Topic-Driven Detection and Analysis of Scholarly Data

Abstract

AbstractThe chapter presents a topic mining approach that can used for a scholarly data analysis. The idea here is that research topics can emerge through an analysis of epistemological aspects of scholar publications that are extracted from conventional publication metadata, such as the title, the author-assigned keywords, and the abstract. As a first contribution, we provide a conceptual analysis of research topic profiling according to the peculiar behaviours/trends of a given topic along a considered time interval. As a further contribution, we define a disciplined approach and the related techniques for topic mining based on the use of publication metadata and natural language processing (NLP) tools. The approach can be employed within a variety of topic analysis issues, such as country-oriented and/or field-oriented research analysis tasks that are based on scholarly publications. In this direction, to assess the applicability of the proposed techniques for use in a real scenario, a case study analysis based on two publication datasets (one national and one worldwide) is presented.