search

KnowWhereGraph / KWG-SHACL

This is the repository of SHACL shapes development/research for KWG.
1 stars 1 forks source link

WildFire Dataset Tested #9

Open ling-cai opened 2 years ago

ling-cai commented 2 years ago

I tested wildfire-mtbs dataset. The test code can be found here. The test report is also in this notebook.

Based on the validate report, I fixed some errors in our constructed shapes, including adding ssn namespaces for some shapes, adding 'sosa:ObservationCollection' to the domain constraint of 'sosa:hasFeatureOfInsterest.' and so on.

In general, the data shape conforms most of our shapes. There are still problems: one about actuation and one about observation (more details are in the notebook above). The latter violation is that each observation has to have exactly one FeatureOfInterest. However, once we use observation collection, we declare the relation 'hasFeatureOfInterest' between observation collection and feature of interest.In this case, there is no need to claim the relation between observation and feature of interest. We may need to change observationConstraints.

@zhurui0509 @cogan-shimizu

I am also not sure what should be in the paper regarding this testing.

zhurui0509 commented 2 years ago

Thanks a lot @ling-cai for this detailed report!

All these seem as expected. Regarding the obvervation and observationCollection, I suggest leave as it is for now, but in the paper, we will discuss about it.

For your assigned section, I suggest you to describe (1). how you modeled and triplifed your data? (2). any thing that would be missing if not using SHACL? (e.g., there might be missing sensor for observations) (3). in addition to SOSA-SHACL, where else SHACL can help better refine your data?

Again, great work! Thank you!