Improving Recommendation Diversity by Highlighting the ExTrA Fabricated Experts

Ya-Hui An,Chao Wang,Quan Yuan,Qiang Dong

doi:10.1109/access.2020.2984365

Abstract

Nowadays, recommender systems (RSes) are becoming increasingly important to individual users and business marketing, especially in the online e-commerce scenarios. However, while the majority of recommendation algorithms proposed in the literature have focused their efforts on improving prediction accuracy, other important aspects of recommendation quality, such as diversity of recommendations, have been more or less overlooked. In the latest decade, recommendation diversity has drawn more research attention, especially in the models based on user-item bipartite networks. In this paper, we introduce a family of approaches to extract fabricated experts from users in RSes, named as the Expert Tracking Approaches (ExTrA for short), and explore the capability of these fabricated experts in improving the recommendation diversity, by highlighting them in a well-known bipartite network-based method, called the Mass Diffusion (MD for short) model. These ExTrA-based models are compared with two state-of-the-art MD-improved models HHP and BHC, with respect to recommendation accuracy and diversity. Comprehensive empirical results on three real-world datasets MovieLens, Netflix and RYM show that, our proposed ExTrA-based models can achieve significant diversity gain while maintain comparable level of recommendation accuracy.

Highlights

Recommender systems (RSes) are powerful tools of helping users confront the challenge of information overload, by uncovering users’ potential preferences on uncollected items and delivering personalized recommendation lists
Liu et al [20] proposed a biased heat conduction (BHC) model to enhance the accuracy of Heat Conduction (HC) model; Zhou et al [11] integrated Mass Diffusion (MD) and HC methods together to generate a hybrid recommendation model, which improves accuracy and diversity simultaneously
WORKS In this paper, we introduce a family of approaches to extract fabricated experts from all users in recommender systems, and highlight them in the mass diffusion model

Summary

INTRODUCTION

Recommender systems (RSes) are powerful tools of helping users confront the challenge of information overload, by uncovering users’ potential preferences on uncollected items and delivering personalized recommendation lists. MD can achieve more accurate recommendations than the traditional item-based Collaborative Filtering (CF) model [16], it can be categorized into a special case of CF with the RA similarity rather than the common Cosine or Jaccard similarity [22] Another model, called the Heat Conduction (HC) [23], is a similar process, but allocating resource in a different way, which results in the exposure of more niches, with rather low accuracy, could not be applied alone in real RSes. Subsequently, Liu et al [20] proposed a biased heat conduction (BHC) model to enhance the accuracy of HC model; Zhou et al [11] integrated MD and HC methods together to generate a hybrid recommendation model, which improves accuracy and diversity simultaneously. The total resource that i would get is the sum of resource from all the collected items

THE ExTrA-BASED DIFFUSION MODEL

DIVERSITY COMPARISON WITH EXISTING MODELS

Findings

CONCLUSIONS AND FUTURE WORKS