Syntactic Multi-view Learning for Open Information Extraction
This work addresses the extraction of relational tuples from open-domain sentences, offering an incremental improvement by better integrating syntactic structures into neural models.
The paper tackled the problem of under-exploiting syntactic information in neural Open Information Extraction by modeling constituency and dependency trees as word-level graphs and using multi-view learning to fuse them with semantic representations, resulting in improved tuple generation as shown in experiments.
Open Information Extraction (OpenIE) aims to extract relational tuples from open-domain sentences. Traditional rule-based or statistical models have been developed based on syntactic structures of sentences, identified by syntactic parsers. However, previous neural OpenIE models under-explore the useful syntactic information. In this paper, we model both constituency and dependency trees into word-level graphs, and enable neural OpenIE to learn from the syntactic structures. To better fuse heterogeneous information from both graphs, we adopt multi-view learning to capture multiple relationships from them. Finally, the finetuned constituency and dependency representations are aggregated with sentential semantic representations for tuple generation. Experiments show that both constituency and dependency information, and the multi-view learning are effective.