A Uniform Query Processing Approach for Integrating Data from Heterogeneous Resources

• Scientists who need to explore several different databases in theirresearch can find it difficult and tedious to extract and combineinformation from various heterogeneous data sources manually.This is a particular problem for researchers in the life sciences,since technical advances in the last decade have resulted in a dramaticincrease in the quantity and variety of data. Many databases of interest are developed independently by differentresearch groups, and the database administrators often want tokeep their databases autonomous so that they can develop and maintainthem without being constrained by other database sources.Therefore, there is a need for software solutions to the problem ofdata integration that facilitate combiningup-to-date data from autonomous, heterogeneous databases located atdifferent sites. A system for data integration from heterogeneous (relational and RDF/S),autonomous and distributed data sources has been designed and implemented inthis work. The main aim in the design and implementation of the system hasbeen to make large parts of query and result processing independent ofthe kinds of data resources that are being used. The queries are heldin a resource independent form through large parts of the query processing.We refer to this as uniform query and result processing. The userstates queries, global queries, against an integrated view of theunderlying data resources. The integrated view does not reveal thestructure of the underlying data sources. A global query is rewritten byusing rules that describe the mapping from concepts in the integrated view toconcepts in the data sources. This is then split into sub-queries thateach relate to one of the data sources. Wrappers translate sub-queriesinto the query languages of the component databases, send these sub-queriesto the component databases and then retrieve the results. Several smallexample federations have been implemented to test the system, one ofwhich is a federation of biological databases. We have focusedon incorporating data in relational databases and RDF Schema data, sincethese are widely used and are becoming increasingly popular formanaging data collections. An outcome of this work is a functioning prototype system thatapplies a uniform query and result processing approach, and hasa modular system design that is easy to use as a starting pointfor modifications and extensions.

Ämnesord

NATURVETENSKAP  -- Biologi -- Bioinformatik och systembiologi (hsv//swe)

NATURAL SCIENCES  -- Biological Sciences -- Bioinformatics and Systems Biology (hsv//eng)

NATURVETENSKAP  -- Data- och informationsvetenskap -- Datavetenskap (hsv//swe)

NATURAL SCIENCES  -- Computer and Information Sciences -- Computer Sciences (hsv//eng)

Nyckelord

Query processing

Functional data model

Rewrite rules

Data integration

Publikations- och innehållstyp

dok (ämneskategori)

vet (ämneskategori)