Heterogeneous relational reasoning in knowledge graphs with reinforcement learning

Saebi, Mandana; Krieg, Steven J.; Zhang, Chuxu; Jiang, Meng; Kajdanowicz, Tomasz; Chawla, Nitesh V.

doi:10.1016/j.inffus.2022.07.001

Cited by 9 publications

(7 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, Wang et al introduced attention to utilize the context information by using GNN (Graph Neural Networks) [39]. HRRL [40] performs path-based inference using GNN to encode neighborhood information. Some scholars have tried to apply graph auto-encoders (GAE) and spatiotemporal graph neural networks (STGNN) to knowledge reasoning tasks.…”

Section: B Representation-learning-based Reasoningmentioning

confidence: 99%

“…The electronics data set from the open data set of Amazon is used for experimental verification [64,65]. They provide both rich meta-information of products and diverse user review records and have been widely used as domain-specific e-commerce datasets for research [5,40,45]. It contains data related to products and reviews.…”

Section: Description and Preprocessing Of The Datasetmentioning

confidence: 99%

See 1 more Smart Citation

Inferring Complementary and Substitutable Products Based on Knowledge Graph Reasoning

Fang,

Yu,

Ding

et al. 2023

Mathematics

View full text Add to dashboard Cite

Complementarity and substitutability between products are essential concepts in retail and marketing. To achieve this, existing approaches take advantage of knowledge graphs to learn more evidence for inference. However, they often omit the knowledge that lies in the unstructured data. In this research, we concentrate on inferring complementary and substitutable products in e-commerce from mass structured and unstructured data. An improved knowledge-graph-based reasoning model has been proposed which cannot only derive related products but also provide interpretable paths to explain the relationship. The methodology employed in our study unfolds through several stages. First, a knowledge graph refining entities and relationships from data was constructed. Second, we developed a two-stage knowledge representation learning method to better represent the structured and unstructured knowledge based on TransE and SBERT. Then, the relationship inferring problem was converted into a path reasoning problem under the Markov decision process environment by learning a dynamic policy network. We also applied a soft pruning strategy and a modified reward function to improve the effectiveness of the policy network training. We demonstrate the effectiveness of the proposed method on standard Amazon datasets, and it gives about 5–15% relative improvement over the state-of-the-art models in terms of NDCG@10, Recall@10, Precision @10, and HR@10.

show abstract

Section: B Representation-learning-based Reasoningmentioning

confidence: 99%

Section: Description and Preprocessing Of The Datasetmentioning

confidence: 99%

Inferring Complementary and Substitutable Products Based on Knowledge Graph Reasoning

Fang,

Yu,

Ding

et al. 2023

Mathematics

View full text Add to dashboard Cite

show abstract

“…This path is selected by the model as the most representative for the recommended product p among the set of all the predicted 𝑘-hop paths L 𝑘 𝑢,𝑝 between user 𝑢 and a not yet interacted product p. Our addressed task, named as Knowledge Graph Reasoning for Explainable Recommendation (KGRE-Rec) consists of two sub-tasks; i) making recommendations for each user and ii) generating a path sequence as the explanation of each recommended product. Path-guided approaches solve the KGRE-Rec problem as a unified task [7,30,36,41,50,51]. Autoregressive Path Generation for Explainable Recommendation.…”

Section: Preliminariesmentioning

confidence: 99%

“…Modern techniques exploit KG paths for clearer explanations, employing models rooted in KGEs. These aim to navigate KGs from users to unseen products, utilizing paths as explanations, achieved through methods like reinforcement learning [7,30,36,41,50] or neural-symbolic systems [38,51]. However, using KGEs poses issues of explanation fidelity and scalability.…”

Section: Related Workmentioning

confidence: 99%

“…Furthermore, annotating relevant path types manually might be time-consuming and impractical in real-world scenarios. To address these issues, modern methods perform the recommendation and path selection steps jointly, through path reasoning [7,30,36,41,50,51]. They rely on reinforcement learning (RL), with agents optimised to identify paths from users to unseen yet relevant products in the KG.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Hands on Explainable Recommender Systems with Knowledge Graphs

Balloccu

Boratto

Fenu

et al. 2022

Proceedings of the 16th ACM Conference on Recommender Systems

View full text Add to dashboard Cite

Path reasoning methods over knowledge graphs have gained popularity for their potential to improve transparency in recommender systems. However, the resulting models still rely on pre-trained knowledge graph embeddings, fail to fully exploit the interdependence between entities and relations in the KG for recommendation, and may generate inaccurate explanations. In this paper, we introduce PEARLM, a novel approach that efficiently captures user behaviour and product-side knowledge through language modelling. With our approach, knowledge graph embeddings are directly learned from paths over the KG by the language model, which also unifies entities and relations in the same optimisation space. Constraints on the sequence decoding additionally guarantee path faithfulness with respect to the KG. Experiments on two datasets show the effectiveness of our approach compared to state-of-the-art baselines. Source code and datasets: AVAILABLE AFTER GETTING ACCEPTED. CCS CONCEPTS• Information systems → Recommender systems; • Applied computing → Law, social and behavioral sciences.

show abstract