ReTraCk: A Flexible and Efficient Framework for Knowledge Base Question Answering

We present Retriever-Transducer-Checker (ReTraCk), a neural semantic parsing framework for large scale knowledge base question answering (KBQA). ReTraCk is designed as a modular framework to maintain high flexibility. It includes a retriever to retrieve relevant KB items efficiently, a transducer to generate logical form with syntax correctness guarantees and a checker to improve the transduction procedure. ReTraCk is ranked at top1 overall performance on the GrailQA leaderboard 1 and obtains highly competitive performance on the typical WebQuestionsSP benchmark. Our system can interact with users timely, demonstrating the efficiency of the proposed framework. 2 * The first three authors contributed equally. This work was conducted during Shuang and Qian's internship at Microsoft Research Asia.