A Conditional Random Field for Discriminatively-trained Finite-state String Edit Distance

Authors

Andrew McCallum, University of Massachusetts - Amherst
Kedar Bellare, University of Massachusetts - Amherst
Fernando Pereira, University of Pennsylvania

Publication Date

2005

Abstract

The need to measure sequence similarity arises in information extraction, object identity, data mining, biological sequence analysis, and other domains. This paper presents discriminative string-edit CRFs, a finitestate conditional random field model for edit sequences between strings. Conditional random fields have advantages over generative approaches to this problem, such as pair HMMs or the work of Ristad and Yianilos, because as conditionally-trained methods, they enable the use of complex, arbitrary actions and features of the input strings. As in generative models, the training data does not have to specify the edit sequences between the given string pairs. Unlike generative models, however, our model is trained on both positive and negative instances of string pairs. We present positive experimental results on several data sets.

Comments

This paper was harvested from CiteSeer

Recommended Citation

McCallum, Andrew; Bellare, Kedar; and Pereira, Fernando, "A Conditional Random Field for Discriminatively-trained Finite-state String Edit Distance" (2005). Computer Science Department Faculty Publication Series. 43.
Retrieved from https://scholarworks.umass.edu/cs_faculty_pubs/43