Search

Automatic Question Answering (QA) system has become quite popular in recent years, especially since the QA tracks appeared at Text REtrieval Conference (TREC). However, using only lexical information, the keyword-based information retrieval cannot fully describe the characteristics of natural language, thus the system performance cannot make people satisfied. It is proposed in this paper a definition of dependency term, based on the dependency grammar, employing the natural language dependency structure, as the improvement of the term, to support the typical information retrieval models. It is in fact a solution for a special application in XML information retrieval (XML IR) field. Experiments show that: dependency-term-based information retrieval model effectively describes the characteristics of natural language questions, and improves the performance of automatic question answering system. [ABSTRACT FROM AUTHOR]

The logical presentation model of XML data is the basis of XML data management. After introducing XML tree models and frequent pattern models, in this paper we have proposed a novel Weighted Element Tree Model (WETM) for measuring the structural similarity of XML documents. This model is a concise form of XML tree models, so the efficiency of the operation on this model is higher than XML tree models. And comparing with frequent pattern models, the WETM enhances the expression ability of structural information of sub trees, which can appreciate the accuracy of similarity evaluation. Moreover, in order to compare the performance of the proposed evaluation algorithm, it is applied to XML documents clustering. The experimental results show that our algorithm is superior to the algorithms based on tree models or frequent pattern models. [ABSTRACT FROM AUTHOR]