Finding similar experts

The task of finding people who are experts on a topic has recently received increased attention. We introduce a different expert finding task for which a small number of example experts is given (instead of a natural language query), and the system's task is to return similar experts. We define, compare, and evaluate a number of ways of representing experts, and investigate how the size of theinitial example set affects performance. We show that morefine-grained representations of candidates result in higher performance, and larger sample sets as input lead to improved precision.