PhD Thesis Final Defense to be held on January 15, 2020, at 12:00

Photo Credit: Ilias Kanellos

The examination is open to anyone who wishes to attend (ICCS-NTUA building)


Abstract: The constantly increasing number of scientific publications affects researchers, students, academic hiring officials and search engines alike in discerning the high-impact works among them. Therefore, there is a need to develop methods to rank scientific papers. Despite a prolific literature on query-independent (or static) paper ranking algorithms, which aim to rank papers based on their impact, no systematic review of the field has been conducted. Past literature lacks in terms of defining impact, often failing to discern among short-term and long-term scientific impact. Further, no extensive experimental evaluation of the various proposed methods has been conducted.

This thesis examines impact-based paper ranking in terms of methods, search engine applications, and its relation to paper abstract readability. In short, the contributions of the thesis are as follows:

Long-term and short-term impact are formally defined and the various ranking and evaluation approaches encountered so far in the literature are examined and classified.

An extensive experimental evaluation is conducted to identify which proposed mechanisms perform best in ranking by short- and long-term impact.

Motivated by the observed improvement margin in ranking based on short-term impact, a novel method is proposed building on recent advances of network science.

The development of specialized and general academic search engines enhanced with short- and long-term impact-based ranking methods is presented.

Finally, paper abstract readability and its relation to paper impact are examined.

Supervisor: Yannis Vassiliou, Professor Emeritus

PhD student: Ilias Kanellos