be a difficult task, even when considering context. Correctly identifying and resolving such ambiguous identifiers is crucial for enabling such applications to advance from the research lab into practical usage.
doi:10.1145/2505515.2505565 dblp:conf/cikm/HermanssonKJJD13 fatcat:vazawm2ix5b5zbw2hvkzdvzcju