i3MAGE: Incremental, Interactive, Inter-Model Mapping Generation

Pinkel, Christoph

Vorschau

PDF
dissertation_christoph_pinkel.pdf - Veröffentlichte Version
Download (6MB)

URL:	https://madoc.bib.uni-mannheim.de/40946
URN:	urn:nbn:de:bsz:180-madoc-409466
Dokumenttyp:	Dissertation
Erscheinungsjahr:	2016
Ort der Veröffentlichung:	Mannheim
Hochschule:	Universität Mannheim
Gutachter:	Stuckenschmidt, Heiner
Datum der mündl. Prüfung:	7 Juni 2016
Sprache der Veröffentlichung:	Englisch
Einrichtung:	Fakultät für Wirtschaftsinformatik und Wirtschaftsmathematik > Practical Computer Science II: Artificial Intelligence (Stuckenschmidt 2009-)
Fachgebiet:	004 Informatik
Normierte Schlagwörter (SWD):	Mapping-Problem , Semantisches Datenmodell , Relationales Datenmodell
Freie Schlagwörter (Englisch):	RDB2RDF , automatic schema mapping , inter-model mapping
Abstract:	Data integration is a highly important prerequisite for most enterprise data analyses. While hard in general, a particular concern is about human effort for designing a global integration schema, authoring queries against that schema, and creating mappings to connect data sources with the global schema. Ontology-based data integration (OBDI), which employs ontologies as a target model, reduces the effort for schema design and usage. On the other side, it requires mappings that are particularly difficult to create. Architects who work with OBDI hence need systems to support the process of mapping development. One key type of tooling to support mapping development is automatic or semi-automatic generation of mapping suggestions. While many such tools exist in the wider sphere of data integration, few are built to work in the case of OBDI, where the inter-model gap between relational input schemata and a target ontology has to be bridged. Among those that support OBDI at all, none so far are fully optimized for this specific case by performing a truly inter-model matching while also leveraging distinct but corresponding aspects of both models. We propose i3MAGE, an approach and a system for automatic and semi-automatic generation of mappings in OBDI. The system is built on generic inter-model matching, and it is optimized in various ways for matching relational source schemata to target ontology schemata. To be truly semi-automatic in every respect, i3MAGE works both incrementally, building mappings pay-as-you-go, and interactively in exchange with a human user. We introduce a specialized benchmark and evaluate i3MAGE against a number of other approaches. In addition, we provide examples, where i3MAGE can be deployed in holistic data integration environments.