Predicting collaboration between researchers by discovering patterns from the literature

Financiers

The main research problem that we will solve in the framework of the proposed project consists of the following components: 1. Generalization of the approach to detecting legality from the literature to cross-domain recommendation of cooperation among researchers and the development of a theoretical model and methodological framework for forecasting (recommending) new and promising collaborations between researchers. In addition, we want to announce new ones topics that come into play for collaboration and an explanation of why a particular collaboration makes sense. 2. Development of a methodology for investing heterogeneous networks based on a metapot that will allow linking the structural and semantic features of cooperation networks. 3. Development of software tools to recommend participation in huge heterogeneous based networks the concept of semantic metapot. 4. Development of an open source web application for cross-domain research collaboration recommendation that will allow the following functionalities: (i) real-time integration of data from relevant data resources; (ii) cleaning, converting, splitting and aggregating data; (iii) efficiency in terms ofmanaging a huge amount of data and (iv) will offer an interface to visualize the results. 5. Application of the developed methodology on two large bibliographic collections (MEDLINE and COBISS) and Stack Overflow network.

Project Partners

• Univerza v Ljubljani, Fakulteta za strojništvo
• Univerza v Ljubljani, Medicinska fakulteta