Class Meeting for 10-710 09-06-2011

This is one of the class meetings on the schedule for the course Structured Prediction 10-710 in Fall 2011.

HMMs, MaxEnt Taggers, and MEMMs

• Slides in Powerpoint
• Slides in PDF

Required Readings

No additional readings from Thursday, but I will focus on this paper:

• Frietag 2000 Maximum Entropy Markov Models for Information Extraction and Segmentation

The Borthwick et al paper I discussed last Thursday is

• Exploiting diverse knowledge sources via maximum entropy in named entity recognition, by A. Borthwick, J. Sterling, E. Agichtein, R. Grishman. In Proceedings of the sixth workshop on very large corpora, 1998.

Background Readings

• LSP: section 3.4 and appendix D through D.2. To understand why these models are often called "max ent," read appendix C.
• Max ent for NLP: Berger et al., 1996, "A Maximum Entropy Approach to Natural Language Processing" (Wiki)
• Max ent for NLP: Ratnaparkhi, 1996, "A Maximum Entropy Model for Part-of-Speech Tagging"
• Max ent for language modeling: Lau et al., 1993, "Adaptive Language Modeling Using the Maximum Entropy Principle" (Wiki)