An Open Relation Extraction System for Web Text Information

Web texts typically undergo the open-ended growth of new relations. Traditional relation extraction methods lack automatic annotation and perform poorly on tasks. We propose an open-domain system (ORES) based distant supervision few-shot learning to solve this problem. More specifically, we utilize tBERT design instance selector 1, implementing labeling in data mining component. Meanwhile, example 2 K-BERT The real-time management component outputs relational data. Experiments show that ORES can filter out higher quality diverse instances for better learning. It achieves significant improvement compared Neural Snowball with fewer seed sentences.