Defining and Using Schematic Correspondences for Automatically Generating Schema Mappings

Abstract

Abstract. Mapping specification has been recognised as a critical bottleneck to the large scale deployment of data integration systems. A mapping is a description using which data structured under one schema are transformed into data structured under a different schema, and is central to data integration and data exchange systems. In this paper, we argue that the classical approach of correspondence identification followed by (manual) mapping generation can be simplified through the removal of the second step by judicious refinement of the correspondences captured. As a step in this direction, we present in this paper a model for schematic correspondences that builds on and extends the classification proposed by Kim et al. to cater for the automatic derivation of mappings, and present an algorithm that shows how correspondences specified in the model proposed can be used for deriving schema mappings. The approach is illustrated using a case study from integration in proteomics.

A Functional Model for Dataspace Management Systems

“…then we have shown in [47] how it is possible to derive the view s1.worker ← s2.partTimeWorker ∪ s2.fullTimeWorker that populates the construct in the s1 side of the mapping in terms of the constructs in the s2 side, as well as the views s2.partTimeWorker ← σ s1.numberOfHours<8 s1.worker s2.fullTimeWorker ← σ s1.numberOfHours≥8 s1.worker that populates the constructs in the s2 side of the mapping in terms of the construct in the s1 side. In summary, in the case of mappings, we sometimes construe them intensionally as views and sometimes extensionally, as a set of mapping elements (i.e., pairs of sets of constructs).…”

“…As mentioned above, in [47] we have shown how viewGen can, in fact, generate executable expressions (in our case, view expressions against the supermodel we use). In the more general case, the operation is expected to select from the set of semantic correspondences those that are available for use by evalIQ in generating queries for evalQ, as described above.…”

“…Such an expression can be understood as an intensional description of the mapping (understood, extensionally, as a set of pairs of constructs). As we have shown in [47], given correspondences whose semantic import leads fairly directly to the derivation of view expressions that reconcile some semantic heterogeneities [41], we can algorithmically derive from them the view expression that allows us to populate the source construct(s) with data obtained from the target construct(s) (and vice versa), and make it a bound feature in the feature set of a mapping. For example, if there is a semantic correspondence relating 'worker' in s1 to 'partTimeWorker' and 'fullTimeWorker' in s2 and if its feature set contains…”

“…Now, invoking f := inferCorrespondences(a) derives from a subset of the associations in a a set f for each of whose elements, on the basis of matching evidence (i.e., the similarity scores associated with the elements of a), it can be postulated that it can be interpreted as a semantic correspondence. In our own work [47], we have focussed on the semantic correspondences identified in [41] (essentially, same name for distinct constructs, distinct names for the same construct, missing constructs, horizontally-or vertically-partitioned constructs). f := compose( f 1, f 2) Given two sets f 1 and f 2 of semantic correspondences,…”

“…Among the different conceptions associated with dataspace management systems (DSMSs), we have pursued one [6,32,33,47] founded on model management research [2,8,9]. Here, we build upon our previous work with a view to describing a comprehensive model of dataspace functionality using an algebraic style.…”