Canonicalizing Open Knowledge Bases with Multi-Layered Meta-Graph Neural Network
This work addresses data quality issues in knowledge bases for applications relying on structured information, but it appears incremental as it builds on existing canonicalization methods with a novel graph-based approach.
The paper tackles the problem of canonicalizing noun and relational phrases in Open Knowledge Bases to reduce redundancy and ambiguity by integrating structural and semantic information into a multi-layered graph, and it reports that their model outperforms existing approaches on a public dataset.
Noun phrases and relational phrases in Open Knowledge Bases are often not canonical, leading to redundant and ambiguous facts. In this work, we integrate structural information (from which tuple, which sentence) and semantic information (semantic similarity) to do the canonicalization. We represent the two types of information as a multi-layered graph: the structural information forms the links across the sentence, relational phrase, and noun phrase layers; the semantic information forms weighted intra-layer links for each layer. We propose a graph neural network model to aggregate the representations of noun phrases and relational phrases through the multi-layered meta-graph structure. Experiments show that our model outperforms existing approaches on a public datasets in general domain.