Finding the Right Experts and Learning to Rank Them by Relevance: Evaluation of a Semi-automatically Generated Ranking Function
Felix Beierle, Felix Engel, Matthias Hemmje · 2014
Abstract. A framework for expert searching developed at the Univer-sity of Hagen supports the use of contextual factors motivated in the field of expertise seeking. A user can query the system with skills which the person to be found should be an expert in. The result is a list of users from the searched knowledge base, ranked by relevance for the given query. In this paper, we address the semi-automatic generation of train-ing data for the learning-to-rank library that does the relevance ranking. We focus on evaluating the quality of the ranking function. 1