Matching papers with reviewers is a complicated task,
with many sub-problems. Conference chairs must solve a
large optimization problem involving constraints on the num-
ber of reviewers per paper and the number of papers per re-
viewer. One of the most important elements of the process,
however, is modeling the expertise of a given reviewer with
respect to the topical content of a given paper. This task
is related to expert nding," an area that has received in-
creased interest in recent years in the context of the TREC
Enterprise Track. In addition, for several years researchers
in articial intelligence have sought to automate, or at least
streamline, the reviewer matching process.