Abstract
User modeling and recommender systems are often seen as key success factors for companies such as Google, Amazon, and Netflix. However, while user-modeling and recommender systems successfully utilize items like emails, news, social tags, and movies, they widely neglect mind-maps as a source for user modeling. We consider this a serious shortcoming since we assume user modeling based on mind maps to be equally effective as user modeling based on other items. Hence, millions of mind-mapping users could benefit from user-modeling applications such as recommender systems. The objective of this doctoral thesis is to develop an effective user-modeling approach based on mind maps. To achieve this objective, we integrate a research-paper recommender system in our mind-mapping and reference-management software Docear. The recommender system builds user models based on the users' mind maps, and recommends research papers based on the user models. As part of our research, we identify several variables relating to mind-map-based user modeling, and evaluate the variables' impact on user-modeling effectiveness with an offline evaluation, a user study, and an online evaluation based on 430,893 recommendations displayed to 4,700 users. We find, among others, that the number of analyzed nodes, the time when nodes were modified, the visibility of nodes, the relations between nodes, and the number of children and siblings of a node affect the effectiveness of user modeling. When all variables are combined in a favorable way, this novel user-modeling approach achieves click-through rates of 7.20%, which is nearly twice as effective as the best baseline. In addition, we show that user modeling based on mind maps performs about as well as user modeling based on other items, namely the research articles users downloaded or cited. Our findings let us to conclude that user modeling based on mind maps is a promising research field, and that developers of mind-mapping applications should integrate recommender systems into their applications. Such systems could create additional value for millions of mindmapping users. As part of our research, we also address the question of how to evaluate recommender systems adequately. This question is highly discussed in the recommender-system community, and we provide some new results and arguments. Among others, we show that offline evaluations often cannot predict results of online evaluations and user studies in the field of research-paper recommender systems. We also show that click-through rate and user rating correlate well (r=0.78). We discuss these findings, including some inherent problems of offline evaluations, and conclude that offline evaluations are probably unsuitable for evaluating research-paper recommender systems, while both user studies and online evaluations are adequate evaluation methods. We also introduce a new weighting scheme, TF-IDuF, which could be relevant for recommender systems in general. In addition, we are first to compare the weighting scheme CC-IDF against CC only, and we research concept drift in the context of researchpaper recommender systems, with the result that interests of researchers seem to shift after about four months. Last, but not least, we publish the architecture of Docear’s recommender system, as well as four datasets relating to the users, recommendations, and document corpus of Docear and its recommender systemAbstract User modeling and recommender systems are often seen as key success factors for companies such as Google, Amazon, and Netflix. However, while user-modeling and recommender systems successfully utilize items like emails, news, social tags, and movies, they widely neglect mind-maps as a source for user modeling. We consider this a serious shortcoming since we assume user modeling based on mind maps to be equally effective as user modeling based on other items. Hence, millions of mind-mapping users could benefit from user-modeling applications such as recommender systems. The objective of this doctoral thesis is to develop an effective user-modeling approach based on mind maps. To achieve this objective, we integrate a research-paper recommender system in our mind-mapping and reference-management software Docear. The recommender system builds user models based on the users' mind maps, and recommends research papers based on the user models. As part of our research, we identify several variables relating to mind-map-based user modeling, and evaluate the variables' impact on user-modeling effectiveness with an offline evaluation, a user study, and an online evaluation based on 430,893 recommendations displayed to 4,700 users. We find, among others, that the number of analyzed nodes, the time when nodes were modified, the visibility of nodes, the relations between nodes, and the number of children and siblings of a node affect the effectiveness of user modeling. When all variables are combined in a favorable way, this novel user-modeling approach achieves click-through rates of 7.20%, which is nearly twice as effective as the best baseline. In addition, we show that user modeling based on mind maps performs about as well as user modeling based on other items, namely the research articles users downloaded or cited. Our findings let us to conclude that user modeling based on mind maps is a promising research field, and that developers of mind-mapping applications should integrate recommender systems into their applications. Such systems could create additional value for millions of mindmapping users. As part of our research, we also address the question of how to evaluate recommender systems adequately. This question is highly discussed in the recommender-system community, and we provide some new results and arguments. Among others, we show that offline evaluations often cannot predict results of online evaluations and user studies in the field of research-paper recommender systems. We also show that click-through rate and user rating correlate well (r=0.78). We discuss these findings, including some inherent problems of offline evaluations, and conclude that offline evaluations are probably unsuitable for evaluating research-paper recommender systems, while both user studies and online evaluations are adequate evaluation methods. We also introduce a new weighting scheme, TF-IDuF, which could be relevant for recommender systems in general. In addition, we are first to compare the weighting scheme CC-IDF against CC only, and we research concept drift in the context of researchpaper recommender systems, with the result that interests of researchers seem to shift after about four months. Last, but not least, we publish the architecture of Docear’s recommender system, as well as four datasets relating to the users, recommendations, and document corpus of Docear and its recommender system
To download full research paper click on the link below.
If you need implementation of this research paper or any of its variants, feel free contact us on contact@codersarts.com.
Comments