Semi-automatically Mapping Structured Sources into the Semantic Web

Abstract

Abstract. Linked data continues to grow at a rapid rate, but a limitation of a lot of the data that is being published is the lack of a semantic description. There are tools, such as D2R, that allow a user to quickly convert a database into RDF, but these tools do not provide a way to easily map the data into an existing ontology. This paper presents a semiautomatic approach to map structured sources to ontologies in order to build semantic descriptions (source models). Since the precise mapping is sometimes ambiguous, we also provide a graphical user interface that allows a user to interactively refine the models. The resulting source models can then be used to convert data into RDF with respect to a given ontology or to define a SPARQL end point that can be queried with respect to an ontology. We evaluated the overall approach on a variety of sources and show that it can be used to quickly build source models with minimal user interaction.

A Graph-Based Approach to Learn Semantic Descriptions of Data Sources

“…We then used our algorithm to learn a source description for each source given the manually created source descriptions of the other sources as training data (The original source descriptions and the results are available at: http://isi.edu/integration/data/iswc2013). Since the semantic labeling is not the focus of this paper, we assume that Algorithm 2 is given the correct semantic type for each attribute (we evaluated the semantic labeling in our previous work [13,24]). …”

“…We compared the results of our approach with the results of Karma [13], a data integration tool that allows users to semi-automatically create source descriptions for sources and services. To make the Karma results comparable to our approach, we also gave Karma the correct semantic types for each attribute.…”

“…Most of the proposed approaches on the Semantic Web focus on the first step of the modeling process. In our previous work [13], we presented an algorithm to construct semantic models of data sources by computing a Steiner tree in a graph derived from the ontology and the semantic types. If the suggested tree is not the correct model of the data, the user interactively imposes constraints on the algorithm through a graphical user interface to build the correct model.…”

Karma: A System for Mapping Structured Sources into the Semantic Web

“…In our example, the correct choice is the first one because the information in the table is about Pathways that involve our Gene. After users make a selection, Karma recomputes the Steiner tree, which is now required to include the class/property selections users make [5]). Figure 5 shows the correct, updated model.…”

“…When Karma is unable to infer the semantic type for a column, users can interactively assign the desired type; Karma uses the assigned type and the data in the column to train a CRF model to recognize the type in the future [4]. The semantic types are used by our Steiner tree algorithm to compute the source model as the minimum tree that connects the assigned semantic types via properties in the ontology (the details of the approach are published elsewhere [5]). Because the minimum model is not always the desired model, Karma provides a user interface to enable users to force this algorithm to include specific properties in the model.…”