Improved Neural Relation Detection for Knowledge Base Question Answering

PDF

Relation detection is a core component for many NLP applications including Knowledge Base Question Answering (KBQA). In this paper, we propose a hierarchical recurrent neural network enhanced by residual learning that detects KB relations given an input question. Our method uses deep residual bidirectional LSTMs to compare questions and relation names via different hierarchies of abstraction. Additionally, we propose a simple KBQA system that integrates entity linking and our proposed relation detector to enable one enhance another. Experimental results evidence that our approach achieves not only outstanding relation detection performance, but more importantly, it helps our KBQA system to achieve state-of-the-art accuracy for both single-relation (SimpleQuestions) and multi-relation (WebQSP) QA benchmarks.
Submitted 20 Apr 2017 to Computation and Language [cs.CL]
Published 21 Apr 2017
Subjects: cs.CL cs.AI cs.NE
Author comments: Accepted by ACL 2017
http://arxiv.org/abs/1704.06194
http://arxiv.org/pdf/1704.06194.pdf

0 comments