A Social Recommendation System for Undergraduate Research


Student thesis: Doctoral Thesis

View graph of relations


Related Research Unit(s)


Awarding Institution
Award date6 Dec 2018


With the rapid development of information technology, we are exposed in the era of information explosion. For undergraduates, basic knowledge of major cannot meet their needs any longer. In order to keep up with the development of times, more and more undergraduates participate in the faculty-mentored research project. Undergraduates’ participation in research projects has shown to be effective in raising students’ interest in research and to be helpful for their future development. The matching between undergraduates and research projects greatly influences student achievement and future academic performance. However, due to the lack of academic experiences, most undergraduates do not have adequate knowledge to understand the topic of research projects and the expertise of research project supervisors. Information asymmetry and incomplete decision-making information make it difficult for undergraduates to find and select suitable research projects, resulting in the mismatch between undergraduate students and research projects. Since most undergraduates do not the experience of doing academic research, their research interests are relatively broad and not very clear. Current keyword-based searching methods are more suitable for undergraduates who clearly know their research interests and understand related research. In addition, the selections of research projects are not only related to the research interest of undergraduates, but also undergraduates have other considerations in the selection process of research projects (that is, the social connections between undergraduates and research project supervisors, the social connections between undergraduates and research project departments/schools, and the quality of research project supervisors).Current keyword-based searching methods only consider the research interests of undergraduates, ignore the social connections between undergraduates and the supervisors of research projects, the social connections between undergraduates and research project departments/schools, and the quality of research project supervisors, and they are difficult to meet the needs of undergraduates.

In order to help undergraduates find and select suitable research projects and promote the match between undergraduates and research projects, this dissertation proposes a social recommendation approach for the selection of research projects based on the research social network platform. The proposed approach is called “social”, and the reasons are that: First, this dissertation captures the information of undergraduates' activities on research social networks, by profiling, modeling and analyzing undergraduates’ research interest, comprehensive research interest profiles of undergraduates can be obtained. Second, this dissertation utilizes individual social network and institutional collaboration network to analyze the social connections between undergraduates and research project at the individual level and the institutional level.

The proposed social recommendation approach includes two stages, namely, the requirement filtering stage and the recommendation stage of the research project. Undergraduates’ selection of research projects is bilateral, and whether the qualifications of undergraduates meet the requirements of research projects affects their choice of research projects. Therefore, this dissertation firstly filters out the research projects for which undergraduates’ qualifications do not meet the requirements. In the recommendation stage, highly relevant, socially connected and high quality research projects are recommended to undergraduates in accordance with the relevance, connectivity and quality modules. In the filtering stage of research project requirements, this dissertation uses rule-based filtering techniques. In the relevance module, this dissertation uses web usage mining techniques and language model methods. In the connectivity module, this dissertation uses Jaccard method in the link prediction area. In the quality module, this dissertation uses bibliometric analysis techniques and altermetric analysis techniques. The proposed approach has been implemented and tested on undergraduate students. The experimental results show that the social recommendation approach is effective in promoting the matching between undergraduates and research projects, as well as motivating them to adopt deep approaches to learning and promoting undergraduates’ engagement in research projects.