User Segmentation for Controlling Recommendation Diversity

Abstract

The quality of recommendations is known to be a ected by diversity and novelty in addition to accuracy. Recent work has focused on methods that increase diversity of recommendation lists. However, these methods assume the user preference for diversity is constant across all users. In this paper, we show that users' propensity towards diversity varies greatly and argue that the diversity of recommendation lists should be consistent with the level of user interest in diverse recommendations. We introduce a user segmentation approach in order to personalize recommendation according to user preference for diversity. We show that recommendations generated using these segments match the diversity preferences of users in each segment. We also discuss the impact of this segmentation on the novelty of recommendations.

Keywords

• Recommendation diversity
• Performance evaluation metrics
• Novelty
• Collaborative Filtering

1. INTRODUCTION

Although there are many methods in the literature that can be used to increase diversity in recommendations [ 1 ], only a few have mentioned the varying degrees of interest users have for diverse recommendation results [ 2 ]. One can imagine two extreme cases of this interest: one user likes to receive as recommendations only science ction movies made within the last 10 years; another user likes a more diverse set of movies from many genres in her recommendation list. Obviously, any attempt to increase the diversity of recommendation list is likely to generate poor results for the rst user with limited interests.

We measure a user's preference for diversity as a function of the diversity of items that the user has rated, and segment the users into groups based on their scores. Recommendations for each group can be generated independently Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for third-party components of this work must be honored. For all other uses, contact the owner/author(s). 2.

DEFINITIONS

Let U and I be the sets of users and items, respectively. The lists of recommendations is denoted as R. Ru is the recommendation items for user u 2 U and user pro le Iu is the list of items that u has rated. Diversity is the measure of dissimilarity between items in a set. For this purpose, we use average pairwise distance of items in a set as Intra-List Distance (ILD ) [ 4 ].

ILD(L) =

1 jLj(jLj

X Xd(i; j) 1) i2L j2L (1)

In addition to diversity, we can measure the impact of user segmentation on the novelty or catalog coverage of recommendation lists. We de ne novelty as the average distance from the items in user pro le to the items in recommendations.

N ov(Iu; Ru) = jRujjIuj 1

X X d(i; j) min(jRuj; jIuj) i2Ru j2Iu (2)

We also consider the popularity of items in the recommendation lists. Popularity of an item i is de ned by P op(i) =

jUij maxj2I (jUj j) where Ui is the set of users who have rated item i.

EXPERIMENTS AND DISCUSSIONS

CONCLUSIONS

This work examines the consequences of segmenting user populations by diversity, as a means of personalizing user interest in and tolerance for diversity. We show that interest in diversity varies widely across users, with a distinct peak and users with preferences both low and high.

Our division of the user population into four segments is a simple but e ective method for increasing diversity for those segments of the population interested in such diversity and decreasing it for those with less interest. The expected e ects on diversity and novelty are seen across three di erent recommendation algorithms.

We plan to explore these e ects in future work in additional datasets and algorithms, as well as alternate methods for personalizing diversity.

References

[1] G. Adomavicius and Y. Kwon . Improving aggregate recommendation diversity using ranking-based techniques . IEEE Transactions on Knowledge and Data Engineering , 24 ( 5 ): 896 { 911 , 2012 .

[2] K. Kapoor , V. Kumar , L. Terveen , J. A. Konstan , and P. Schrater . I like to explore sometimes: Adapting to dynamic user novelty preferences . In Proceedings of the 9th ACM Conference on Recommender Systems , pages 19 { 26 . ACM, 2015 .

[3] S. Rendle , C. Freudenthaler , Z. Gantner , and L. Schmidt-Thieme . Bpr: Bayesian personalized ranking from implicit feedback . In Proceedings of the twenty- fth conference on uncertainty in arti cial intelligence , pages 452 { 461 . AUAI Press, 2009 .

[4] C.-N. Ziegler , S. M. McNee , J. A. Konstan , and G. Lausen . Improving recommendation lists through topic diversi cation . In Proceedings of the 14th international conference on World Wide Web , pages 22 { 32 . ACM, 2005 .