LD-FusionTool
Data Fusion and Conflict Resolution tool for Linked Data
About LD-FusionTool
The value of Linked Data lies in the ability to link pieces of data, even across data sources. A data integration process can provide a unified view on the data. Tools covering all steps of the integration process need to be designed and developed in order to leverage the potential for semantic applications and the enterprise environment where data integration is crucial.
LD-FusionTool covers the Data Fusion step in the integration process for RDF, where data are merged to produce consistent and clean representations of objects, and conflicts which emerged during data integration need to be resolved. This involves several tasks that LD-FusionTool tackles:
- Resolution of schema and identity conflicts - different identifiers (URIs) for resources and properties from used ontologies (discovered, e.g., by data deduplication and schema matching) need to be resolved.
- Resolution of data conflicts - multiple (potentially) conflicting values of a property need to be resolved according to chosen policies. Conflicts can emerge, e.g., when integrating different sources or after deduplication.
- Quality assessment - the output may optionally contain metadata estimating quality of the produced RDF triples. Unlike other tools, LD-FusionTool evaluates the quality of fused data (as opposed to source data) and levarages infomration available during data fusion.
- Provenance tracking - the origin of all fused RDF triples can be traced back to the original sources.
In addition, LD-FusionTool leverages the structure of the RDF graph by considering dependency of properties and resources in order to simplify data fusion and improve the quality of results.
How LD-FusionTool works
The figure in this section depicts a high-level architecture of LD-FusionTool.
For more information about how LD-FusionTool works, please refer to the following publications:
- New Directions in Linked Data Fusion (poster at ISWC 2014)
- Linked Data Integration with Conflicts (submission to Journal of Web Semantics, 2014)
- Linked Open Data Aggregation: Conflict Resolution and Aggregate Quality (COMPSAC 2012, 1st IEEE International Workshop on Methods for Establishing Trust with Open Data)
- Linked Data Fusion in ODCleanStore (demo at ISWC 2012)
Running LD-FusionTool
LD-FusionTool is distributed as an executable java .jar archive. You can download the binary along with examples here.
LD-FusionTool can be run with Java, e.g., with the following command:
java -jar odcsft-application-<version>-executable.jar
Running it without parameters prints out the usage message with possible options.
What and how should be fused is configured with an XML configuration file. See examples, which also serve as the documentation of the configuration XML, and wiki page about conflict resolution configuration.
LD-FusionTool in UnifiedViews
LD-FusionTool was originally developed for a Linked Data processing framework ODCleanStore (hence it was originally named ODCS-FusionTool), which has been superceded by ETL tool UnifiedViews developed as a joint project of Charles University, Semantic Web Company GmbH, Semantica.cz s.r.o., and EEA s.r.o.
The version for UnifiedViews is available as a Data Processing Unit at GitHub.