Automatic hierarchical categorization of research expertise using minimum information

Gustavo O. de Siqueira

Sergio Canuto

pp. 103-115

in: Jaap Kamps, Giannis Tsakonas, Yannis Manolopoulos, Lazaros Iliadis, Ioannis Karydis (eds), Research and advanced technology for digital libraries, Berlin, Springer, 2017

Abstract

Throughout the history of science, different knowledge areas have collaborated to overcome major research challenges. The task of associating a researcher with such areas makes a series of tasks feasible such as the organization of digital repositories, expertise recommendation and the formation of research groups for complex problems. In this paper we propose a simple yet effective automatic classification model that is capable of categorizing research expertise according to a hierarchical knowledge area classification scheme. Our proposal relies on discriminative evidence provided by the title of academic works, which is the minimum information capable of relating a researcher to its knowledge area. We also evaluate the use of learning-to-rank as an effective mean to rank experts with minimum information. Our experiments show that using supervised machine learning methods trained with manually labeled information, it is possible to produce effective classification and ranking models.