Neural Graph Reasoning: Complex Logical Query Answering Meets Graph Databases

• DOI: 10.48550/arxiv.2303.14617
• Publication Date: 2023-01-01
• Authors: Hong‐Yu Ren; Mikhail Galkin; Michael Cochez; Zhaocheng Zhu; Jure Leskovec
• Keywords: Graph Embedding

Complex logical query answering (CLQA) is a recently emerged task of graph machine learning that goes beyond simple one-hop link prediction and solves far more complex multi-hop reasoning over massive, potentially incomplete graphs in latent space. The received significant traction the community; numerous works expanded field along theoretical practical axes to tackle different types queries modalities with efficient systems. In this paper, we provide holistic survey CLQA detailed taxonomy studying from multiple angles, including (modality, domain, background semantics), modeling aspects (encoder, processor, decoder), supported (operators, patterns, projected variables), datasets, evaluation metrics, applications. Refining task, introduce concept Neural Graph Databases (NGDBs). Extending idea databases (graph DBs), NGDB consists Storage Engine. Inside Storage, design store, feature further embed information embedding store using an encoder. Given query, Query Engine learns how perform planning execution order efficiently retrieve correct results by interacting Storage. Compared traditional DBs, NGDBs allow for flexible unified features diverse store. Moreover, when incomplete, they can robust retrieval answers which normal DB cannot recover. Finally, point out promising directions, unsolved problems applications future research.