Paper Session 7: Diversity

Date: Wednesday, Aug 30, 2017, 09:30-10:15
Location: Main Room
Chair: Dietmar Jannach

  • LPFewer Flops at the Top: Accuracy, Diversity, and Regularization in Two-Class Collaborative Filtering by Bibek Paudel, Thilo Haas and Abraham Bernstein

    In most existing recommender systems, implicit or explicit interactions are treated as positive links and all unknown interactions are treated as negative links. The goal is to suggest new links that will be perceived as positive links. However, as signed social networks and newer content services become common, it is important to distinguish between positive and negative preferences. Even in existing applications, the cost of a negative recommendation could be high when people are looking for new jobs, friends, or places to live.

    In this work, we develop novel probabilistic latent factor models to recommend positive links and compare with existing methods on five different openly available datasets. Our models are able to produce better ranking lists and are effective in the task of ranking positive links at the top and negative links at the bottom. Moreover, we find that modeling signed social networks and user preferences this way has the advantage of increasing diversity of recommendations. We also investigate the effect of regularization on the quality of recommendations, a matter that has not received enough attention in the literature. We find that regularization parameter heavily affects the quality of recommendations in terms of both accuracy and diversity.

  • SPGeographical Diversification in POI Recommendation: Toward Improved Coverage on Interested Areas by Jungkyu Han and Hayato Yamana

    In recommending POIs, the accuracy is important. However, merely accurate recommendations are useless in real life. Thus, beyond accuracy – such as the diversity of the recommended POIs – is as important as accuracy for a good recommendation. Although existing diversification methods can help POI recommender systems recommend more diverse POIs, they lack “geographical diversification” that results in concentration of “diverse” recommended POIs on “a small portion” of the most active areas of the target-user. This is caused by the neglect of POI locations in the diversification i.e., existing diversification methods try to diversify the categories of recommended items (POIs). However, geographical diversification is indispensable for the user whose activity district consists of many sub-areas and who requires variety of recommended POIs over his/her activity sub-areas. In this paper, we propose a novel proportional geographical diversification method to recommend variety of POIs located in his/her activity district but variety of sub-areas in the district proportional to the frequency of his/her activity to each sub-area. We compared our proposed method with existing diversification methods with real datasets. The evaluation result shows that no methods except for our proposed method can significantly increase geographical diversity at the expense of tolerable accuracy loss.

Back to Program

Diamond Supporter
Platinum Supporters
Gold Supporter
Silver Supporter
Special Supporters