search

w3c / sdw

Repository for the Spatial Data on the Web Working Group
https://www.w3.org/2020/sdw/
148 stars 81 forks source link

Temporal-relations test suite #1238

Closed dr-shorthair closed 3 years ago

dr-shorthair commented 3 years ago

A set of RDF resources that can be used to test implementations of OWL-Time

dr-shorthair commented 3 years ago

@nicholascar some resources for an OWL-Time test-suite

nicholascar commented 3 years ago

Cool! We will have to implement similar tests for GeoSPARQL 1.1. now that the bar has been set...

dr-shorthair commented 3 years ago

@tguild @lvdbrink @jabhay please note that some of the resources in these tests are in the namespace http://www.w3.org/2021/time/test/ which may need to be reserved?

tguild commented 3 years ago

Create a placeholder html doc for now and I can publish as a placeholder. We can do content negotiation if we later wish to serve rdf, ttl etc

dr-shorthair commented 3 years ago

@tguild see https://github.com/w3c/sdw/blob/time-test-suite/time/test/index.html

dr-shorthair commented 3 years ago

@nicholascar @HolgerKnublauch could rules to test the temporal relations be written in SHACL?

nicholascar commented 3 years ago

@tguild see https://github.com/w3c/sdw/blob/time-test-suite/time/test/index.html

Better, see https://raw.githack.com/w3c/sdw/time-test-suite/time/test/index.html

nicholascar commented 3 years ago

@nicholascar @HolgerKnublauch could rules to test the temporal relations be written in SHACL?

I suppose they could be but I would personally find it easier to write a demonstration implementation in just SPARQL where I would calculate the time:intervalStarts etc. from the base data (probably all TIME object properties) and then compare the calculated results to each xxx-true.ttl / xxx-false.ttl file by graph comparison. Likely I just don't know how easy it is to embed SPARQL in SHACL or to create custom SHACL functions...

I could implement not a query language implementation though but a tool: an RDFlib extension. There I would simple make a new RDFlib class that calculates TIME relations for given data. It would obviously use the equivalent logic to a demonstration implementation in SPARQL but would use Python code against its own graph objects (faster than translating SPARQL). It would export created properties as graphs so results could be tested with this suite. I could then use this pattern (of a dedicated class) to calculate GeoSPARQL results too in efficient Python spatial toolkit code.

I'd then like to register the implementation here somewhere.

dr-shorthair commented 3 years ago

In the test files I have used only the time:inXSDDateTimeStamp to fix the positions of the time:Instant individuals that define the beginning and end of each time:Interval. So for these tests it is only necessary to find ordering relationships between literals of type xsd:dateTimeStamp (once you've chased down the paths to find them). I think that is built in to SHACL? An advantage of SHACL is explicit error messages.

nicholascar commented 3 years ago

An advantage of SHACL is explicit error messages.

Sure, and TIME validates in SHACL would be useful but you don’t need this test suite in order to create them, just implement the ontology rules!

When I think this test suite will be most useful will be in tool implementation, as per RDFlib. How else other than having test data to work on will we know that the tools are performing correctly?

tguild commented 3 years ago

Published the doc

dr-shorthair commented 3 years ago

@chris-little I thought I eliminated those - where did you find them?

chris-little commented 3 years ago

@chris-little I thought I eliminated those - where did you find them?

@dr-shorthair They have now gone. I must have been looking at an earlier commit.