11 Comments
User's avatar
Benjamin Cogrel's avatar

Nice post! As a nuance, SKOS doesn't declare skos:broader as transitive so as to be robust to "dirty" thesauri. Instead they introduced a super-property (skos:broaderTransitive) for doing the inference automatically: https://www.w3.org/TR/skos-primer/#sectransitivebroader .

But for good thesauri, it is usually fair to think of skos:broader as transitive.

Expand full comment
Jessica Talisman's avatar

Great add here. Thank you!

Expand full comment
Burak Cansoy's avatar

Thank you for that gem!

Expand full comment
𝚐𝚎𝚘𝚛𝚐𝚎   @𝚓𝚎𝚗𝚜𝚎𝚗's avatar

Hey there. Masterful flow here. I'm a bit jealy of your finding a new voice for these ageless practices. We all need to hear it to breath life back into old science.

Expand full comment
Jessica Talisman's avatar

Oh wow, thank you. This was a really hard one to write. I appreciate

Expand full comment
John OGorman's avatar

Is this semantics or logic? If I don't know that 'bridge' can be a game or a structure, then narrower instances. like 'contract bridge' and 'suspension bridge' would fall under the same broader concept.

"For example, a SKOS‐aware reasoner can automatically infer that if Concept A is a skos:broader of Concept B, and Concept B is a skos:broader of Concept C, then Concept A is also a broader concept of Concept C. This kind of lightweight inferencing accelerates symbolic rule engines, enabling efficient query answering and classification, over a large thesauri."

Expand full comment
Jessica Talisman's avatar

Both. Ontologies introduce logical reasoning. Ontologies are semantic. Therefore ontologies are logical, semantic reasoning models. If you want to differentiate a game from a structure in SKOS, you add Bridge (Structure) and Bridge (Game) as a simple disambiguation setting. Simple parentheticals to differentiate.

See :

ANSI Z39.19: https://www.niso.org/publications/ansiniso-z3919-2005-r2010

ISO 25964: https://www.niso.org/schemas/iso25964

Expand full comment
Zane Hall's avatar

"a SKOS thesaurus will deliver immense value, with or without AI. And with SKOS, you have built a knowledge graph, or what I like to call “knowledge graph lite”."

Great perspective. This might have some parallels in the structured data / BI world. Thanks!

Expand full comment
Jessica Talisman's avatar

Immensely valuable for BI. Great observation, Zane!

Expand full comment
Michael Thompson's avatar

Just what I needed today - was deep diving with someone on ontologies! Thank you! This part of your post, reflecting on the persisting importance of ontology : "Emerging AI research is discovering that symbolic AI is needed to drive accurate reliable AI results and even more critically, to bolster AI’s natural language conversational elements" - any recent readings you can recommend on that direction? That was a big and interesting statement in itself!

Expand full comment
Jessica Talisman's avatar

Hi Michael, there’s tons of research out there. I would read Gary Marcus’s paper linked in my article.

This paper is a great, comprehensive literature from 2024, about neuro-symbolic and sub-symbolic AI.

https://arxiv.org/abs/2501.05435

Expand full comment