Search Disabled |
|
|
|
Search Options | Help | Site Map | Cultivate Web Site | ||||
|
|
|||||
| Home | Current Issue | Index of Back Issues |
| Issue 8 Home | Editorial | Features | Regular Columns | News & Events | Misc. | ||
By Marieke Guy - November 2002
Marieke Guy reports on a symposium seeking to bring together key organisations in the field and discuss the state of the art in digital imaging.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
On a hot Thursday morning in late June I travelled down to London for the JISC International Digital Image Symposium. The presentations were all held at One Great George Street, a very stylish building in the heart of Westminster, and handily right next to St James Park. The symposium itself was organised on behalf of the JISC's Images Working Group and was an endeavour to bring together key organisations in the field and discuss the state of the art in digital imaging.
After registration and coffee Jane Williams, who at the time was the Chair of the Images Working Group and the head of JISC Assist (she has since left), started the ball rolling with a brief introduction to the JISC and their mission to fund projects that would allow seamless access to resources at any time.
The first session of the day covered delivery and digital asset management and was chaired by Catherine Grout, Programme Director for the JISC Information Environment. In her opening Catherine looked back at the JISC's changing approach to digital images. The JISC started thinking about images over 5 years ago and Catherine herself is responsible for the strategy for distributed image services. To encourage debate and activity the JISC Images Working Group was set up and currently deals with an image population of 500,000+. Although at present the preponderance seems to be in humanities images Catherine feels the future will be in images for medicine. To end her presentation Catherine highlighted some of the main digitised image collections (JISC Image Digitisation Initiative (JIDI) [1], The Visual Arts Data Service (VADS) [2], licensed collections (Scottish Cultural Resources Access Network (SCRAN) [3] and The Art Museum Image Consortium (AMICO) [4]. She also mentioned brokered-access activities supported by the JISC.
Oya Rieger, manager of the Library Office of Distributed Learning at Cornell (the 10th largest research library in the world) followed with a presentation on imaging to asset management. In her role at Cornell she deals with over 3 million images. She looked at changes in user expectations, information landscape and to distributed learning.
Following Oya, Paulo Buonora of the Italian State Archives, Rome, then gave an introduction to various digitisation projects with which he had been involved, such as Imago II. Paulo was very open about the successes and failures experienced and raised contentious issues such as whether the technology used to view images would be obsolete too soon, and admitted that at times he wondered if they would have been better off using microfiche instead. He also acknowledged that at the start of the project, (over 5 years ago now), they had not followed any rules or standards and had actually destroyed the original TIFF images; a mistake that would take a lot more justifying today.
Debbie Campbell, from the National Library of Australia, then discussed delivery and digital asset management in the light of her main project, Picture Australia [5]. The presentation covered a number of contentious issues such as the need for persistent identifiers. Picture Australia currently does this with the hope of migrating to an international solution when it arrives.
The last speaker before the panel session was Simon Grant of the Tate Gallery [6], London. He is Head of Information and Systems for the Tate's four sites. Simon spoke about current work at the Tate, which he called the Tate's insight initiative, an attempt at creating collections content and digitising the entire Tate collection of 55,000 works. The project has cost around £1 million. The Tate are also looking at developing their image management system (IMS) and using it as a model. They have recently worked on an indexing schema based on Collage and also on moving to an asset management and content management system. One other area that the Tate is very interested in is the picking up of customer information through personalisation.
The panel discussion touched on a number of the areas that had been raised by the speakers. An interesting debate arose about whether projects should see their digitisation as final or as something only for that time and that they would be going back to redo things in time. The phrase used was quick and dirty versus standards and quality. Debbie Campbell felt that you need to put digitisation into context and that sometimes it is important to include images that are not that great but have important historical value. If you are going for a 'quick fix' you should try to automate the process of redoing. Oya Rieger said that it was not Cornell's intentional to go back and redo work but a project can only do its best at that moment. The most important thing is to get images out there and get audiences interested.
There was also a brief discussion on Geographical Information Systems (GIS) and whether more organisations are intending to add referencing to their images. It is now possible to use postcodes to link images together. One delegate asked is this what users really want? Another explained that archaeology projects are moving ahead in this area and the Digimap Project [7], European Council of Information Associations (ECIA) [8], and the Alexander project are worth looking at if the area interests you.
After lunch we reassembled in the presentation room for an afternoon of metadata and technology. The chair for the first session was Seamus Ross, Director of Humanities Computing and Information Management at the University of Glasgow.
Howard Besser, an associate professor at Berkeley, began the session as the keynote speaker. He gave a whirlwind tour of important recent activities in the world of image metadata. His talk covered the key problems we are facing: namely discovery, longevity and interoperability, and a hundred and one other areas of interest. Howard explained what an important role standards have in helping with the key areas as well as veracity, provenance and recording in a consistent manner. He followed on from the previous panel session by asking how you avoid creating everything again over time. Although he could offer no easy answer metadata obviously has an important role to play. Howard also began a brief discussion of the different types of metadata scheme currently used including MARC, Dublin Core (DC) and the making of America (MOA II) leading on to Metadata Encoding and Transmission Standard (METS) [9] and slipped in tuppence worth about Computer Based Imaging retrieval (CBIR).
The following speakers, who talked on different aspects of metadata, were intending to cover material from the general to the specific.
Merillee Proffitt of the RLG Cultural Materials Initiative [10] talked about RLG Cultural materials and metadata standards. She introduced RLG 101, a non-profit corporation based in the U.S. that holds world-class collections of cultural materials. Some of the challenges that RLG have faced include diverse descriptive practices, meaningful integration across projects, digital representation of materials (surrogates), reliable distributed infrastructure, multiple audiences, international rights and responsibilities. A lot of these problems have been solved using METS, a metadata scheme which is useful for structural (sequential) metadata.
The next speaker was Michael Day from UKOLN. He talked about a metadata review carried out for the Focused Images for Learning and Teaching (FILTER) project, which had looked at selected metadata initiatives relating to images. The initiatives had been divided into a number of sections: Cultural heritage (AMICO, SCRAN, SPECTRUM [11], Visual Resources Association (VRA) [12]), digitisation (National Information Standards Organization (NISO) [13]), geospatial (Federal Geographic Data Committee (FGDC) [14]) and educational (IEEE Learning Technology Standardisation Committee (LTSC), Learning Objects Metadata (LOM) [15]). Michael explained that the main conclusions drawn from the review were format diversity which is really a healthy sign because metadata schemes need to be fit for purpose and images tend to be highly heterogeneous.
Paul Shabajee, a research fellow at the Institute for Learning and Research Technology (ILRT) [16] working on the ARKive Project [17] then talked about metadata, resource discovery and repurposing. His main point was that although metadata is fundamental to resource discovery in large or distributed data sets it is not enough by itself. He suggested various scenarios in which the use of raw metadata would not help users to find an image. The issues covered are fundamental to any type of cross-disciplinary repurposing and cross- searching. Paul found that there is no easy answer; he himself is looking to shared ontologies and the semantic web.
The panel session discussed Paul's various scenarios when metadata would not be enough. Howard felt that they all were answered in different ways. For example in one scenario a user tries to find all instances of birds that fly; because flying is taken for granted with birds this metadata would not have been stored. One way of approaching the problem would be in the design of the user interface. The panel also discussed the problems of interpreting metadata schemas and the fact that ultimately metadata does not matter to certain people, the consumer does not care.
After a quick coffee break we began the technical/blue skies session chaired by David Dawson of Resource.
Ed Bremner from Technical Advisory Service for Images (TASI) [18] began the session by talking about next generation image file types. He confirmed our feelings that technology has moved on (for example with music formats we have moved from wax to 78 to LP to CD to minidisk and DVD) but that image standards are actually less of a problem than metadata standards. He asked us to consider how 'standard' our current image standards actually are and concluded that they are not very standard; (for example with the TIFF format are we talking about baseline or extended?) So with this in mind do we need new file types for digital images? Ed explained that arguably no we did not, because users do not care and the current files are satisfactory. The current archiving standard is fairly satisfactory as a TIFF but it would be useful to create a delivery standard. This standard should really be a quality open standard with general support, improved compression, contain secure metadata tagging based on xml, have international rights management, be zoomable and streamable, and have multi-layer error resilience. Ed's suggested answer to all these criteria is JPEG 2000.
Next Paul Child, Project Manager of Artworld [19], presented on the use of scaleable vector graphics (SVG). The Artworld project is storing learning and teaching resources in XML and the object catalogue will deliver objects in SVG. SVG is a non-proprietary language for describing two-dimensional graphics in XML and covers 3 types of objects: vector graphics shapes, images and text. His presentation gave some of the background to its use and looked at Cocoon an XML publishing framework.
The subsequent speaker was Rob Mildren, Project manager for Scottish Archive Network (SCAN) [20] (not to be confused with SCRAN!), a heritage lottery-funded project that digitises wills and testaments in Scotland 1500 - 1900. Rob talked about the implementation of a quality control system recently set up by SCAN where all QC is carried out within 24 hours of digitisation, 30% of images are checked, and retakes are completed within 24 hours. Quality control metadata is also recorded and the results are documented and retained. These processes are all highly automated. SCAN's plans for the future include more content and performance improvements and reducing the time of digitisation. They also intend to start using JPEG 2000.
Shigeaki Iwai, a digitisation consultant for the Higher Education Digitisation Service (HEDS) [21] reconsidered the digital imaging methods used for library and museum materials. He looked at the imaging of rare materials, different types of scanners, different types of digital camera and the related issues. The most important of these was the safe preservation of originals, consideration of light and heat, imaging speed, image quality and economics.
The panel discussion considered how digital imaging projects can move forward with the technology they use. After the close of the first day of the symposium a gathering with buffet and drinks was held at the Café Royal, near Piccadilly Circus.
For the first session of the second day a considerably less jovial group of delegates congregated in the auditorium: England had lost to Brazil in the world cup that morning. However spirits were soon lifted by some hearty discussion into what exactly user needs are and how we can carry out evaluation.
The session was chaired by Jane Williams and the first speaker was Max Marmor, Director of Collection Development for ArtSTOR [22]. Max gave an introduction to ArtSTOR, which is a user-centred digital image library that is academic- and museum-based. It is related to JSTOR which concentrates on journal storage.
The second presentation was given by Seamus Ross, Director of Humanities Computing and Information Management (HATII) who had chaired the previous day's metadata session. Seamus talked about user needs analysis to impact assessment and started off by asking the question why evaluate? Some answers he gave were to establish what is being achieved, to justify investment, enhance current position, consider interface design, inform future developments and measure project impact (e.g. on teaching, research, lifelong learning). Seamus explained that evaluation is not done in the abstract and that we need a set of criteria established well before the start of a project. Evaluation depends on the scope of project, aims, time and resources. Seamus also discussed methods of evaluation, which include observation and focus group discussion. He suggested we use multiple methods.
Margaret Graham, Principal Lecturer at the University of Northumbria, then gave a talk on images and users which looked at a small user survey funded by the Chartered Institute of Library and Information Professionals (CILIP) [23]. The survey's purpose was to discover the information-seeking behaviour of image users. Margaret explained that there are many diverse image collections in the UK and that describing images is a difficult task primarily due to a professional skills shortage. It seems that digitisation alone does not make image collections easier to manage. In the Visual Information Seeking Orientated Research (VISOR) study, 4 types of search were identified: specific, semi-specific, conceptual/abstract and vague. Images were selected on a number of different criteria including saliency, quality, recency and accuracy. The main conclusions arrived upon by the survey were that users need further training in the use of tools, that funding was needed to allow cataloguing to be carried out alongside digitisation and that there needs to be more investigation into CBIR (Content-Based Image Retrieval).
The final speaker for the session was Susan Haigh from the National Library of Canada who talked about evaluating user satisfaction. The National Library of Canada has been working on Images Canada [24], a site similar in scope to Picture Australia. Susan discussed some of the assumptions they had made about their users at the start of the project. Many had been very wrong. For example they had assumed that users would understand the relationship between searching and metadata. In fact users had very little comprehension of what metadata was. Users were not interested in multiple views of the same object and did not appreciate the concept of a gateway. They did not want to view too many images and 6 thumbnails per page were enough. There was also some discussion of bilingual issues.
In the panel discussion it was suggested that evaluation is important but questions were asked about how far we should go on implementing something based on that discovery. The reasons for this are many but centre on money and the fact that often the user is wrong (but you have to listen to them anyway). One conclusion drawn from the presentations was that it is important to share lessons but current models put us under pressure to be successful. Delegates felt that sharing mistakes is just as useful, if not more so, than sharing successes.
In the final stretch before lunch Bernard Smith Director General for the Information Society [25], European Commission, gave a keynote speech on trying to make Europe into a knowledge state. Bernard explained that cultural sectors need to define quality standards as well as strive for preservation and access. If quality was not defined soon by institutions then the Brussels quality framework would be enforced. Bernard also introduced Framework 6 (FP6), which begins a move away from project thinking to initiative thinking. For FP6 there have been 15,000 expressions of interest so far, proposed projects are strong on preservation and there is a lot of interesting material on digital imaging, but, so far, little on digital libraries.
After lunch Debbie Campbell chaired the final session of the symposium on the use of images in learning and teaching.
Jill Evans, project leader of Bristol Biomed [26], kindly stepped in for Catherine Grout as the keynote speaker and gave an interesting presentation on the issue of sustainability for short-term projects. Her current project Biobank was a JISCfair call and is looking at depositing images and finding a more sustainable model for Biomed. Part of the reason for the Biobank bid was that Jill had begun to worry about her project becoming obsolete and full of out of date resources with continuation funding being hard to find. Most digital image projects are aware that quality is costly and that over time they will have problems filling in the gaps in their collections. Jill feels that the traditional approach is not working. One option that Biobank has come up with is looking for internal solutions rather than external solutions. Biobank could embed the resource in the community itself and get the community to take ownership. Users themselves will then add metadata and images. Naturally this has quality implications and the way round this will be to have distributed teams of 'trusted evaluators'. Jill explained that this approach would be a trade-off between quality and accessibility and does raise a number of pertinent issues: which is the most important: a timely resource or an impeccable resource that is out of date? Can any system ever be self-sustaining? And should quality ever be compromised? Jill suggested that if a resource contains images that are mediocre or of poor quality it is not a huge issue to users. Users would say that having an image is the most important thing, above quality, especially for learning and teaching. Image resources should be community peer-reviewed against agreed standards and criteria.
Phill Purdy, manager at VADS then gave a presentation on new developments in the Promoting Image Collections for Learning and Teaching in the Visual Arts (PICTIVA) project. He demonstrated a new lightbox facility, which provides multiple personal image sets for users to retain, annotate, distribute and manage. PICTIVA is also implementing user login/registration and trying out CBIR to allow users to find similar images against the parameters of colour, contrast and shape.
Elaine Mowat, Tertiary Education Officer for SCRAN then provided lots of interesting ways that the resources at SCRAN had been use in learning activities. SCRAN materials are copyright cleared for educational use and you can technically do anything you like to them. Elaine explained that SCRAN's approach to learning is to encourage emotional engagement. Their resources have been used in this way in interactive handouts, online tutorials, virtual field trips stories and case studies, group discussions and brainstorming, as images of the week, and in paper-based materials.
The final speaker of the symposium was Charles Merullo of Hulton Getty. The Hulton Archive [27] is one of the largest collections of photography and illustrative material in the world. It contains over 40 million images including prints, engravings, cartoons, illustrations, maps, periodicals and other ephemera. Charles talked us through some of the fascinating items in the collection and suggested ways in which they could be used in learning and teaching.
The panel discussion covered how we can get teachers to teach with digital images. It was felt that we need a robust set of images that relate to courses along with sets of tools and resources and help for teachers themselves.
At the end of day two it was agreed that the Digital Image Symposium had been a huge success and covered many of the key issues that arise for projects creating digital images resources today. Although many areas (such as quality versus quick and dirty, the benefits of CBIR, the use of evaluation and what metadata schema should be used) will continue to be discussed time and time again, conferences such at these bring the fundamental questions to the foreground. After all…"Good questions outrank easy answers."
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Marieke Guy
QA Focus/NOF-digitise Advisor
UKOLN
University of Bath
Bath, BA2 7AY
United Kingdom
m.napier@ukoln.ac.uk
<http://www.ukoln.ac.uk>
Marieke Guy is QA Focus/NOF-digitise Advisor, within the Policy and Advice Team. She is currently on maternity leave.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
For citation purposes:
Guy, M. "JISC International Digital Image Symposium, 20-21 June 2002", Cultivate Interactive, issue
8, 15 November 2002
URL: <http://www.cultivate-int.org/issue8/digimsym/>
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Related articles:
If you would like to view similar articles to this one click on a key word below:
< - digital preservation - digital imaging - digital asset management - digitisation - CBIR - GIS - METS - metadata - image formats - >
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
|
|