Position-aware Attention and Supervised Data Improve Slot Filling

TLDR

Knowledge graphs are crucial for many applications, yet automatic population of knowledge bases from documents has progressed slowly. The study aims to simultaneously tackle two longstanding obstacles in relation extraction. The authors introduce a high‑capacity LSTM model with entity position‑aware attention and construct TACRED, a 119,474‑example crowdsourced dataset for TAC KBP relations. Combining the new dataset with the enhanced model yields substantially better relation extraction, raising the TAC KBP 2015 slot‑filling system’s F1 from 22.2 % to 26.7 %.

Abstract

Organized relational knowledge in the form of “knowledge graphs” is important for many applications. However, the ability to populate knowledge bases with facts automatically extracted from documents has improved frustratingly slowly. This paper simultaneously addresses two issues that have held back prior work. We first propose an effective new model, which combines an LSTM sequence model with a form of entity position-aware attention that is better suited to relation extraction. Then we build TACRED, a large (119,474 examples) supervised relation extraction dataset obtained via crowdsourcing and targeted towards TAC KBP relations. The combination of better supervised data and a more appropriate high-capacity model enables much better relation extraction performance. When the model trained on this new dataset replaces the previous relation extraction component of the best TAC KBP 2015 slot filling system, its F1 score increases markedly from 22.2% to 26.7%.