The first EUscreen Open Workshop took place in Mykonos City, Greece, on June 23 and 24. Over 60 participants travelled to Mykonos to attend the workshop. On the first day, the programme was made up of presentations about activities and developments within EUscreen whereas on the second day, the programme contained presentations about relevant developments within other European projects. This report contains a summary of all presentations given during the workshop.
The EUscreen content selection strategy and guidelines
The programme started with a welcome by Sonja de Leeuw, project co-ordinator of EUscreen. The first presentation on the EUscreen content selection strategy and guidelines was given by Linda Kaye and Luis Carrasqueiro from the British Universities Film and Video Council. Together with Royal Holloway University of London who is the work package leader for work package (WP) 3, partners active in this WP have been responsible for the development of this policy and guidelines and after extensive consultation with all partners, they have created a handbook for content providers. This handbook was officially presented during the session.
The content selection policy is based on three strands: a selection of 14 historical topics, comparative virtual exhibitions and virtual exhibitions by content providers. The historical topics cover European television history in the broadest sense with topics varying from Arts and Culture and Health to Wars and Conflict, and Society and Social Issues. This presentation included a number of examples of how to use these topics to select content. The comparative virtual exhibitions will provide EUscreen users with a new insight into television in Europe by selecting a number of themes and then using these to compare television practices in different European countries. The virtual exhibitions by content providers offer a platform for those responsible for participating archives to create their own exhibitions with material they want to highlight.
The EUscreen metadata schema
The handbook also contains a description of the EUscreen metadata model. Marco Rendina from Cinecittà Luce presented the structure of the schema which is based on EBUCore and it’s fully interoperable with the latest Europeana Data Model. A part of the metadata will be available in both the original language and in English, like the title and a summary of the content. The thesaurus will be automatically translated. The presentation also contained a number of examples of how to use the schema for different types of digital items, like videos, photos and documents.
The next two presentations provided participants with an update on the portal development. Vassilis Tzouvaras from the National Technical University of Athens showed the back-end tool that will be used by content providers to upload content and metadata into the portal. With the tool, it is possible to import large sets of metadata and map them into the EUscreen metadata schema. The tool will be available for use in August.
Daniel Ockeloen from Noterik and Sanna Marttila from TAIK then gave a presentation about how the front-end is being designed which included a presentation of the wire frames of the proposed front-end. They highlighted the complexity of this design process given the variety of demands from potential end-users and the range of different functionalities they require. The front-end is therefore being designed to be as flexible as possible to support different kinds of use. All users will be able to use all functionalities. Special templates for specific user groups are being designed, showing different functionalities. These templates can also be customised to fulfil individual user needs.
Roadmap, workflow and future challenges
The last presentation on the first day focused on the workflow for the content selection process. Elaine Burrows and Eve-Marie Oesterlen presented a plan for the next two and a half years, including a pilot period,to help content providers to reach their targeted number of digital items to be contributed to the EUscreen repository before the end of the project. The workflow they presented is based on the results of a survey that was distributed amongst content providers.
The day ended with a Q&A session chaired by Sonja de Leeuw. Most questions came from content providers about the workflow, the digitalisation process and the migration of Video Active content into EUscreen. The main thrust of those questions was that it will be a challenge dealing with the differences that exist between current practices prevalent amongst those responsible for the archives.
The second day of the workshop contained two sessions: one on metadata and one on applications.
Metadata schemes in different contexts
The first session was chaired by Marco Rendina and contained four presentations on the development of metadata schemes. The first presentation focused on EBUCore, which is also used as the basis for the EUscreen metadata schema. Jean-Pierre Evain from the European Broadcasting Union showed the evolution of this schema, from 1998 up to 2010. Started as a relatively small initiative, EBUCore is now used by a substantial number of broadcasters and European projects. The schema is being developed by an expert community on metadata. This community provides expertise on the representation and generation of data, indexing and retrieval and workflows and the lifecycle of data. The expectation is that the current use of EBUCore is only the tip of the iceberg and that the use of this schema will continue to increase in the future.
Franca Debole from ISTI-CNR and Julia Welter from the Deutsches Film Institute showed the metadata schema that is used in the European Film Gateway (EFG). This schema, based on the FRBR model enables the representation and the connection of different versions of the same cinematographic work, since different film archives may own the same work in a different version or parts of the same work. This metadata schema is also used to link other sources to the content, like the authority files of names, institutions and events. In contrast to EUscreen, the EFG doesn’t have a content selection policy, because of a lack of available digital content and complex IPR issues.
Tobias Bürger from Salzburg Research gave an overview of the activities of the Media Annotation Working Group of W3C. The goal of this working group is to develop the means to create interoperability between metadata schemes for different kinds of content, like video, audio and images. This working group uses two approaches to support interoperability: through semantic web language and through the use of pivot upper ontology.
The last presentation in this session was given by Carlo Meghini who is a member of the European Data Model (EDM) working group. He described some of the challenges the working group is facing. Rich functionalities, like those in Europeana, require rich representations. These rich representations are already available inside the separate archives but combining these representations creates a problem with interoperability. The EDM working group aims to solve this problem by applying a strategy based on four pillars: collecting the various metadata schemes, connecting and mapping them, providing a standard based on these mapping activities and linking the various collections to each other.
Applications for the automatic extraction of metadata
The second session highlighted a number of approaches to the automatic extraction of metadata. The session was chaired by Vassilis Tzouvaras.
Natasa Sofou from the National Technical University of Athens showed the latest results in a project called WeKnowIT, which can automatically link images on Flickr to each other, based on visual similarity. With this tool, new photos can be automatically localised. Experiments will also be carried out with EUscreen material to see how this technique works on videos.
The presentation by Guus Schreiber from the Free University of Amsterdam about the use of linked data in the project NoTube was planned to be given via Skype. Unfortunately, due to problems with the Internet connection, this presentation couldn’t take place.
Another method to extract metadata is automatic concept detection. Marco Bertini presented the results of this technique that is used in two projects: VIDI-Video and IM3I. Within VIDI-Video, 1000 audio and video concepts were trained and used to extract metadata. This kind of extraction facilitates fine-grained searching through video on a shot level. This technique delivers high performance but currently takes too long to process content in daily archival practice. The IM3I project is trying to reduce processing time by using a very limited set of concepts, creating less fine-grained metadata. Implicitly, this presentation showed the current existing dilemma for archivists in the extraction of metadata, notably time versus quality.
The two case studies from PrestoPRIME showed a user-centered approach to the extraction of metadata. Johan Oomen from the Netherlands Institute for Sound and Vision gave a presentation about Waisda?, a social tagging game with Dutch television content, where users tag things they see and hear. When two players add the same tag at the same time, points are earned. The tags were also evaluated by archivists from Sound and Vision. Results from this evaluation showed that tags added to factual programmes like documentaries, proved to be more valuable than tags added to reality shows, like a dating show for Dutch farmers. Günther Mühlberger from the University of Innsbruck Library presented a case study on the digitalisation of VHS tapes and the extraction of metadata. This was presented in the context of a scenario his team is developing for the preservation of material from universities. This scenario was tested during a pilot project involving the Institute of Slavonic Languages and aimed at digitizing a large set of tapes, as well as extracting metadata for higher education purposes.
Future EUscreen events
This workshop was the second of nine workshops which will be organised during the course of the project. Every workshop will focus on one or two aspects of the EUscreen project. This one focused on metadata schemes and content selection policies. The next EUscreen event will be the 1st EUscreen International Conference on Content Selection Policies and Contextualisation in Rome on October 7-8 2010.
The videos of the presentations will be available soon.