2017/07/14 by Alon Talmor, Mor Geva, Talmor, Alon +3
Computer Science · Decision Sciences · #Computation and Language (cs.CL) #Data Quality and Management #FOS: Computer and information sciences #Natural Language Processing Techniques #Topic Modeling
paper · pdf · doi:10.48550/arxiv.1707.04412
openalex publication_date 2017/07/14 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Semantic parsing shines at analyzing complex natural language that involves\ncomposition and computation over multiple pieces of evidence. However, datasets\nfor semantic parsing contain many factoid questions that can be answered from a\nsingle web document. In this paper, we propose to evaluate semantic\nparsing-based question answering models by comparing them to a question\nanswering baseline that queries the web and extracts the answer only from web\nsnippets, without access to the target knowledge-base. We investigate this\napproach on COMPLEXQUESTIONS, a dataset designed to focus on compositional\nlanguage, and find that our model obtains reasonable performance (35 F1\ncompared to 41 F1 of state-of-the-art). We find in our analysis that our model\nperforms well on complex questions involving conjunctions, but struggles on\nquestions that involve relation composition and superlatives.\n