LLD XG -- 20 Jan 2011

<emma> Previous: 2011-01-13 http://www.w3.org/2005/Incubator/lld/minutes/2011/01/13-lld-minutes.html

<antoine> :-)

<emma> Scribe: Michael

<emma> scribenick: michaelp

Admin

<Asaf> ??P35 is me

Emma: Acceptance of minutes of previous telecon.
... Objections?
... Minutes accepted.

<antoine> s/Munites/Minutes

http://www.w3.org/2005/Incubator/lld/minutes/2011/01/13-lld-minutes.html

--RESOLVED

RESOLUTION: http://www.w3.org/2005/Incubator/lld/minutes/2011/01/13-lld-minutes.html

Emma: Special conference for people from Asia timezones

TomB: Problems with audio, timezones. Schedule extra teleconference that is well-scribed.

<scribe> Scribenick: michaelp

<scribe> Scribe: michaelp

<kcoyle> what time would work?

TomB: We have to make sure that UC of this group are not rejected just because of time zones and geography.
... We will figure out a time.

USE CASE CLUSTERS

CLUSTER Bibliographic data: curators Gordon, Martin

http://www.w3.org/2005/Incubator/lld/wiki/Cluster_BibData

GordonD: Cluster is pretty well finished.
... We added an extra UC on migration of legacy library data. That may add some extra work, however.
... Questions?

Emma: Could you highlight main issues?

<Asaf> Question: Do we expect any impact by this group's work on the work of the FR* and RDA groups?

GordonD: Two main issues. Firstly, unit of processing is bibliographic record in library environment.
... Every change in an attribute in a record requires duplication of the entire record for local use.
... Causes duplicated data, which is problematic for triple migration.
... Creates identical triples.
... But main issue is still shift from record to triples.

<Asaf> GordonD: Thanks! That's very encouraging!

GordonD: The UC clustering has been valuable with regards to the FR* and RDA groups.

<Zakim> TomB, you wanted to ask about levels between statement and record, e.g., WEMI

TomB: What about intermediate levels between records and statements, like aggregations like FRBR entities?

GordonD: Records is aggregation of statements, whether work, expression, ..., records.
... Traditionally, we have been looking at aggregations. LLD challenges this assumptions by looking at statements on their own.

<emma> http://www.w3.org/2005/Incubator/lld/wiki/Granularity_of_library_metadata

GordonD: Granularity is not downwards from single resource, but goes up from aggregations, etc., themselves.
... Aggregations like collections, series. Levels of granularity of these is often mixed up by libraries.

<Zakim> antoine, you wanted to ask about record issue in "problem and limitations"

Antoine: Granularity important issue for data conversion.
... We should but reference to this from cluster page.
... Other issue: Relevant technologies. Case about de-duplication. Single case or are there more like this?

GordonD: This came up in more than this UC.

<Zakim> jeff_, you wanted to consider owl sameAs as a factor in granularity

Jeff: owl:sameAs and levels of granularity.
... It would be nice to pull resources together, but this issue makes it difficult.
... We have numbers of documents that contain triples. It is useful to be able to relate those without linking raw triples.
... Information can be spread across multiple records, but refer to the same thing.
... It would be nice if triples could be pulled together into a unified record.

GordonD: I see a number of problems.
... We shift focus to the triple, then we build aggregations of triples that suit the needs of an organization.
... This gives us a great deal of flexibility in rebuilding records.

<TomB> Gordon: we shift attn to triple, then build up aggregations. So if you want MARC-like view, you have an application profile - what attributes need to be part of the profile? Flexibility in building records - my last slide in presentn on "disintegration of the record".

GordonD: The same triple can appear in many different "record" structures.

<kcoyle> :-)

GordonD: Aggregations can produce conflicts as they are built at will.
... Conflicting statements, missing data.
... because of missing or present owl:sameAs

marcia: We have a project where multiple databases have to be aggregated.
... People have more then one option in terms of vocab selection.

<TomB> Is Gordon referring to http://gordondunsire.com/pubs/pres/EvolCatRec.ppt (for "disintegration of the record")?

marcia: Individual statements become the elements.

<jeff_> OWL provides mechanisms for reconciling equivalent/sub classes across vocabularies.

GordonD: Mapping of properties will become feature of specific namespaces (vocabs)

<TomB> Gordon: Developing environment will have FRBR, ISBD, and RDA elements in parallel - people will have a choice which to use. Mappings need to be part of those namespaces. Who does the mapping? If that isn't done, we have a third layer of complexity - subjects may have URIs but are sameAs - same with predicates and objects.

GordonD: If namespaces are semantically linked, it gives users free choice which to use..

<TomB> ...Application Profiles are important because they specify which namespace elements taken from. Concern about unresolved debate about APs versus OWL.

GordonD: The profile will say which namespace the element is taken from, but there is the unresolved debate in terms of DC application profiles and their relationship to OWL.

<scribe> ACTION: Gordon and Martin to curate bibliographic data cluster for end of December [recorded in http://www.w3.org/2005/Incubator/lld/minutes/2010/10/23-lld-minutes.html#action05] [CONTINUES]

<scribe> ACTION: Gordon and Martin to curate bibliographic data cluster for end of December [recorded in http://www.w3.org/2005/Incubator/lld/minutes/2010/10/23-lld-minutes.html#action05]

<scribe> --continues

Trying to find the Action, sorry

<scribe> ACTION: Gordon and Ross to review use cases in light of migration of legacy data [recorded in http://www.w3.org/2011/01/13-lld-minutes.html#action01] [DONE]

<GordonD> Tom, yes

<emma> http://www.w3.org/2005/Incubator/lld/wiki/Use_Case_Migrating_Library_Legacy_Data

emma: I wanted to mention the new social uses cluster.
... They will have a call for UCs.

PAGES RELATED TO USE CASES

-- Terminology: "value vocabulary; metadata element set, and dataset"

http://www.w3.org/2001/sw/wiki/Library_terminology_informally_explained#Vocabularies.2C_Element_sets.2C_Datasets

Emma: Mark, would you like to report?

markva: Feedback on mailing list was good.
... We are on a good path.

<antoine> +1

<Asaf> Dispersing some of the fog around the terminology was very, very important. It would be very important for the readers of our final report, too. Mark++!

<markva> :-)

<scribe> ACTION: Mark and Emmanuelle to clean definition of value vocabulary and use of attribute [recorded in http://www.w3.org/2005/Incubator/lld/minutes/2011/01/06-lld-minutes.html#action02]

--> http://www.w3.org/2001/sw/wiki/Library_terminology_informally_explained#Vocabularies.2C_Element_sets.2C_Datasets

<scribe> --done

<antoine> I think almost everyone in the XG participated that discussion :-)

<scribe> ACTION: Alex, Jeff, Martin, MichaelP elaborate on general purpose IT architecture for dealing with linked data with caching feature (short sketch for final report) [recorded in http://www.w3.org/2005/Incubator/lld/minutes/2010/10/24-lld-minutes.html#action04] [CONTINUES]

<Asaf> antoine: sure! I commended Mark for undertaking to synthesize and process all the input.

<antoine> http://ckan.net/group/lld -> good progress there!

<scribe> ACTION: Volunteers to send login information (openid credentials) to William Waite to curate LLD group on CKAN [recorded in http://www.w3.org/2010/11/25-lld-minutes.html#action04] [CONTINUES]

<scribe> ACTION: Everyone to update the Events page (http://www.w3.org/2001/sw/wiki/LLD/Events) on the wiki regularly [recorded in http://www.w3.org/2005/Incubator/lld/minutes/2010/10/24-lld-minutes.html#action05] [CONTINUES]

Emma: Don't forget to add to that page.

<antoine> http://www.w3.org/Submission/CBD/

Antoine: I just wanted to mention the issue of bounding statements for releasing linked data.

<kcoyle> interesting

Antoine: Goal is to create an "optimal unit of specific knowledge about that resource to be utilized"

<edsu> antoine: thanks, hadn't seen that before

Mark: I know about that spec. I used it in WordNet release.
... CBD connects statements about a subject, skipping blank nodes.
... Triples that have the same URI as subject. Very simple and algorithmic.

<Asaf> This was mentioned (without knowing about CBD) in our mailing list discussion, right?

edsu: Will that increase traffic to do something useful with the data?
... More roundtrips required to resolve other connected URIs.

markva: Depends on the application and your goals.

edsu: Human readable display, for example.
... You would have to do a number of further lookups.

markva: In WordNet case, you don't want to get the whole thing.

Jeff: GordonD was talking about synchronizing info from different sources. There is a UC for foaf:focus and SKOS.
... Even if you do the inferencing on the real world object, foaf: focus would help to trace back the origin of these statements.

<antoine> Michaelp: perhaps this is a provenance issue

<antoine> ... how you keep stuff together

<antoine> ... How much data you send back is a very interesting engineering issue

<antoine> ... It somehow dilutes the identifier

<antoine> ... it may suddenly become much more when you get more info for it than what you'd expect

<antoine> Emma: there are lots of ways to define what is useful for a URI

<edsu> "useful" is usefully vague imho :-)

<antoine> Michaelp: that relates to SPARQL DESCRIBE

<antoine> ... It may become really complex if you have a bunch of unconnected identifiers

<Asaf> edsu++

<marcia> good discussions!

<Asaf> bye everyone!

<markva> bye!

<emma> rrsagent please draft minutes

- DRAFT -

LLD XG

20 Jan 2011

Attendees

Contents

Admin

USE CASE CLUSTERS

CLUSTER Bibliographic data: curators Gordon, Martin

PAGES RELATED TO USE CASES

-- Terminology: "value vocabulary; metadata element set, and dataset"

Summary of Action Items

Scribe.perl diagnostic output