Heterogeneous Relational Reasoning in Knowledge Graphs with Reinforcement Learning
Preprint
Abstract
Path-based relational reasoning over knowledge graphs has become increasingly popular due to a variety of downstream applications such as question answering in dialogue systems, fact prediction, and recommender systems. In recent years, reinforcement learning (RL) has provided solutions that are more interpretable and explainable than other deep learning models. However, these solutions still face several challenges, including large action space for the RL agent and accurate representation of entity neighborhood structure. We address these problems by introducing a type-enhanced RL agent that uses the local neighborhood information for efficient path-based reasoning over knowledge graphs. Our solution uses graph neural network (GNN) for encoding the neighborhood information and utilizes entity types to prune the action space. Experiments on real-world dataset show that our method outperforms state-of-the-art RL methods and discovers more novel paths during the training procedure.Keywords
All Related Versions
This publication has 9 references indexed in Scilit:
- Graph Few-Shot Learning via Knowledge TransferProceedings of the AAAI Conference on Artificial Intelligence, 2020
- Few-Shot Knowledge Graph CompletionProceedings of the AAAI Conference on Artificial Intelligence, 2020
- Modeling Relation Paths for Knowledge Graph CompletionIEEE Transactions on Knowledge and Data Engineering, 2020
- Reinforcement Knowledge Graph Reasoning for Explainable Recommendation2019
- Multi-Hop Knowledge Graph Reasoning with Reward Shaping2018
- One-Shot Relational Learning for Knowledge Graphs2018
- DeepPath: A Reinforcement Learning Method for Knowledge Graph Reasoning2017
- Composite Task-Completion Dialogue Policy Learning via Hierarchical Deep Reinforcement Learning2017
- Mastering the game of Go with deep neural networks and tree searchNature, 2016