Learn More
This article documents a study investigating the specific requirements of a Question Answering System which relies on sentence level document retrieval. We compared 50 TReC factoid queries with the Aquaint document sentences which contain their answers. The main morphological and semantic relations linking query terms to document terms are identified.(More)
We have carried out a comparison of interactive search in a homogenous information retrieval domain using a keyword search engine on the one hand and a concept ontology system on the other. The experimental design was that of the TREC Interactive Track. While the results showed that keyword search was superior on this occasion, we have identified the ideal(More)
The basic architecture of our factoid system is standard in nature and comprises query type identification, query analysis and translation, retrieval query formulation, document retrieval, text file parsing, named entity recognition and answer entity selection. Factoid classification into 69 query types is carried out using keywords. Associated with each(More)