MSc Johannes Härtel

MSc defense Johannes Härtel

Metadata

  • Date: 14 November 2016 (Monday)
  • Time: 18.15
  • Room: B 013
  • Presenter: Johannes Härtel
  • Supervisor: Ralf Lämmel
  • 2nd reviewer: Marcel Heinz

Title

Relating Data Sources along Multiple Dimensions of Heterogeneity

Abstract

Heterogeneity between data sources can be motivated by different reasons: Heterogeneity in syntax is present if both are not written in the same lan- guage (e.g. software and natural language); in terminology if naming of en- tities is not unique; in coverage if both are describing overlapping, but not exactly the same content; in perspective if both are intended to fulfill differ- ent purposes and in granularity if the level of detail varies. Related work is reviewed on methods and parameters on how to measure or decrease such influence when relating data. This work presents the Ascomara approach, a highly configurable method for drawing a relation between data sources with (i) syntactic heterogeneity present handled comparable to works of Antoniol et al. and (ii) with heterogeneity in granularity present handled comparable to works of Kuhn et al. and Salton et al. Further, the approach defines a set of views that allows the inspection of the relation with the intention to support developers when improving a data source. Variation points of the approach are given in terms of several parameters and views that serve as input and output. One case-study analyses the parameters impact on hetero- geneity dimension’s influence on the relation. A second case-study describes the views and an improvement of a data source.

Download

In preparation.