TOWARDS EFFECTIVE RELATIONAL KEYWORD SEARCH USING SEMANTICS

Abstract

Existing relational keyword search methods do not consider the Object-Relationship-Attribute (ORA) semantics in the database, and thus suffer from the problems of returning incomplete answers, an overwhelming number of answers, and even incorrect answers. Besides, they do not consider the normal forms of database relations, and return different answers for different schemas of the same data content. Hence, users may fail to find the answers that satisfy their search intention. In this thesis, we address these problems by exploiting the ORA semantics. We introduce the ORM data/schema graph to capture the ORA semantics in the database. We investigate how objects in the database are related via relationships, and propose a semantic path ranking scheme. Further, we extend the keyword query language to include metadata keywords, and identify the keyword context and search target in keyword queries. Finally, we incorporate keyword queries with aggregates and GROUPBY, and process these queries correctly.