Enhancing Freebase Question Answering Using Textual Evidence

Existing knowledge-based question answering systems often rely on small annotated training data. While shallow methods like information extraction techniques are robust to data scarcity, they are less expressive than deep understanding methods, thereby failing at answering questions involving multiple constraints. Here we alleviate this problem by empowering a relation extraction method with additional evidence from Wikipedia. We first present a novel neural network based relation extractor to retrieve the candidate answers from Freebase, and then develop a refinement model to validate answers using Wikipedia. We achieve 53.3 F1 on WEBQUESTIONS, a substantial improvement over the state-of-theart.