Syntactic impact on sentence similarity measure in archive-based QA system

  • Authors:

  • Guang Qiu;Jiajun Bu;Chun Chen;Peng Huang;Keke Cai

  • Affiliations:

  • College of Computer Science, Zhejiang University, Hangzhou, China;College of Computer Science, Zhejiang University, Hangzhou, China;College of Computer Science, Zhejiang University, Hangzhou, China;College of Computer Science, Zhejiang University, Hangzhou, China;College of Computer Science, Zhejiang University, Hangzhou, China

  • Venue:
  • PAKDD'07 Proceedings of the 11th Pacific-Asia conference on Advances in knowledge discovery and data mining
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

There's now an increase in the number of Question Answering communities where large archives of question and answer pairs are collected up over time. These archives help traditional type-specified Question Answering (QA) systems to overcome type constraints and enable a service of general types. Semantic similarity measures between sentences dominate the overall performance of such Archive-based QA systems in finding similar questions in the archive to users' requests. Available approaches to sentence similarity measurement mainly utility word-to-word similarity measures directly in a bag-of-words way. In this paper, we take the syntactic evidence into account and carry out an examination on the impact of syntactic information on the sentence similarity measurement. We also compare the performance of our syntactic information incorporated approach with some baseline retrieval models. Experiments show that our approach outperforms other models both in mean average precision (MAP) and recall.