Profile-Based Selection of Expert Groups

Abstract

In a wide variety of daily activities, the need of selecting a group of k experts from a larger pool of n candidates ($k<n$) based on some criteria often arises. Indicative examples, among many others, include the selection of program committee members for a research conference, staffing an organization’s board with competent members, forming a subject-specific task force, or building a group of project evaluators. Unfortunately, the process of expert group selection is typically carried out manually by a certain individual, which poses two significant shortcomings: (a) the task is particularly cumbersome, and (b) the selection process is largely subjective thus leading to results of doubtful quality. To address these challenges, in this paper, we propose an automatic profile-based expert group selection mechanism that is supported by digital libraries. To this end, we build textual profiles of candidates and propose algorithms that follow an IR-based approach to perform the expert group selection. Our approach is generic and independent of the actual expert group selection problem, as long as the candidate profiles have been generated. To evaluate the effectiveness of our approach, we demonstrate its applicability on the scenario of automatically building a program committee for a research conference.