Efficient and effective keyword search in XML database

Abstract

In this thesis, we propose Tree+IDREF data model to capture ID references for efficient and effective keyword search in XML. In this model, we propose novel Lowest Referred Ancestor (LRA) pair, Extended LRA (ELRA) pair and ELRA group semantics to find search results of keyword queries. Efficient algorithms are presented to compute the search results based on our semantics. Then, we exploit underlining schema information to identify meaningful units of result display. We study rules based on object classes and relationship types captured in ORA-SS to formulate result display for SLCA, ELRA pair and ELRA group results. Besides, we develop a keyword search demo system with DBLP real-world XML database for research community to search for publications and authors based on our search semantics and result presentation rules. The demo prototype is available at: http://xmldb.ddns.comp.nus.edu.sg Finally, experimental evaluation shows the superiority of our approach in search efficiency and result quality.