Linked Data refer to a set of best practices for publishing and connecting structured data on the Web using Semantic Web standards (public URIs, links to external resources, RDF formats), with the purpose of fostering reuse, linkage and consumption of data.
Linked Data consist nowadays in a globally available knowledge graph, where several datasets are represented in RDF standards, can be accessed and understood by machines and, most importantly, are connected across disciplines. We believe that Linked Data can help making the explanation process more automatic, relieving the effort that the experts put into interpreting patterns.
Dedalo uses an Inductive Logic Programming approach. Patterns are used as positive and negative examples to learn from (as in Machine Learning).
The background knowledge used for reasoning is build using an A* search over the Linked Data graph.
Dedalo traverses the graph and finds characteristics shared by the examples, and then selects the explanations that are most suitable to the positive examples (so they do not apply to the negative examples).
In this demo, we show scenarios from different domains in which, starting from patterns obtained from some data mining processes, Dedalo was able to generate plausible explanations for the grouping of those data.
Now choose a use-case among the following.
Why those people have written papers together?
Below you will see how a Network Partitioning algorithm has grouped researchers from the KMi department according to the papers they have written together (the connections). The closer two authors are, the more papers they have co-authored.
Can you guess why some people are more closer than others? If not, just pass the mouse on one of the researchers.