Using Syntactic and Semantic Relation Analysis in Question Answering

Our participation at TREC thi integrating dependency and analysis of external resources int system. In TREC-13, we have p dependency relation matching t extraction for factoid and list que showed that the technique is ef extraction within the corpus. H also identified some problems an this technique. First, dependency does not perform well on short have only very few key terms. Th integrate query expansion and ma resources to provide addit information to these short ques technique cannot be directly a answer nuggets from external w pages contain much more noise a corpus, statistical based dep matching tends to make a lot of e previously trained model on the we do not have sufficient trainin model on the web. Therefore w semantic relation analysis dependency relation analysis t nuggets for factoid and list ques Finally, we adopt a soft pattern (Cui et al., 2005) for definition se the definitional QA task. We focus on the following thr year’s TREC: (1) We incorporate dependency into query expansion based resources, and we perform qu both document and passage r more contextual information (2) We propose the use of analysis to extract answer n and list questions from exter These external answer nugg directly projected back to th questions) or be used as add evidence for corpus-based (for factoid questions). {sunrenxu,jing Department of Computer Science School of Computing National University of Singapore jian,tanyeefa,cuihang,chuats,kanmy}@comp.nus.edu.sg