Coverage, redundancy and size-awareness in genre diversity for recommender systems

Abstract

There is increasing awareness in the Recommender Systems field that diversity is a key property that enhances the usefulness of recommendations. Genre information can serve as a means to measure and enhance the diversity of recommendations and is readily available in domains such as movies, music or books. In this work we propose a new Binomial framework for defining genre diversity in recommender systems that takes into account three key properties: genre coverage, genre redundancy and recommendation list size-awareness. We show that methods previously proposed for measuring and enhancing recommendation diversity -including those adapted from search result diversification-fail to address adequately these three properties. We also propose an efficient greedy optimization technique to optimize Binomial diversity. Experiments with the Netflix dataset show the properties of our framework and comparison with state of the art methods.

Personalised Diversification Using Intent-Aware Portfolio

“…As pointed out in [12], certain diversity metrics are correlated and it can be useful to analyse them against each other. In Figure 4 we plot DNG and SDI, against S-recall.…”

“…In [12], Vargas et al focus on genre-based diversity and propose the Binomial framework as a way of measuring and enhancing recommendations in terms of genre coverage, redundancy and list size-awareness. A recommendation list can be seen as a sequence of Bernoulli trials, where a trial models the selection of an item covering each genre.…”

A Survey on Recommendation Methods Beyond Accuracy

“…For instance, a movie recommender focused on (2) recommends movies in a variety of representative genres such as action, comedy, romance in a single RecList while a recommender focused on (1) can recommend movies in various sub-genres in action. (3) User representative diversity, items that cover the dis-similar (diverse) preferences of a given target user should be placed in the RecList [79]. As an example, it is preferable that both sci-fi and romance movies be included in the RecList if the target previously watched movies from the sci-fi and romance genres.…”

“…The thinking behind the algorithms has been introduced to the recommendation field, through several studies [7], [76], [79].…”

“…Another problem is that, in most studies, personalization is covered by the accuracy aspect. With the other aspects-such as diversity and novelty-only a few methods [50], [72], [79] pay attention to personalized diversity or novelty. Correlation between Off-line Performance and User Perception: Given the highly subjective nature of beyondaccuracy aspects, many of the surveyed studies listed, side by side, the results of off-line performance (measured in terms of the metrics described in Sect.…”