Paper: Name Perplexity

ACL ID N09-2039
Title Name Perplexity
Venue Human Language Technologies
Session Short Paper
Year 2009

The acuracy of a Cros Document Corefer- ence system depends on the amount of context available, which is a parameter that varies greatly from corpora to corpora. This paper presents a statistical model for computing name perplexity clases. For each perplexity clas, the prior probability of coreference is estimated. The amount of context required for coreference is controled by the prior corefer- ence probability. We show that the prior prob- ability coreference is an important factor for maintaining a god balance betwen precision and recal for cros document coreference sys- tems.