- From: Antonin Delpeuch <antonin@delpeuch.eu>
- Date: Thu, 3 Dec 2020 18:54:53 +0100
- To: "public-reconciliation@w3.org" <public-reconciliation@w3.org>
Hi all, Last month I had the opportunity to present our work to the Ontology Matching workshop, part of the International Semantic Web Conference (ISWC). http://om2020.ontologymatching.org/ (I am not sure if the video is available anywhere). This community develops algorithms to match ontologies, but is also interested in other tasks such as matching between instances of concepts (which is closer to our notion of reconciliation). They run yearly competitions where systems are evaluated on test datasets: - the Ontology Alignment Evaluation Initiative (http://oaei.ontologymatching.org/2020/) - the SemTab challenge (http://www.cs.ox.ac.uk/isg/challenges/sem-tab/) My aim with this presentation was to convince them of the interest in using web APIs like ours in their systems. In particular, I think it would be great if such a web API could be used for communication between the matching system and the evaluator (which submits reconciliation problems to the system). The idea has generated some enthusiasm but of course quite some discussions need to happen to get clear on how much overlap there can be between the two use cases. Vladimir has noted on GitHub that we could take some inspiration from the SemTab tasks to add new features to the protocol: https://github.com/reconciliation-api/specs/issues/59 Among others: - joint reconciliation of multiple columns - dedicated endpoint to guess the type of a column, or the relation between two columns (as a property) - reconciling without specifying the properties linking the various columns supplied I am optimistic about what could come out of those discussions, since those are two communities doing very similar things but which have not really been talking together so far! What do you think? Best, Antonin
Received on Thursday, 3 December 2020 17:55:08 UTC