Mind-Map Based User Modeling and Research Paper Recommendations

Mind-maps have not received much attention in the user modeling and recommender system community, although they contain lots of information that could be valuable for user modeling and recommender systems. For this paper, we explored the effectiveness of standard user modeling approaches applied to mind-maps, and developed novel user modeling approaches that consider the unique characteristics of mind-maps. We applied and evaluated the approaches with our mind-mapping and reference management software Docear. Docear displayed 431,112 research paper recommendations, based on 4,701 users' mindmaps, from March 2013 to August 2014. The evaluation shows that standard user modeling approaches are reasonably effective when applied to mind-maps, with click-through rates (CTR) between 1.05% and 4.12%. However, when adjusting user modeling to the unique characteristics of mind-maps, a higher CTR of 9.82% could be achieved. The adjustments included, among others, a restriction of the user model size to 35 terms. These terms were extracted from the users’ most recently moved nodes within the past 90 days. Nodes were weighted based on their depths and the number of siblings. The terms of the nodes were weighted with a novel weighting scheme (TF-IDuF) that might also be relevant for recommender systems in general. Overall, our results reinforced our astonishment that mind-maps are being disregarded by the user modeling and recommender system community, and that currently no mind-mapping tools feature any recommender systems. Our research shows that mind-map specific user modeling has a high potential. We hope to initiate a discussion that encourages researchers to do research in this field and that encourages developers to integrate recommender systems to their mind-mapping tools.