28 August 2013
Report from the Semantic Web SIG session in Singapore
This was the 3rd open session of the Semantic Web SIG (SWSIG). The session started with a short introduction by Emmanuelle Bermès, convenor of the group.
The premises of a Semantic Web related group in IFLA emerged within the IT section in Milan (2009) after a very successful satellite meeting on “Emerging trends and technology: libraries between Web 2.0, semantic web and search technology”. The section then proposed to other interested groups within IFLA, including the Cataloguing section and Classification and Indexing section, to organise a joint session on “Libraries and the Semantic Web” in Göteborg (2010). This session was a great success and led to the idea of creating a Special Interest Group.
The SWSIG first gathered during an open session in Puerto Rico (2011). This was a requirement for the SIG to be officially approved by IFLA, which was the case by the end of year 2011. More than 100 people attended the meeting. In Helsinki (2012), the IT section who is hosting the SWSIG officially appointed Emmanuelle Bermès as the convenor and a second meeting was held, attended by about 70 people.
Both meetings in Puerto Rico and Helsinki took about the same form: a series of short reports by people attending the Congress, about their realisations in their libraries in the field of Semantic Web. Such realisations were still quite scarce in 2011, but in 2012, the SWSIG demonstrated that they were a lot of ongoing initiatives. These meetings took the form of a series of lightning talks, outcome of which is available from the SWSIG's Web page on the IFLA website.
Finally, the IT section organised a satellite meeting in Singapore prior to this year's congress, with the focus: “User interaction based on Library Linked Data”. The event was well attended and produced interesting talks and ideas.
Feedback from the SWSIG open session meeting in Helsinki was that the people in the room actually needed more discussion time in order to make the most of the experiments that were presented. People wanted to be able to ask their own questions and obtain answers to their own needs, rather than hearing a pre-conceived presentation.
Based on this feedback, the convenor proposed that this year's open session should be a discussion session.
The attendants were then split into 4 thematic groups, each group being animated by a member of the SWSIG who did attend the previous years. The 4 topics were:
- standards and mappings
- linking and cooperation
- using and reusing Linked Data
- tools and recipes.
An additional group was created with the topic “Linked Data for beginners”.
Each group was given a set of questions and had 40 minutes to discuss the topic. Then each group named a person who reported on the discussion.
It was announced that the outcomes of these discussions would be used as input to shape the content of the satellite meeting that the SWSIG is organising next year in Paris prior to the Lyon Congress.
The group started with a discussion on which vocabularies libraries are currently using when publishing Linked Data. It appears that even if they tend to use typical library vocabularies such as Dublin Core, RDA... different libraries (for instance The British Library and Bibliothèque nationale de France) have chosen different combinations of vocabularies. So maybe there is no perfect way to go and it is a good thing to publish a full dataset in whatever format. So, the group stated that it is important to focus on creating good mappings between vocabularies so that people can reuse the data.
The group was encouraged to discuss the standards that IFLA is developing in this area. FRBR, ISBD, UNIMARC... most IFLA standards are now available as Linked Data ; everybody in the group was actually aware of this fact and thinks it's a useful task. However, these standards are not really used at the moment, so the group discussed why and how to improve the situation. The group suggested the development of mappings and guidelines for usage, and advocating their use in different ways.
Linking and cooperation
The list of questions handed to the group encouraged them to discuss both linking and cooperation, but in the timeframe available they mainly discussed linking.
The linking debate revolves around what to do with respect to Wikipedia: to link or not to link to Wikipedia. Positive examples of institutions having chosen to do it include BnF with their Opencat prototype and the BBC.
One of the challenges of Linked Data is how to make our OPACs more attractive in a trusted way. It is necessary to get the users into our environment. Libraries could use their collection as a focus to attract users, as well as fulfil their responsibility of sharing their collections with the world.
The group had a sense that this kind of development could be a task for subject librarians, who could help users identify resources of interest on the web in their domain area. The limitation here seems to be lying in subject librarians not feeling safe enough yet in that environment to carry on this type of activity.
Using and reusing
The group hesitated between focussing on the reuse of library metadata more effectively in the short term, or rather to embrace the subject space and the replacement of subject classification. Finally they decided to focus on authority data which seems to be at the core of the subject.
The group then discussed the difference between Linked Data and Open Data, stating that Linked Data is a technology and Open Data is a policy, so it makes sense to combine the two. It is the group's belief that maximum reuse comes from assuaging an open license.
Making authority files available as Linked Data under an Open License should bring the maximum benefit from libraries to the community.
The group also shortly discussed provenance, disclaimers, and different types of links.
Recipes and tools
This group also discussed the relationship between Linked and Open Data, coming to the same conclusion as the previous one.
There are existing registries for Linked Data, existing datasets that can be reused. Web tools can be used to manipulate the data. The SameAs.org service was mentioned. Several vocabularies are available including Dublin Core, FRBR and Bibframe, all providing a good range of reusable metadata elements.
The group thinks it is important that librarians are aware of RDF, SPARQL, OWL in general (at least certain profiles of systems librarians), as the Semantic Web is about building complex domain models. Such skills will provide ways to discuss Linked Data as a common ground with technicians and thus offer great opportunities.
Tools in this environment are expected to be as transparent and open as other Web tools we have (allow viewing the source code for instance). SPARQL allows using library knowledge to ask very complex questions. The notion of records will evolve to named graphs.
LD for beginners
This group was intended as a place to give an introduction to Semantic Web and Linked Data technology. It was the most attended group in this session.
Discussion started with a question asked to the participants: what was their interest in Linked Data ? Some were executives who were looking for a high-level explanation of the strategic incentive for Linked Data. Other had heard about it or read about it but didn't understand or didn't have time to investigate. They were curious about actual realisations that would help them understand the benefit of Linked Data for libraries.
A short definition of the Semantic Web was given and the benefits to libraries were discussed: this technology helps make them part of the web; make the data more visible outside the library domain. More technical questions were asked about how to transform MARC data or link to other resources. Different types of libraries are involved (national libraries and others).
The group concluded on a necessity for tutorials on that matter within IFLA (during the satellite in 2014): such tutorials should be targeted according to the expected audience (library managers or executives, cataloguers or metadata librarians.)
As a conclusion, this was a successful session. The participants declared they were interested in the discussion topics. It appears that many people attended just out of curiosity and the SWSIG should address the need for basic information about Semantic Web and Linked Data, targeted to a non-expert audience.
Based on this report, topics of interest for next year may include:
- use cases, best practices and recipes for publishing library data, especially authority data, on the web using Linked Data
- mapping vocabularies in order to facilitate the reuse of library data
- recommendations and experiments to make library online services more attractive in a trusted way by relying on Linked Data
- Linked Data, Open Data and how they can maximise the benefit of library data for the community at large
- demonstrating the benefits of library Linked Data for end users
- tackling the challenge of creating links with other datasets, within or beyond the library domain
- tutorials for a targeted audience (strategic - technical basic - technical advanced)