Unifying Task-Oriented Knowledge Graph Learning and Recommendation

Qianyu Li,Hengjie Song,Xiaoli Tang,Tengyun Wang,Haizhi Yang

doi:10.1109/access.2019.2932466

Qianyu Li, Hengjie Song + Show 3 more

Open Access

https://doi.org/10.1109/access.2019.2932466

Copy DOI

Abstract

Incorporating knowledge graphs (KGs) into recommender systems (knowledge-aware recommendation) to improve the recommendation accuracy and explainability has attracted considerable research efforts. However, existing methods largely assume that KGs are complete when transferring knowledge from them, which may lead to suboptimal performance for those KGs, can be hardly complete in real-life scenarios. In this paper, we present a robustly co-learning model (RCoLM) that takes the incompleteness nature of KGs into consideration when incorporating them into recommendation. The RCoLM aims at transferring knowledge between recommendation task and knowledge graph completion (KG completion) task by utilizing a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">transfer learning</i> model. An earlier version of this paper appeared in KDD 2019. This version is an extension of the previous submission and two major innovations are presented here. At first, distinct from previous knowledge-aware recommendation methods, which mainly focus on transferring knowledge from KGs to item recommendations, the RCoLM attempts to exploit user-item interactions from recommendations for KG completion, and unifies the two tasks in a joint model for mutual enhancements. Second, the RCoLM provides a general task-oriented negative sampling strategy on KG completion task, which further improves the adaptive ability of the proposed algorithm and plays an essential role for obtaining superior performance in various sub-tasks of the KG completion. The extensive experiments on two real-world public datasets demonstrate that RCoLM outperforms not only state-of-the-art knowledge-aware recommendation methods but also existing KG completion methods.

Highlights

Knowledge Graph (KG), a heterogeneous structure, has gradually gained attention of researchers in recent years due to the ability to support a bulk of structured data effectively [1]
Since City Light, The Great Dictator and Modern Times are all directed by Charles Chaplin, we may be able to understand Alice’s preference on director Charles Chaplin and predict that Charles Chaplin is the director of The Pilgrim, which transfers user-item interactions from recommendations to KG completion. To deal with these issues, we propose an unified model, which is named as Robustly Co Learning Model (RCoLM), to transfer knowledge between recommendation task and KG completion task for mutual enhancements
To unify item recommendation task and KG completion task, we investigate a strategy for modeling the relationship between the two representations of items/entities, namely robustly co-learning model (RCoLM) and we assume that they have the same latent representations

Summary

Introduction

Knowledge Graph (KG), a heterogeneous structure, has gradually gained attention of researchers in recent years due to the ability to support a bulk of structured data effectively [1]. A KG is a type of multi-relational directed graph composed of a large number of entities and relations [1]. Each edge in the KG is represented as a triple in the form of (head entity, relation, tail entity), called a fact, indicating that the head entity and the tail entity is connected through the relation. A large number of KGs, such as YAGO,. KGs have shown great potential by equipping items with rich auxiliary information (i.e., transferred knowledge), being a promising solution to improve the accuracy and explainability of recommender systems

Methods

Results

Conclusion