14:49:06 <RRSAgent> RRSAgent has joined #linkml-wot
14:49:10 <RRSAgent> logging to https://www.w3.org/2024/09/25-linkml-wot-irc
14:49:10 <tpac-breakout-bot> RRSAgent, do not leave
14:49:11 <tpac-breakout-bot> RRSAgent, make logs public
14:49:12 <tpac-breakout-bot> Meeting: One Year Update: Using LinkML in Web of Things Specifications
14:49:12 <tpac-breakout-bot> Chair: Ege Korkan, Mahda Noura
14:49:12 <tpac-breakout-bot> Agenda: https://github.com/w3c/tpac2024-breakouts/issues/23
14:49:12 <Zakim> Zakim has joined #linkml-wot
14:49:13 <tpac-breakout-bot> Zakim, clear agenda
14:49:13 <Zakim> agenda cleared
14:49:13 <tpac-breakout-bot> Zakim, agenda+ Pick a scribe
14:49:14 <Zakim> agendum 1 added
14:49:14 <tpac-breakout-bot> Zakim, agenda+ Reminders: code of conduct, health policies, recorded session policy
14:49:14 <Zakim> agendum 2 added
14:49:14 <tpac-breakout-bot> Zakim, agenda+ Goal of this session
14:49:15 <Zakim> agendum 3 added
14:49:15 <tpac-breakout-bot> Zakim, agenda+ Discussion
14:49:15 <Zakim> agendum 4 added
14:49:15 <tpac-breakout-bot> Zakim, agenda+ Next steps / where discussion continues
14:49:16 <Zakim> agendum 5 added
14:49:17 <tpac-breakout-bot> tpac-breakout-bot has left #linkml-wot
15:13:31 <EgeKorkan> EgeKorkan has joined #linkml-wot
15:29:16 <JKRhb> JKRhb has joined #linkml-wot
15:30:06 <McCool> McCool has joined #linkml-wot
15:35:46 <EgeKorkan> scribenick: McCool
15:36:29 <anatoly-scherbakov> anatoly-scherbakov has joined #linkml-wot
15:36:34 <anatoly-scherbakov> present
15:36:37 <anatoly-scherbakov> present+
15:37:10 <mahda> mahda has joined #linkml-wot
15:37:19 <mahda> present+ Mahda_Noura
15:37:20 <EgeKorkan> rrsagent, draft minutes
15:37:22 <RRSAgent> I have made the request to generate https://www.w3.org/2024/09/25-linkml-wot-minutes.html EgeKorkan
15:37:44 <EgeKorkan> rrsagent, make minutes public
15:37:44 <RRSAgent> I'm logging. I don't understand 'make minutes public', EgeKorkan.  Try /msg RRSAgent help
15:38:13 <gkellogg> gkellogg has joined #linkml-wot
15:38:19 <betehess> betehess has joined #linkml-wot
15:38:28 <McCool> ege: two chairs, virtual
15:38:30 <gkellogg> present+ Gregg_Kellogg
15:38:53 <betehess> present+ Alexandre_Bertails
15:39:12 <McCool> ... ege intro - at Siemens, with Mahda noura, also research scientist at Siemens
15:39:28 <McCool> ... slides are public, also a link on event details in W3C system
15:39:49 <dape> dape has joined #linkml-wot
15:39:51 <JKRhb> present+ Jan_Romann
15:40:05 <dape> present+ Daniel_Peintner
15:40:21 <mahda> rrsagent, make logs public
15:40:34 <JKRhb> rrsagent, draft minutes
15:40:36 <RRSAgent> I have made the request to generate https://www.w3.org/2024/09/25-linkml-wot-minutes.html JKRhb
15:41:22 <McCool> ege: (policies and procedures review)
15:41:27 <McCool> topic: agenda
15:41:48 <McCool> ege: goal is to share our experiences with linkml, related to schemas
15:42:15 <McCool> ... some previous presentations
15:42:16 <anatoly-scherbakov> new
15:42:20 <McCool> ... (see slides)
15:42:52 <McCool> ege: in TPAC 2023 explained this use case, in schemata discussion
15:44:03 <McCool> ege: use case we have in WoT is managing spec documents, ontology documents, SHACL shapes, JSON Schemas, etc.
15:44:19 <McCool> ... all need a publication procedure, and need to be synced
15:44:26 <McCool> ... and will get worse, as we are adding a registry
15:44:42 <McCool> ... and that will involve people outside of W3C
15:45:07 <McCool> ... we had a custom-built tool but it is difficult to maintain
15:45:45 <McCool> ege: Mahda has done an analysis of other tools - see link in slides
15:45:57 <McCool> ... trying to understand options relative to requirements
15:46:09 <McCool> ... but linkml had the best feature set
15:46:21 <McCool> ... so we have been working on trying to use it for our use case
15:46:48 <McCool> mahda: goal is to translate from multiple sources to a single source of truth
15:47:01 <McCool> ... want automation with LinkML generators
15:47:16 <McCool> ... want to avoid manually crafting various artifacts
15:47:31 <McCool> ... simpler process with three main steps
15:47:47 <McCool> ... starting with a LinkML schema and then a set of different generators
15:47:59 <McCool> ... including generation of the standards document itself
15:48:37 <McCool> ... have a repo, https://github.com/w3c/wot-thing-description-toolchain-tmp
15:48:53 <McCool> ... install uv package manager, then "uv run wotis"
15:49:09 <McCool> ... (see slides for CLI interface)
15:49:34 <McCool> ... limitations: HTML generation does not use W3C style yet.  Need to add customization
15:49:44 <McCool> q+
15:50:29 <EgeKorkan> ack m
15:50:50 <McCool> mm: will the HTML gen also include static export from ReSpec and linting etc?
15:51:38 <McCool> mahda: first step is generating HTML sources, in particular aligning schemas and HTML
15:52:37 <McCool> ... we also have a lot of test files to manage to check that schemas are valid
15:53:24 <McCool> mm: just want to confirm that process will end at respec version and the rest of the steps will be manual
15:53:47 <McCool> mahda: (demonstrates tool)
15:54:07 <McCool> ... in WoT we have several different ontologies
15:54:31 <McCool> ... linkml starts with the definition of set of prefixes, using an OO approach
15:54:58 <McCool> ... three main concepts: classes, attributes, and types; there are also "slots" that can be reused across classes
15:55:01 <VladimirAlexiev> VladimirAlexiev has joined #linkml-wot
15:55:30 <McCool> ... can relate linkml classes to ontology definitions in RDF
15:55:37 <VladimirAlexiev> very nice stuff. Electrical CIM/CGMES wants to transition from Enterprise Architect to LinkML, but just starting. Discussion at https://github.com/Sveino/Spec4CIM-KG/issues/9
15:56:05 <McCool> ... can say whether slots are required, constraints, type (e.g. string) and so on
15:56:26 <McCool> ... note also that linkml schema is specified in YAML
15:56:56 <VladimirAlexiev> What SHACL can be generated from LinkML? Can I add my own constraints (eg SHACL SPARQL) and get them emitted together with simple SHACL generated from the schema?
15:56:58 <McCool> ... then we can generate a set of resources
15:57:40 <McCool> ... types can also relate to SHACL types
15:57:58 <McCool> ... most of our testing has been in JSON Schema
15:58:01 <VladimirAlexiev> can you configure the jsonld context not to use `@vocab` (some say that's "promiscuous") but a prefix `wot:` ?
15:58:03 <JKRhb> rrsagent, draft minutes
15:58:04 <RRSAgent> I have made the request to generate https://www.w3.org/2024/09/25-linkml-wot-minutes.html JKRhb
15:58:10 <McCool> q+
15:58:46 <EgeKorkan> q+ VladimirAlexiev
15:59:13 <McCool> mahda: not all of these resources that we generated are fully tested
15:59:22 <McCool> ... can also generate documentation
15:59:37 <McCool> ... later plan to use jinja or bikeshed
15:59:44 <McCool> ... to customize output
16:00:02 <McCool> ... can also generate diagrams using mermaid
16:00:14 <McCool> ... for relationships between classes
16:00:50 <McCool> mahda: weaknesses - linkml schema is quite expressive, but...
16:01:10 <McCool> ... post-processing is still needed, and need customizations to meet W3C requirements
16:01:22 <EgeKorkan> q?
16:01:43 <McCool> ... also need to fix up e.g. JSON-LD context document
16:02:19 <McCool> ... unfortunately, linkml does not define all our requirements
16:02:39 <McCool> mahda: in slides have a list of additional features we would need
16:02:55 <McCool> ... one important thing is natively supporting JSON-LD multilanguage strings
16:03:49 <McCool> ... if we explictly define all the language tags ourselves it makes the schema very large
16:04:06 <McCool> ... next, missing scoped JSON-LD contexts
16:04:28 <betehess> q+ to ask if LinkML is expressive enough to model itself in the LinkML, and then can be validated using let's say SHACL
16:04:47 <McCool> ... e.g. in TD, properties has its own context
16:05:40 <McCool> ... also require open mappings for context extensions
16:06:02 <McCool> ... to integrate external vocabularies, e.g. for protocol bindings or security schemes
16:06:18 <McCool> ... in SHACL, can use "closed: false"
16:07:12 <McCool> ... still requires a fairly large effort to get everything to work well
16:07:28 <McCool> ... once schema gets large, it can get verbose
16:07:40 <VladimirAlexiev> Don't use native JSON numbers since it's unpredictable what datatype will be attached (eg 12345678901234567890 is `double`) and the number can be changed (approximated). Use JSON strings with `@type`
16:07:48 <McCool> ... to control, need to define some intermediate classes
16:08:19 <McCool> ... have tried to remove syntax of RDF, but mental model is similar
16:08:39 <McCool> q-
16:08:42 <McCool> q+
16:08:51 <gkellogg> q+
16:09:17 <McCool> vlad: there is desire to move to linkml in our community
16:09:32 <McCool> ... had question about how shacl is generated
16:09:49 <McCool> ... whether you can attach extra constraints and how those get combined
16:10:00 <McCool> mahda: don't generate complex SPARQL
16:10:16 <EgeKorkan> rrsagent, please set these logs world-visible
16:10:17 <McCool> s/SPARQL/SHACL/
16:11:13 <McCool> vlad: on your slides I noticed some JSON-LD numbers without quotes, e.g. may get converted to double, e.g. 1.2 might get converted, might round things off, etc.
16:12:18 <McCool> ege: do you mean example on slide 24?
16:12:42 <McCool> vlad: yes, for example, could be converted to scientific notation
16:13:29 <McCool> gk: work on JSON-LD
16:13:34 <EgeKorkan> q?
16:13:49 <McCool> ... JSON-LD does specify how to deal with numbers, but inherent representation of numbers in JSON is double
16:14:08 <McCool> ... there are some other rules, but it's correct that numbers are not "stable"
16:14:34 <McCool> ... whether you are testing your shapes as numbers or not, seems that is best after transformation to RDF
16:14:52 <McCool> ... can set properties to treat a string as a number
16:15:02 <McCool> q?
16:15:06 <McCool> ack v
16:15:06 <EgeKorkan> ack v
16:15:44 <McCool> ege: would like to discuss how to continue discussion, we have a regular call, please contact me
16:16:03 <McCool> bete: can I use linkML to validate linkML models?
16:16:28 <McCool> mahda: linkml schema itself is written in linkml, and there are linters
16:16:39 <EgeKorkan> q?
16:16:43 <EgeKorkan> ack b
16:16:45 <McCool> ack b
16:16:46 <Zakim> betehess, you wanted to ask if LinkML is expressive enough to model itself in the LinkML, and then can be validated using let's say SHACL
16:18:52 <McCool> mm: how do we manage assertions, etc?
16:19:08 <EgeKorkan> ack m
16:19:15 <McCool> ege: will still have to embed in text, and use another tool to extract them from HTML for implementation report, etc.
16:19:30 <McCool> gk: we are doing an update to support YAML and CBOr
16:19:40 <McCool> ... do you have any input on things we should consider
16:20:28 <McCool> mm: should discuss this in joint WoT/JSON-LD call later
16:20:39 <McCool> ege: but for linkml, don't yet do that
16:21:01 <McCool> mahda: do have one issue regarding round-tripping
16:21:21 <McCool> ... conflict between at-type and json-schema type
16:21:39 <McCool> ... our current solution does not work with framing
16:21:42 <EgeKorkan> q?
16:22:17 <EgeKorkan> ack g
16:22:17 <McCool> gk: question was really if linkml is being used to work around issues that JSON-LD is not dealing with
16:22:49 <McCool> mahda: lot of limitations in linkml right now, e.g. related to containers and sets
16:23:14 <McCool> mm: limitations of linkml or json-ld?
16:23:21 <McCool> mahda: linkml limitations
16:23:38 <McCool> q?
16:23:55 <McCool> q+
16:24:48 <McCool> mm: one comment is that we have an information model, then a serialization
16:25:22 <McCool> ... but CBOR-LD is definitely interesting, but needs a registry of contexts that can be used
16:25:49 <McCool> ege: we also want to have a registry and recommended prefixed
16:26:03 <McCool> ege: have created a slide with a summary
16:26:14 <McCool> ege: any other next steps?
16:26:42 <McCool> mm: re CBOR, should we also be reaching out to IETF?
16:26:47 <McCool> gk: yes, that would be good
16:27:27 <McCool> mm: may also want to reach out to DID, they have similar requirements
16:28:31 <McCool> mm: for later, we want to discuss canonicalization
16:28:56 <McCool> ... rdf canonicalization is a bit heavyweight for our purposes
16:29:42 <McCool> gk: although that is not JSON-LD per se
16:29:42 <EgeKorkan> q?
16:29:52 <McCool> mm: and also unrelated to linkml, probably
16:30:02 <McCool> ege: thanks everyone, adjourn
16:34:33 <gkellogg> gkellogg has joined #linkml-wot
16:51:24 <gkellogg> gkellogg has joined #linkml-wot
17:01:44 <gkellogg> gkellogg has joined #linkml-wot
17:01:47 <JKRhb> JKRhb has joined #linkml-wot
17:02:32 <betehess> betehess has joined #linkml-wot
17:27:42 <gkellogg> gkellogg has joined #linkml-wot
17:43:17 <betehess> betehess has joined #linkml-wot
18:08:12 <gkellogg> gkellogg has joined #linkml-wot
18:09:44 <gkellogg> gkellogg has left #linkml-wot
18:15:05 <betehess> betehess has joined #linkml-wot
19:15:45 <betehess> betehess has joined #linkml-wot
20:15:34 <betehess> betehess has joined #linkml-wot
20:21:04 <betehess> betehess has joined #linkml-wot
21:30:31 <betehess> betehess has joined #linkml-wot
21:38:44 <betehess> betehess has joined #linkml-wot
22:09:10 <betehess> betehess has joined #linkml-wot
22:32:08 <EgeKorka_> EgeKorka_ has joined #linkml-wot
22:44:14 <betehess> betehess has joined #linkml-wot
22:52:05 <betehess> betehess has joined #linkml-wot
22:55:19 <EgeKorkan> EgeKorkan has joined #linkml-wot
22:55:44 <EgeKorkan> EgeKorkan has joined #linkml-wot
22:58:09 <EgeKorkan> EgeKorkan has joined #linkml-wot
23:00:52 <betehess> betehess has joined #linkml-wot
23:03:15 <EgeKorka_> EgeKorka_ has joined #linkml-wot
23:03:48 <EgeKorka_> EgeKorka_ has joined #linkml-wot
23:55:30 <betehess> betehess has joined #linkml-wot