-
Notifications
You must be signed in to change notification settings - Fork 0
Home
Learning Expressive Linkage Rules from Sparse Data. Petar Petrovski and Christian Bizer. Under review at the Semantic Web Journal (SWJ), 2018.
We provide three running examples: phones, headphones and tvs. Running any of the examples is a two step process.
First prepare the matching project by copying one of the example projects found in silk.2.6-GenLinkSA/silk-learning/src/main/resources/to the silk working directory. In Linux/Mac systems the default working directory is installed on ~/.silk, and on Windows C:\Users<Username>.silk. For preparation of other silk projects please see the Silk framework wiki.
Next run silk.2.6-GenLinkSA/silk-workspace/src/main/scala/de/fuberlin/wiwiss/silk/workspace/scripts/CrossValidation.scala to perform a cross validation experiment (java CrossValidation.scala). As of now changes of the experiment can be done in the script itself. We are working on a version that could be parametrized.
The silk project for the phones matching task can be found silk.2.6-GenLinkSA/silk-learning/src/main/resources/phones.
The silk project for the headphonesphones matching task can be found silk.2.6-GenLinkSA/silk-learning/src/main/resources/headphones.
The silk project for the tvs matching task can be found silk.2.6-GenLinkSA/silk-learning/src/main/resources/tvs.