Seminar, 9 Nov 2009

Report of the Vocabulary Mapping Framework (VMF) Seminar held at the British Library, St. Pancras, London, 9 November 2009

Introduction and content provider perspective

Library perspective

The VMF matrix – what it is and how it may be used

Questions and discussion session 1

  • Audience member: It is necessary to stress the importance of version control of the VMF. It is crucial that the currency and accuracy of the VMF model is maintained.
  • Audience member: In cases where a mapping does not result in a best fit it would be useful to receive information as to why such a mapping does not fit fully.
  • Godfrey Rust: Current query testing uses SPARQL [see SPARQL query language for RDF].
  • Audience member: Can you explain more about the workings of mapping, for example, rules?
  • Godfrey Rust: This is uncertain. How do we use rules most effectively? The answer to this is that we really do not know yet as VMF has not been used /tested. Envisage a mixture of applying semantics and techniques alongside smart use of rules and queries. Will have to see what works.
  • Audience member: Do you foresee human coding in the rules?
  • Godfrey Rust: See this more as a question of techniques and applying these consistently.
  • Gordon Dunsire: Consumers need to know the implications of the rules.
  • Godfrey Rust: If users are given a choice this needs to be consistently clear, for example, always to map to higher level.
  • Audience member: If a user does not get an exact match when mapping they need to be clear what the difference is.
  • Audience member: In the process of mapping data is not always "homogeneous". Standards are changing. Will it be necessary to modernise schemes being mapped?
  • Godfrey Rust: If the change in pattern can be identified, then provided this can be defined, we can map.
  • Audience member: Question about "grooming" of data – how will the VMF be operated?  For instance will it be operated by a non-profit making organisation or perhaps by a consortium of users?
  • Godfrey Rust: This is the topic of the next session. Personally, envisage it to be rather like Wikipedia in operation.
  • Audience member: Will the qualifiers for mapping result in exact match?
  • Godfrey Rust: Mappings into VMF will always be exact. You will know whether parent/child/sibling.
  • Audience member: Comment – the purpose of VMF is to enrich data and there is a need to map identifiers (mapping individual identities of objects). VMF has to work in a larger infrastructure.
  • Godfrey Rust and Gordon Dunsire: Agree. If Scheme A wants to map to Scheme B via VMF, they do not "talk to each other". The schemes will find within VMF where in the hierarchy the mapping lies.
  • Audience member: What about mapping authority data. The standards used are not provided to be semantically coherent e.g. cataloguing standards. A validation stage is needed when mapping authority data.
  • Godfrey Rust: It is critical to get mappings authorised and this needs to be made explicit in the ontology. This is also especially important in terms of concepts and rights and if any money is to be exchanged.
  • Audience member: Does the ontology need to deal with authority in data?
  • Godfrey Rust: The VMF needs to deal with semantics to support this.
  • Mark Bide: There is such a vocabulary in MARC for authorship of a particular record.
  • Audience member: As VMF becomes more used will there be any validation to ensure that the hierarchy will be maintained and not made more distant by additions?
  • Godfrey Rust: There has to be. This is an essential requirement of VMF.

Where do we go from here?

  • Slide presentation (PowerPoint Show): Next steps by Norman Paskin (Tertius Ltd; International DOI Federation).

Questions and discussion session 2

  • Mark Bide: Posed the question to the attendees – Is there something with VMF worth carrying on? Nearly 100 percent of attendees were positive over VMF.
  • Mark Bide: Would like views on moving forward.
  • Audience member: Suggest setting VMF up as an open source project. This would deal with many of the issues raised in Norman Paskin's presentation. It would also increase the range of people interested. The work might also continue through funding.
  • Audience member: Expressed doubt over the authority of mappings in an open source project.
  • Norman Paskin: Open source does need to have some control.
  • Audience member: Do see successful schemes operating as open source – cite Bibliographic Ontology. Open source does have many positive aspects.
  • Audience member: Open source projects do not have a formal governance structure.
  • Godfrey Rust: We would like the spirit of open source but part of the value of VMF lies in its authority. How do we get this via open source?
  • Audience member: Referred to potential contacts to be made with Europeana.
  • Audience member: Standardisation may result in governing structure. This gives others who want to use VMF some assurances and continuity.
  • Audience member: There are so many standards. We need a broker of standards.
  • Mark Bide: Should VMF be a membership organisation?
  • Audience member: What comes back as a result of the mapping? This information could be useful for promotion.
  • Godfrey Rust: The VMF could be used to demonstrate the level of use of applications in the short term.
  • Audience member: Consortium of projects and partners striving to form a repository of factors/elements contributing to the semantic web.
  • Audience member: Need to refer to the World Digital Library.
  • Mark Bide: Will need to licence the VMF so that people can use it. We do not have to charge for it.
  • Audience member: VMF is mapping metadata elements not vocabulary and in this way it is misleading. There is no mention of ISAD [e.g. see ISAD(G): General international standard archival description] schemes – these should be incorporated.
  • Godfrey Rust: VMF is for mapping concepts.  Controlled vocabularies exist in metadata. What VMF maps depends on what is presented for mapping. Can map as little or as much as want.
  • Gordon Dunsire: VMF is just one tool of many. Others include SKOS, taxonomies,
  • Audience member: Do you see VMF as taking over the objectives of the HILT Project?
  • Gordon Dunsire: Yes, could be. VMF could be very positive in evolutionary environment.
  • Audience member: Change management for each schema can be met by VMF.
  • Audience member: Need to find a way to normalise standards between electronic publishers.
  • Mark Bide: We are under pressure to deal with such issues raised in this discussion. A new e-ONIX has just been released. Need to maintain compatibility between different versions.  There has been much increased effort made to use data between schemes – semantic enrichment.
  • Gordon Dunsire: VMF allows inferences of classifications from different communities. This has taken off with the digital environment.  As long as the information is in the "silo" it is possible to calculate cost benefits more easily. Once material leaves the "silo" many issues are opened up.
  • Audience member: VMF is attempting to join up the "silos" by interoperability. Is there a danger that the hub could become a giant ‘silo’?
  • Gordon Dunsire: Do not see a single solution. It has to be hybrid. A giant "silo" is a remote possibility.
  • Norman Paskin: Are we engaging with the semantic web community enough?
  • Gordon Dunsire: Do not think we are.
  • Mark Bide concluded by thanking the delegates and the speakers for their contributions to a valuable discussion.

Recorder: Katharine Gryspeerdt (British Library)