Login
Computer Science @ Rochester
Monday, November 29, 2004
11:00 AM
CSB 209
Andrew McCallum
U. Massachusetts, Amherst
Toward Unified Graphical Models of Information Extraction and Data Mining
Although information extraction and data mining appear together in many applications, their interface in most current systems would better be described as serial juxtaposition than as tight integration. Information extraction populates slots in a database by identifying relevant subsequences of text, but is usually not aware of the emerging patterns and regularities in the database. Data mining methods begin from a populated database, and are often unaware of where the data came from, or its inherent uncertainties. The result is that the accuracy of both suffers, and significant mining of complex text sources is beyond reach.

In this talk I will describe work in relational, undirected graphical models for information extraction and data mining. After briefly introducing conditional random fields, I will describe three pieces of work that make steps toward joint models which make more unified decisions: (1) an extension of CRFs to factorial state representation, enabling simultaneous part-of-speech tagging and noun-phrase segmentation, (2) a random field method for noun co-reference resolution that has strong ties to graph partitioning, (3) an example of improving co-reference by modeling uncertainty about segmentation, and improving segmentation by leveraging co-reference.

If there is time, I'll end by describing some very recent work on a new method of social network analysis on message data that models not only the presence of links between people, but also the words transmitted on those links. Results on the Enron email corpus, provide evidence that clearly relevant topics are discovered, and also that the model better predicts peoples' roles.

Joint work with colleagues at UMass: Charles Sutton, Ben Wellner, Khashayar Rohanimanesh, Michael Hay, Fuchun Peng, Xuerui Wang and Andres Corrada.

Bio: Andrew McCallum is an Associate Professor at University of Massachusetts, Amherst. He was previously Vice President of Research and Development at WhizBang Labs, a company that used machine learning for information extraction from the Web. In the late 1990's he was a Research Scientist and Coordinator at Justsystem Pittsburgh Research Center, where he spearheaded the creation of CORA, an early research paper search engine that used machine learning for spidering, extraction, classification and citation analysis. He was a post-doctoral fellow at Carnegie Mellon University after receiving his PhD from the University of Rochester with Dana Ballard in 1995. He is an action editor for the Journal of Machine Learning Research. For the past nine years, McCallum has been active in research on statistical machine learning applied to text, especially information extraction, document classification, finite state models, and semi-supervised learning.

Web page: http://www.cs.umass.edu/~mccallum.