Finding Expert via Topical Affinity Propagation on Thesis Advisors

• Published in: 2018 International Conference on Artificial Intelligence and Data Processing (IDAP) pp. 1 - 5
• Main Authors: Mungen, Ahmet Anil, Gundogan, Esra, Kaya, Mehmet
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.09.2018
• Subjects: Bayes methods; effect value; finding experts; Knowledge based systems; local approach method; Metadata; Reinforcement learning; Social network services; worship
• DOI: 10.1109/IDAP.2018.8620839

The academic world has expanded regarding the number of academicians and academic studies in the world day by day. Therewithal, the field of expertise is further elaborated by the development of science. Finding the correct reviewer or referee when academic work is to be conducted or evaluated, is crucial for evaluation of the study. In the academic literature, finding experts is one of the most popular subjects in recommendation systems. All of these studies deal with the works of the academicians, the characteristics of the works and the years of works. Just looking at the metadata of works is not enough to find an expert in the academic world alone. Another way to understand how an academics is an expert is to look at the thesis subject of academics' students in the other world advisor-student relationships. Expertise obtained through statistical approaches to our study is based on a propagation-based Topical Affinity Propagation (I-TAP) approach. Thus, expert candidates have examined not only their knowledge but also the expertise of students working area. The study was also scored with the TAP, and TAP data supported statistical methods to for expert finding. The proposed method is presented with a data set of 500,000 theses from universities of Turkey, and the experimental result is conferred relatively by F-Measure.