Events metadata: Extra tags, event mining, and Bioschemas.org

[tnabtaf]

@bebatut would like the ability to identify events that are of different types. Her particular concern is to distinguish training events from non-training events. Our current event tag set (see #137) does not distinguish event types. It does contain a GTN tag, which is true/yes if the event is a training event presented by a GTN member, but that does not identify all training events.

In addition, it might be useful to identify other event types, like talks and hackathons.

Meanwhile, @dannon resists overbuilding/overthinking our events metadata. So, can we add this information in such a way that the changes are minimal?

Furthermore, Boioschemas has an event specification that we should really tie in to.

Bioschemas.org

Has an Event Specification that might be in draft form

Has an eventType field, which can have one of 4 values:

Value	Meaning
Workshops and courses	A workshop or a course event
Meetings and conferences	A meeting or a conference event
Receptions and networking	A reception or a networking event
Awards and prizegivings	An award or a prize giving event

That would work. A bit verbose, and does not have hackathons (probably fit under Meetings and conferences, although we'd

Do we want to go full on for Bioschema.org metadata?

Hmm. So should we embrace bioschema's metadata for all event data? That would allow sites like TESS to extract information directly from our events listings.

Maybe, sort of.

Bioschema does not support YAML. It does support JSON-LD and "Microdata", which is additional info in the generated HTML. Metalsmith expects YAML.

Some options:

1. Just stick with YAML. If TESS wants our events, they can parse our events directory in GitHub and map our terms to Bioscemas events annotations.
2. Include Microdata in the generated HTML in the events listing page. I think this would be unnecessarily complicated.
3. Generate JSON-LD for each entry in the events list. This would not be displayed, but would be scrapable.
4. Talked to @dannon. He suggested generating a JSON feed that uses Bioschemas JSON_LD to describe events. It would describe all events in the Galaxy Hub.

Mapping from YAML to JSON would be straightforward.

We'll do option # 4: Generate a JSON Bioschemas events feed for all events.

New event tag in Hub:

event_types

• Required
• Meaning
  • What type(s) of event/presentation is this?
• Use
  • Could be used to render icons in the events list. (Check font-awesome.)
• Format:
  • A JSON array of event types
  • Event type is one of:
  • Workshops and courses
  • Meetings and conferences
  • Receptions and networking
  • Awards and prizegivings (I don't think we'll have any of these)
  • Presentations (This is not a Bioschema event type. Map it to Meetings and conferences)
  • Hackathons (This is not a Bioschema event type. Map it to Meetings and conferences)
• Examples:
  • event_types: ['Workshops and courses']
  • event_types: ['Workshops and courses', 'Hackathons', 'Meetings and conferences', 'Presentations', 'Receptions and networking']
  • would describe a GCC.

Other related work:

• Need to document what the current event tag set is (it's not at all clear in #137 what we finally decided on).
• Automate verification of event metadata.
  • I have a script that checks all the event pages' metadata for conformance to standard.
  • @bebatut has a script that converts event location metadata to lat/lon.
  • We should combine those scripts to verify metadata, and report anything that can't be automatically mapped.

Schema.org

BioSchema's events are built on top of Schemas.org Events. Relevant subtypes are

• BusinessEvent - has no additional properties of its own
• CourseInstance
• EducationEvent - has no additional properties of its own
• ExhibitionEvent - includes tradeshows.
• Festival
• PublicationEvent - the event of publication for a CreativeWork of any type e.g. a broadcast event, an on-demand event, a book/journal publication via a variety of delivery media.
• SaleEvent
• SocialEvent

Just including this information in so we don't lose it.

[tnabtaf]

@njall actually implemented a lot of this at the May 2018 GTN hack in Cambridge. I asked him for some time to tell me what he needed, and then he implemented it while I went to the bathroom (well, in an absurdly short amount of time).

Bioschemas.org training event metadata was again discussed, and revisions suggested, at the November 2018 GOBLET meeting. @njall guided the discussion and (I believe) would present the suggested revisions to the Bioschemas group at the 2018 Paris Biohackathon. Once those settle, we need to apply them to the events listing.

In the meantime, we can work from the drafts that were updated at the GOBLET meeting.