Information integration: the semantic-model approach
Rada Y. Chirkova, Dongfeng Chen · 2008
This dissertation describes a multiple-coordinator system for large-scale information integration and interoperability, presents algorithms for query processing and optimization based on semantic-model approach, and experimentally evaluates the algorithms on real-life and synthetic data. In addition to supporting gradual large-scale information integration and efficient inter-source processing, the semantic-model approach eliminates the need for mediation in deriving the global schema, thus addressing the main limitation of information-integration systems. This dissertation focuses on performance-related characteristics of several alternative approaches proposed for efficient query processing in the semantic-model environment. The theoretical results and practical algorithms are of independent interest and can be used in any information-integration system that avoids loading all the data into a single repository. Query-processing approaches proposed in this dissertation are applicable both to the original stored data and to materialized views, including restructured views, which are a framework for representing, e.g., the pivot operation available in many database-management systems. These approaches account for the practical issues of information overlap across data sources and of inter-source processing. While most of these algorithms are platform- and implementation-independent, XML-specific optimization techniques that allow for system-level tuning of query-processing performance are proposed as well. Finally, using real-life datasets and the implementation of an information-integration system shell, this dissertation provides experimental results that demonstrate that these algorithms are efficient and competitive in the information-integration setting.