An Algorithm that Learns What's in a Name

Daniel Bikel, Richard Schwartz, and Ralph M. Weischedel

BBN Systems & Technologies
70 Fawcett Street
Cambridge MA 02138

email:  weisched@bbn.com

Abstract

In this paper, we present a hidden Markov model that learns to
recognize and classify names, dates, times, and numerical quantities.
We have evaluated the software IdentiFinder=81 in English (based on
MUC-6, MUC-7, and broadcast news data) and in Spanish (based on data
distributed through MET-1), and on speech input (based on broadcast
news).  We report results here on standard materials only to quantify
performance on data available to the community, namely, MUC-6 and
MET-1.  Results have been consistently better than reported by any
other learning algorithm.  IdentiFinder's performance is competitive
with approaches based on handcrafted rules on mixed case text and
superior on text where case information is not available.

We also present a controlled experiment showing the effect of training
set size on performance, demonstrating that as little as 100,000 words
of training data is adequate to get performance around 90% on
newswire.  Though we present our understanding of why this algorithm performs so
well on this class of problems, we believe that significant improvement
in performance may still be possible.