Contents

RDF Application Profiles - 2014-04-16 telecon

2014-04-16 Wednesday, 9:00-10:00 Boston, 15:00-16:00 Berlin

This meeting will be held as a "Google Hangout on Air" to allow screensharing with many viewers via a live video stream to YouTube. YouTube viewers will be able to submit questions and participate in discussion via IRC. A few "active participants", such meeting chairs and presenters, will be able to broadcast their voices directly on the Hangout (see "Joining instructions" below). The number of active participants is limited by the Google Hangout platform.

Key links

RDF Application Profile Task Group (Kai Eckert)

Kai will briefly present a proposal for a Task Group charter.

Bibframe application profiles (Eric Miller)

Eric will demo software created for Bibframe application profiles.

Scribe notes

RDF Application Profile telecon - 2014-04-16 Wednesday, 9:00 EDT

https://www.youtube.com/watch?v=179RJzwOhdM

Agenda:

Kai: Introduction to document giving charge to the task group

http://docs.google.com/document/d/1818LQDK-Ame3OPWtKDPWD9SiWJISblbj9aljj0-hLUc/edit anyone can comment; see Kai if you wish to edit

Time frame? Kai suggests two years.

Comment after this call; finalize at next call

Presentation by: Eric Miller

TomB: Eric was involved in DCMI in early discussions about application profiles.

Bibframe scribe: a prototype linked data editing tool driven by profile mechanism. You can load pre-defined profiles for different materials; show different capabilities. Produces template-based form.

goal: subset of RDF that solves the task but stays true to RDF basics

It is a work in progress - proof of concept

It allows subsetting vocabs to support individual needs or purposes

There is tension between under-specificity and over-specificity

BIBFRAME

  • underspecifies RDF schema, provide way to create profiles, then observe what people do
  • tool-driven approach
  • most discussion is about model, but little about the end-user

MARC - has lasted 45 years, and has created a community

  • need not only to create bib framework, but something that also creates a community
  • not just libraries, but LAM
  • now we need to take advantage of web

Bibframe

  • not "just" MARC replacement, but a set of building blocks
  • both complex description and simple

Tools that hide complexity, but that are RDF-compatible underneath

  • editing and input tool
  • load pre-defined profiles into tool
  • form-based input + external services (vocabularies); auto-complete from vocabularies
  • order - important for cataloging, but not for RDF
  • allows value constraints: Ranges for fields, etc. Done through "descriptionTemplateRef" in template. Template specifies permitted values
  • schema allows all properties and values; subset adds constraints, permissible values
  • profiles allow libraries to add their own properties for specific needs

profile templates with

  • resource template
  • property template
  • value template

Q: Is there a JSON specification?

A: I’m waiting on LC to let me know their preference on FLOSS licenses before releasing the code and documentation.

Tried in a previous project to implement DSP, but it was difficult. But there is a lot of overlap (the needs to be fulfilled are similar)

JSON is not the only serialization.

Property templates can have specific value templates, value template can take diff forms: Array of permissible resources, data type, default value...

Great value in profiling as a way to have discussion with end users who are less interested in the technology and more in the impact on their work and their goals

Value of LD on web increases as links increase; this platform focuses on accelerating linking. "Value proposition for any data asset on the web is proportionate to the number of links to it."

Different data types can be treated differently (e.g. images), and metadata can be extracted from them automatically. Tool is more about creating connections than about adding descriptive metadata by hand.

Q: Has this been tried with catalogers?

  • one: head of cataloging at LC, but no stress test yet; providing the idea that cataloging is not MARC
  • UC Davis will be using Bibframe in their project

Q: Wasn't there work like this at MIT?

A: There are other projects working on things like this. We try not to use words like "ontology", "linked data", but focus on user experience.

Q: Matthias: last time we showed RDFForms templates; have you looked at this? Perhaps we could compare these initiatives. Is there a presentation mode? Is the goal to be a stand-alone solution or an embedded solution?

A: This is designed to be embeddable in a wide range of workflows. Have not focused on presentation mode; have other presentation tools. Emphasis of integrating external vocabulary lists, to seem as a single list, with type-ahead. Use rest-ful interfaces to external services for now. Implications in terms of reliability and network performance: libraries will need to put "big iron" around these to provide caching services, aggregation services, etc. That's in the future.

Q: Matthias: Are profiles re-usable? Do profile snippets include link to data sources? If so, are they less re-usable?

A: There is tension between re-usability and utility. Re-usability at entity level not ontology level. There is no "vocabulary picker" in this. Where does a profile begin, and where end? End-user search points are not included: those are in the application. Where do you draw the line between profile and applications? e.g. lookahead is not in profile, but in cataloging application.

Q: Is UI built directly from profile?

A: Yes.

Q: Matthias: do you have location for languages?

A: Yes, but as .de .sp. Not in profile, but in application.

  • uses angular JSON translated to RDF

Q: Could these structures be used for validation?

A: if you mean pattern consistency, than yes, this moves in that direction. But people use validation to mean other things.

Q: Cardinality?

A: Yes; cardinality is in profile - but limited to repeatability.

Joining instructions for active participants

Due to a limitation in in the Google Hangout platform, only ten "active participants" can participate with audio (and video). "Active participants" should:

  • have a Google Plus account;
  • install the "hangout plugin" and test in advance that it works;
  • join the hangout at 8:40 New York (14:40 Berlin) to test that it works. A notification will arrive if Gmail or Google+ is open. We will also send an email with a direct link to the hangout. Just enter as soon as you get it so we can make sure everything works ahead of time.

Other links