Overview of the approachAbstract
Open relation extraction (ORE) aims to assign semantic relationships between arguments, essential to the automatic construction of knowledge graphs. The previous methods either depend on external NLP tools (eg, PoS-taggers) and language-specific relation formations, or suffer from inherent problems in sequence representations, thus leading to unsatisfactory extraction in diverse languages and domains. To address the above problems, we propose a Query-based Open Relation Extractor (QORE). QORE utilizes a Transformers-based language model to derive a representation of the interaction between arguments and context, and can process multilingual texts effectively. Extensive experiments are conducted on seven datasets covering four languages, showing that QORE models significantly outperform conventional rule-based systems and the state-of-the-art method LOREM [8]. Regarding the practical challenges [1] of Corpus Heterogeneity and Automation, our evaluations illustrate that QORE models show excellent zero-shot domain transferability and few-shot learning ability.