Ozgur Yilmazel

Evaluation of Turkish Text Information Retrieval: Relational Databases Versus Apache Lucene Ozgur Yilmazel

Many of today's applications have a need for full-text search capabilities for various reasons. Although full-text search has traditionally been the domain of Information Retrieval, nowadays popular Relational Database Management Systems started to implement functionalities that support full-text indexing and searching. The present book covers a comparison of the text retrieval performances of relational databases and Information Retrieval Systems, as well as a comparison of the execution times during indexing and retrieval tasks over a Text REtrieval Conference (TREC)-like test collection for Turkish that contains 408,305 documents and 72 ad hoc queries. The effects of language specific processing for different systems are investigated. Also the effects of different query lengths and operators on retrieval performance are investigated. It is found that language specific preprocessing improves retrieval performance for all systems. Relational Databases are generally slower with longer queries.