Faithfully Explainable Recommendation via Neural Logic Reasoning

Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 3083–3090 June 6–11, 2021. ©2021 Association for Computational Linguistics 3083 Faithfully Explainable Recommendation via Neural Logic Reasoning Yaxin Zhu∗, Yikun Xian∗, Zuohui Fu∗, Gerard de Melo, Yongfeng Zhang Rutgers University, New Brunswick {yaxin.a.zhu,yikun.xian,zuohui.fu}@rutgers.edu {gerard.demelo,yongfeng.zhang}@rutgers.edu Abstract Knowledge graphs (KG) have become in- creasingly important to endow modern rec- ommender systems with the ability to gener- ate traceable reasoning paths to explain the recommendation process. However, prior re- search rarely considers the faithfulness of the derived explanations to justify the decision- making process. To the best of our knowledge, this is the first work that models and evaluates faithfully explainable recommendation under the framework of KG reasoning. Specifically, we propose neural logic reasoning for explain- able recommendation (LOGER) by drawing on interpretable logical rules to guide the path- reasoning process for explanation generation. We experiment on three large-scale datasets in the e-commerce domain, demonstrating the ef- fectiveness of our method in delivering high- quality recommendations as well as ascertain- ing the faithfulness of the derived explanation. 1 Introduction Compared with traditional recommender systems (RS), explainable recommendation is not only ca- pable of providing high-quality recommendation results but also offers personalized and intuitive ex- planations (Zhang and Chen, 2020). Incorporating a knowledge graph (KG) into recommender sys- tems has become increasingly popular, since KG reasoning is able to generate explainable paths con- necting users to relevant target item entities. At the same time, there is increasing demand for systems to ascertain the faithfulness of the generated expla- nation, i.e., assess whether it faithfully reflects the reasoning process of the model and is consistent with the historic user behavior. However, previous work has largely neglected faithfulness in KG-enhanced explainable recom- mendation (Xian et al., 2020a; Fu et al., 2020a). A number of studies (Lakkaraju et al., 2019; ter Ho- eve et al., 2018; Wu and Mooney, 2018) argue that ∗Equal contribution faithful explanations should also be personalized and gain the capability to reflect the personalized user historic behavior. However, to the best of our knowledge, none of the existing explainable recommendation models based on KGs have con- sidered faithfulness in the explainable reasoning process and its evaluation on the generated explain- able paths. For instance, PGPR (Xian et al., 2019; Zhao et al., 2020) infers explainable paths over the KG without considering personalized user be- havior, and its prediction on next potential entities is merely based on the overall knowledge-driven rewards. CAFE (Xian et al., 2020b) builds user module profiles to guide the path inference pro- cedure. However, as illustrated in Subramanian et al. (2020), such neural module networks only implicitly abstract the reasoning process and lack of considering the faithfulness of explanations. In this paper, we propose a new KG-enhanced recommendation model called LOGER to produce faithfully explainable recommendation via neural logic reasoning. To fully account for heterogeneous information and rules about users and items from the KG, we leverage an interpretable neural …

Content truncated to 500 words