Textual Question Answering for Semantic Parsing in Natural Language Processing

Question answering (QA) problem is one of the most frequent tasks in natural language processing (NLP) over language input. To understand the natural language, semantic parsing is an extensively used process. For deriving syntactic and semantic structures of texts, synchronous frameworks are being used frequently. With the aim of answering questions efficiently, in this paper, an approach by semantic parsing of texts with a view to leveraging semantic information is introduced. The proposed approach uses the method based on lambda calculus for semantic parsing to derive logical forms of sentences. The questions are analyzed by collecting significant features to find out correct answers from the facts. Unlike traditional approaches, this paper extends lambda calculus based dependency parser methodology for question analysis. This proposed approach for answering questions is based on generalized searching techniques. Proposed system achieves 95% accuracy for Yes/No questions with an overall 83% mean accuracy for the five tasks of bAbI-10k question answering dataset surpassing the existing approach by about 11%.