Exploring Research Networks with Data Science: A Data-Driven Microservice Architecture for Synergy Detection

Abstract

The determination of synergies in research networks is often approached by citation and co-authorship analysis using scientific publications. Although the latter methods allow a detailed insight into scientific cooperation and communities, the major part of the data, the text body of the publication, is mostly ignored. Hence, this text body contains valuable data, which can be processed in order to describe entities in a research network, detect synergies between those entities and make use of these synergies, e.g. in further cooperation. This paper aims at the description of a prototypic architecture, which uses data science to detect hidden topics from publication data. These topics are used to describe entities (e.g. institutes or projects) in a research network. Topical proximity between those entities is computed by the application of classification. In order to enable actors to explore the results of these processes, the architecture automatically generates graph-based visualizations of the derived synergies. This paper contains a description and definition of the technical elements and the process chain of the architecture to derive these synergies in network data. This is accompanied by a short outline of results with the prototype.