Abstract:Knowledge graph question answering (KGQA) aims to answer natural language questions from structured knowledge graphs (KGs). Traditional KQGA methods are usually limited to single-hop queries and cannot handle complex questions involving multi-hop reasoning well. To overcome this issue, multi-hop KGQA based on reinforcement learning (RL) has been proposed. However, multi-hop KGQA based on RL still faces some challenges. Firstly, due to the insufficient availability of latent environmental information during the reasoning process, the agent finds it challenging to make coherent and correct decisions. Secondly, the agent only receives rewards from the environment upon reaching the answer entity during the exploration, leading to slow or even obstructed learning. To address these shortcomings, we construct multi-perspective information based on the state of the environment, and integrate multi-perspective information with RL framework, thereby creating the Multi-Perspective Information Fusion Reasoning Network (MPIFRN). MPIFRN achieves the goal via three steps. (1) We construct three different views of information, i.e., expectation embedding, instruction-guided embedding, and path-aware embedding. These environmental cues provide more reliable support for decision-making. (2) We still adopt the method of mapping entities and relations into the knowledge graph embedding space to answer multi-hop questions. At each step of reasoning, we use a scoring function to measure the plausibility of each "triple" ¡topic entity, question, candidate entity¿ in the embedding space. (3) Furthermore, we employ the asynchronous advantage actor-critic (A3C) algorithm to guide the agent in selecting the most promising entities and to expand the reasoning paths in parallel by updating policy and value network parameters, thereby facilitating multi-hop knowledge graph question answering. We conduct extensive experiments on KGQA benchmark datasets, providing substantial evidence to demonstrate the effectiveness of our approach.

MACRE: Multi-hop Question Answering via Contrastive Relation Embedding

Multi-hop question answering over incomplete knowledge graph with abstract conceptual evidence

Unrestricted multi-hop reasoning network for interpretable question answering over knowledge graph

Improving embedded knowledge graph multi-hop question answering by introducing relational chain reasoning

Exploiting Hybrid Semantics of Relation Paths for Multi-hop Question Answering Over Knowledge Graphs

UniKGQA: Unified Retrieval and Reasoning for Solving Multi-hop Question Answering Over Knowledge Graph

QAGCN: Answering Multi-Relation Questions via Single-Step Implicit Reasoning over Knowledge Graphs

Incorporating multi-perspective information into reinforcement learning to address multi-hop knowledge graph question answering

Multi-hop question answering of bridge inspection by adopting knowledge graph embedding technology

Hic-KGQA: Improving multi-hop question answering over knowledge graph via hypergraph and inference chain

Dynamic Semantic Graph Construction and Reasoning for Explainable Multi-hop Science Question Answering

Knowledge Graph Multi-Hop Question Answering Based on Dependent Syntactic Semantic Augmented Graph Networks

Path-Enhanced Multi-Relational Question Answering with Knowledge Graph Embeddings

Multi-hop Question Answering for SRLGRN Augmented by Textual Relationship Modelling

TransferNet: an Effective and Transparent Framework for Multi-hop Question Answering over Relation Graph

RConE: Rough Cone Embedding for Multi-Hop Logical Query Answering on Multi-Modal Knowledge Graphs

Translational Relation Embeddings for Multi-Hop Knowledge Base Question Answering

DSAMR: Dual-Stream Attention Multi-hop Reasoning for knowledge-based visual question answering

Relation-Aware Question Answering for Heterogeneous Knowledge Graphs

BeamAggR: Beam Aggregation Reasoning over Multi-source Knowledge for Multi-hop Question Answering

Multi-hop Question Answering over Knowledge Graphs using Large Language Models