As DNA sequencing projects complete more genomes, the emphasis is
shifting to understanding the functions of and relationships between genes
and related molecules. This recent boom in biomedical research has
improved both the breadth and depth of biomolecular knowledge with each
new publication (much of this literature being in electronic format).
While this is certainly beneficial researchers, it also creates a load of
information too overwhelming for a research group --- let alone an
individual --- to follow entirely.

  To help promote successful research directions, we would like to
have intelligent machines that can automatically cull the wealth of
electronic literature, annotate documents, and build structured
knowledge-bases about interactions between proteins, genes, RNA, cell
loci, etc. This information can populate more structured reference
databases, be analyzed with data mining techniques to discover new
relationships, and even launch new research directions.

  The first step in such a system, however, is identifying the terms
of interest (e.g. proteins) in these texts. In this talk I will address
the task of protein name recognition, discuss some of my previous work on
the problem, and outline my current line of research (with some
preliminary results).