- From: Henry Story <henry.story@bblfish.net>
- Date: Sat, 18 Jan 2020 00:51:38 +0100
- To: semantic-web <semantic-web@w3.org>
Hi, I came across Coherent Logic recently. Apparently it is as expressive as First Order Logic. And I found that it was used by Jos De Roo’s EYE implementation of an N3 reasoner. [1] I was wondering what the feedback of its use was in the field, and return on experience on how it fit into the Semantic Web stack. I came across it by reading an excellent 2017 paper by Evan Patterson [2] "Knowledge Representation in Bicategories of Relations” Where David Spivak (MIT) has put together some very elegant work showing how Category Theory could be applied to Databases, and in a number of articles tying these to RDF and SPARQL, the problem has been that his Database instances are functors from a small Category playing the role of a Schema into the Category Set, where objects are Sets and morphisms are functions. This does not fit well with RDF as many relations such as foaf:knows are not functional. By adapting this functorial semantics and instead of using normal Categories for Schemas, Patterson uses Bicategories of relations which can have morphisms between morphisms (giving us inference). Then when representing DB instances as functors into the Category Rel, where objects are Sets and morphisms are relations, we get much closer to RDF. Indeed Patterson starts off his discussion with Description Logics. (Note by the way that both Spivak and Patterson, point to a fundamental concept in Category Theory known as the Grothendieck construction that takes a tabular database and turns it into the flattened structure of RDF, this itself being essential in analyses of SQL or SPARQL Queries) Now the first part of the paper shows that ”regular logic is the internal language of bicategories of relations”. The final section shows that ”distributive relational ontology logics (ologs)” correspond to Coherent Logic. This way of putting things gives a special place to ”regular logic” and ”coherent logic”. So I searched around and found the latter used by Jos de Roo’s N3 reasoner EYE, which seems to somewhat confirm Patterson’s modeling of RDF. Henry Story [1] See Twitter thread https://twitter.com/bblfish/status/1215024256985247745 [2] https://www.epatters.org/assets/papers/2017-relational-ologs.pdf
Received on Friday, 17 January 2020 23:51:49 UTC