MRM

Abstract

In order to deal with the diversified nature of XML documents as well as individual user preferences, we propose a novel Multi-Ranker Model (MRM), which is able to abstract a spectrum of important XML properties and adapt the features to different XML search needs. The model consists of a novel three-level ranking structure and a training module called Ranking Support Vector Machine in a voting Spy Nä1ve Bayes Framework (RSSF). RSSF is effective in learning search preference and then ranks the returned results adaptively. In this demonstration, we present our prototype developed from the model, which we call it the MRM XML search engine. The MRM engine employs only a list of simple XML tagged keywords as a user query for searching XML fragments from a collection of real XML documents. The demonstration presents an indepth analyses of the effectiveness of adaptive rankers, tailored XML rankers and a spectrum of low level ranking features