Questions on data for task D

28 views
Skip to first unread message

Ryn Gray

unread,
Jun 17, 2025, 1:42:39 PMJun 17
to LLMs4OL Challenge
Hello,

I have been reviewing the data for task D (mostly SWEET so far) and would like some clarification regarding the symmetric edge types. For instance, the property "equivalentClass" is generally symmetric in the owl ontology, meaning that if is equivalent to (i.e., ), then should also be equivalent to (i.e., ). However, I noticed that in the relations file, the edges are generally defined in only one direction.

Could you please clarify how I should handle this data? Specifically, does the system expect a specific order for the head and tail of the relationships, or will it account for both directions?

Thank you for your assistance!

LLMs4OL Challenge

unread,
Jun 21, 2025, 12:38:39 PMJun 21
to Ryn Gray, LLMs4OL Challenge
Dear Ryn,

Thanks for your question.

You are right -- equivalentClass is symmetric. While the dataset includes only one direction, the evaluation script accounts for both. So, predicting either (A, equivalentClass, B) or (B, equivalentClass, A) will be accepted. You are free to handle this in your training data!

About the order, yes, other than symmetric relationships, order matters! (Head, Relationship, Tail). And here is are list of symmetric relationships for your reference: "equivalentclass", "sameas", "disjointwith"

I hope this clarifies your questions.

Best Regards
Hamed
On behalf of the LLMs4OL challenge organization team!



--
You received this message because you are subscribed to the Google Groups "LLMs4OL Challenge" group.
To unsubscribe from this group and stop receiving emails from it, send an email to llms4ol-challe...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/llms4ol-challenge/bb9446f8-145f-4f24-8d6f-1c8fba47a9dbn%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Reply all
Reply to author
Forward
0 new messages