Robust Reading: Identification and Tracing of Ambiguous Names

A given entity, representing a person, a location or an organization, may be mentioned in text in multiple, ambiguous ways. Understanding natural language requires identifying whether different mentions of a name, within and across documents, represent the same entity. We develop an unsupervised learning approach that is shown to resolve accurately the name identification and tracing problem. At the heart of our approach is a generative model of how documents are generated and how names are "sprinkled" into them. In its most general form, our model assumes: (1) a joint distribution over entities, (2) an "author" model, that assumes that at least one mention of an entity in a document is easily identifiable, and then generates other mentions via (3) an appearance model, governing how mentions are transformed from the "representative" mention. We show how to estimate the model and do inference with it and how this resolves several aspects of the problem fro...

• Type: Conference
• Year: 2004
• Where: NAACL
• Authors: Xin Li, Paul Morie, Dan Roth
• Keywords: Ambiguous Ways, Generative Model, NAACL 2004, NAACL 2007, Unsupervised Learning Approach