Analyzing the generalizability of the network-based topic emergence identification method

Sukhwan Jung,Aviv Segev,Francesco Osborne,Davide Buscaldi,Michael Cochez,Diego Reforgiato Recupero,Mehwish Alam,Harald Sack,Diego Refogiato Recupero

doi:10.3233/sw-212951

Abstract

Topic evolution helps the understanding of current research topics and their histories by automatically modeling and detecting the set of shared research fields in academic publications as topics. This paper provides a generalized analysis of the topic evolution method for predicting the emergence of new topics, which can operate on any dataset where the topics are defined as the relationships of their neighborhoods in the past by extrapolating to the future topics. Twenty sample topic networks were built with various fields-of-study keywords as seeds, covering domains such as business, materials, diseases, and computer science from the Microsoft Academic Graph dataset. The binary classifier was trained for each topic network using 15 structural features of emerging and existing topics and consistently resulted in accuracy and F1 over 0.91 for all twenty datasets over the periods of 2000 to 2019. Feature selection showed that the models retained most of the performance with only one-third of the tested features. Incremental learning was tested within the same topic over time and between different topics, which resulted in slight performance improvements in both cases. This indicates there is an underlying pattern to the neighbors of new topics common to research domains, likely beyond the sample topics used in the experiment. The result showed that network-based new topic prediction can be applied to various research domains with different research patterns.

Highlights

Scientific knowledge evolves through the contribution of researchers around the globe; discoveries are made to expand existing research topics or to contribute towards creation of new topics
The classification results were measured excluding y = 2020 as performance is significantly lower for all domain datasets in the last year with Acc = 0.4068, area under the ROC curve (AUC) = 0.8028, and F1 = 0.5589
This is because the Microsoft Academic Graph (MAG) dataset used in the experiment has only partial records of 2020 publications up to February

Summary

Introduction

Scientific knowledge evolves through the contribution of researchers around the globe; discoveries are made to expand existing research topics or to contribute towards creation of new topics. Gradual expansion or transition of research topics based on the foundation of past knowledge guarantees validity and soundness of the research This is amplified by the fact that researchers within a community can be unaware of research breakthroughs in other related fields [28]. Identifying and predicting emergence of new topics depend on understanding a set of themes shared by related research communities, which are defined as the research topics. They can appear in various forms, including philosophical categories of research, theoretical developments of research models, applications of. Segev / Analyzing the generalizability of the network-based topic emergence identification method technology, and specific algorithms Identifying such topics in academic publications is a crucial part of research activity. A better understanding of such knowledge allows more targeted research aimed at high demand topics, which is needed in both academic and industrial fields

Objectives

Methods

Results

Conclusion