Learning Relation Prototype from Unlabeled Texts for Long-Tail Relation Extraction

Relation Extraction (RE) is a vital step to complete Knowledge Graph (KG) by extracting entity relations from texts. However, it usually suffers from the long-tail issue. The training data mainly concentrates on a few types of relations, leading to the lack of sufficient annotations for the remaining types of relations. In this paper, we propose a general approach to learn relation prototypes from unlabeled texts, to facilitate the long-tail relation extraction by transferring knowledge from the relation types with sufficient training data. We learn relation prototypes as an implicit factor between entities, which reflects the meanings of relations as well as their proximities for transfer learning. Specifically, we construct a co-occurrence graph from texts, and capture both first-order and second-order entity proximities for embedding learning. Based on this, we further optimize the distance from entity pairs to corresponding prototypes, which can be easily adapted to almost arbitrary RE frameworks. Thus, the learning of infrequent or even unseen relation types will benefit from semantically proximate relations through pairs of entities and large-scale textual information. We have conducted extensive experiments on two publicly available datasets: New York Times and Google Distant Supervision. Compared with eight state-of-the-art baselines, our proposed model achieves significant improvements (4.1 percent F1 on average). Further results on long-tail relations demonstrate the effectiveness of the learned relation prototypes. We further conduct an ablation study to investigate the impacts of varying components, and apply it to four basic relation extraction models to verify the generalization ability. Finally, we analyze several example cases to give intuitive impressions as qualitative analysis. Our codes and data can be found in https://github.com/CrisJk/PA-TRP.