Adaptive diversification of recommendation results via latent factor portfolio

Abstract

This paper studies result diversification in collaborative filtering. We argue that the diversification level in a recommendation list should be adapted to the target users' individual situations and needs. Different users may have different ranges of interests -the preference of a highly focused user might include only few topics, whereas that of the user with broad interests may encompass a wide range of topics. Thus, the recommended items should be diversified according to the interest range of the target user. Such an adaptation is also required due to the fact that the uncertainty of the estimated user preference model may vary significantly between users. To reduce the risk of the recommendation, we should take the difference of the uncertainty into account as well.In this paper, we study the adaptive diversification problem theoretically. We start with commonly used latent factor models and reformulate them using the mean-variance analysis from the portfolio theory in text retrieval. The resulting Latent Factor Portfolio (LFP) model captures the user's interest range and the uncertainty of the user preference by employing the variance of the learned user latent factors. It is shown that the correlations between items (and thus the item diversity) can be obtained by using the correlations between latent factors (topical diversity), which in return significantly reduce the computation load. Our mathematical derivation also reveals that diversification is necessary, not only for risk-averse system behavior (nonadpative), but also for the target users' individual situations (adaptive), which are represented by the distribution and the variance of the latent user factors. Our experiments confirm the theoretical insights and show that LFP succeeds in improving latent factor models by adaptively introducing recommendation diversity to fit the individual user's needs.

A Survey on Recommendation Methods Beyond Accuracy

“…Ishikawa et al [33] exploited information diffusion theory to recommend novel technical articles to interested users; additionally, Shi et al [66] used portfolio theory from the field of economics to tune recommendations (50), [16](288), [20] (382) between risk-taking and risk-aversion.…”

“…Portfolio theory provides a framework by which to tune the degree of risk inherent in the retrieved information. The portfolio concept introduced by Wang et al was extended to the study of recommendations by Shi et al [66].…”

Personalised Diversification Using Intent-Aware Portfolio

“…A variation using a covariance matrix created on explicit item-aspect assignments was proposed and evaluated in [9]. Another adaptation of the Mean-Variance framework has been proposed in [10] for latent factor models by employing variance of learned user latent factors to address users' individual situations.…”

“…they are independent of the user's preferences. A user-dependent covariance has been proposed in [10], however this is in the context of latent factorisation models.…”

Presenting Diversity Aware Recommendations: Making Challenging News Acceptable

“…Previous studies have found that users vary in terms of the degree of diversification that is optimal for them, and that this can be deduced from their previous rating behavior (see e.g., [13,21]). Users also vary in terms of personality traits that are fixed, such as the Big Five [12].…”