#social 2020-11-03

2020-11-03 UTC
ilmu2, puck, sl007 and Joe joined the channel
#
Joe Hello all. Joe from Crossref here jwass@crossref.org . I'm interested in exploring Activity Streams for a possibly unconventional use.
#
Joe As a spec it looks like a really nice way of expressing "things that happened" with JSON-LD for the vocabulary. But it looks like it was designed to communicate human-originated activity, and the stock vocab is social-media oriented. I'm interested in using it to record activity mostly originated by machine proceses, recording the progress of metadata through a system. So it's neither human nor social.
#
Joe Has anyone here used it in a similar way?
Joe and sl007 joined the channel
#
nightpool[m] I will admit it doesn't strike me as the best use, but I guess it depends on the metadata and the system.
Joe and sl007 joined the channel
#
csarven Joe nightpool[m] AS2 is not restricted to human activities.
#
nightpool[m] it's not! but if you're trying to build, like, a distributed database with it there are probably better protocols
#
nightpool[m] there are also more general protocols that are less tied to the specific requirements of social networks, like WebSub and LDN
#
csarven nightpool[m]: It sounds like Joe is interested in using AS2, extending AS2, using another vocab, or remixing vocabs. All of which are possible. Yes, LDN. In fact, there were LDN implementation reports covering scientific / machine activities
#
csarven LDN+AS2 can also be used for research/scholarly activities.
#
csarven Noting that Joe is no longer in channel.
Joe joined the channel
#
Joe I'm back, thanks. Web client timed out.
#
Joe I had a chat with Rhiaro, looks like LDN might be a slightly better fit for these notifications.
#
Joe I am very keenly aware of the scholarly metadata aspect to linked data. Whilst ackowledging that, I'm trying to solve an adjacent problem.
#
Joe Could you point me to the scientific / machine activities uses of LDN?
#
csarven Data and "meta"data. Ack'ing that different communities call metadata what some communities would just call it data.
#
Joe And I guess the only meaningful difference in terms of data modelling between AS2 and LDN is the base vocab you get in the context?
#
csarven Wil lneed to dig.. but iirc https://github.com/albertmeronyo/linked-edit-rules and https://github.com/jpcik/ldn-streams were looking into that area re implementations.
#
Loqi [albertmeronyo] linked-edit-rules: Linked Edit Rules: a methodology to publish, link, combine and execute edit rules on the Web as Linked Data to verify consistency of statistical datasets. Compliant Linked Data Notifications (LDN) sender.
#
csarven But you can use literally any vocab. Search https://lov.linkeddata.es/dataset/lov/ .. and describe the 'notification' with one of htose.
#
csarven or multiple..
#
csarven AS2 is a vocab that you can use in LDN notifications.
#
csarven The LDN spec includes some AS2 examples... IIRC :P
#
Joe Thanks for that. I'm just picking through it all now, those vocab pointers are a great.
#
csarven But like I said.. you don't need to be fixed to AS2. Just find the most appropriate vocabs for your need. As last resort, just create your own vocab
#
Joe Thanks. Yeah, I'm trying to avoid reinventing too much wheel. I'll see if AS2 has enough overlap.
#
Joe Another Q I had was about filtering a stream of AS2 / LDN according to some kind of query. SPARQL looks like the right language to use, though I've never tried using it.
#
Joe I think it should work when applied to a single LDN object with no object graph?
#
csarven I'm guessing probably not by itself. But you could use it as a the generic bits. Use another vocab to express the specific bits of whatever the machine is generating.
#
Joe (Yes, I expect to extend the vocab one way or another.)
#
csarven See this example: https://www.w3.org/TR/activitystreams-core/#example-using-multiple-vocabularies
#
csarven That remixes AS2, PROV-O, Web Annotation
#
csarven So generic machine activities could also be captured with PROV-O. All depends on what the notification/activity/event needs to say.
#
csarven Hmm, no SPARQL would be to query.
#
Joe So I want to say "subscribe to a feed where xyz:member-id == 'xyz:member:123' and type == 'did-a-thing'." (Forgive the vocab, I'm new to JSON-LD.
#
Joe That looks superficially like SPARQL. Or could easily be implemented as generic GET query parameters. But creating a subresource for each one would be inflexible.
#
Joe And I guess I'd be defining a query grammar implicitly then.
#
csarven You don't need SPARQL. Think of SPARQL along the lines of SQL but on RDF graphs
#
csarven heh
Joe joined the channel
#
Joe Thanks, I think I'll avoid it then.
#
csarven Joe: Right.. not for what you're trying to achieve.. not for filtering with the protocols/vocabs mentioned above.
ilmu2 joined the channel