MINING SCIENTIFIC PUBLICATIONS TO ANALYSE THE COLLABORATION IN RESEARCH COMMUNITIES – THE CASE OF THE INFORMATION SYSTEMS COMMUNITY

Abstract

Scientific communities are social structures composed by people and/or institutions connected through relationships, who have common interests and objectives, sharing information and knowledge. Following this view, social networks have been used to analyze and understand these communities, particularly through their scientific publications. From another perspective, recently born scientific communities are still consolidating their themes of interest and have little understanding about the existence as well as the potential of collaboration networks within the community. In these communities, the information analysis from publications face the challenge of low structure and the absence of consolidated parameters for grouping themes of interest. This work presents an approach to analyze scientific communities through text mining of their publications. The collaborative networks of authorship are automatically extracted and the publications context (classification) automatically found. A computational tool has also been developed to support these analyses. A case study with the Brazilian research community in Information Systems was conducted based on the past editions of the Brazilian Symposium on Information Systems.