JISC

The Community Led Image Collections report made recommendations for image collections to enable sharing of resources and metadata within the community. On the basis of the recommendations, this study will focus on engineering education.

The Community Led Image Collections (CLiC) report  made recommendations for image collections to enable sharing of resources and metadata within the community. The main recommendations were:

• A simple directory of community image collections be established on a subject basis. There are potential roles for HEA subject centres and for Intute and others
• Metadata about images be shared. RSS was seen as likely to be more practical than OAI-PMH at the moment. 70% of respondents to the CLiC study said they’d be happy to share metadata as a means of driving traffic to their site
• Images be shared

On the basis of these recommendations, this study will focus on how applicable to engineering education these suggestions are specifically by:

• observing attitudes to sharing image resources within the engineering community
• examining potential barriers and enablers to sharing
• investigating the potential for a visual directory of image collections
• encouraging image collections to provide an RSS image feed for demonstration purposes
• presenting the issues arising from the study and making suggestions for future practice

The methodology for gathering the information presented in this study was, in summary, to:

• Commission reviews of selected Image collections relevant to Engineering in order to reappraise whether these collections are of interest to engineering educators
• Create a prototype image RSS-aggregator as a proof of concept and demonstrator of a visual directory of relevant image collections.
• Engage with image collection owners to try to encourage them to implement RSS feeds to share image metadata
• Evaluate the experience of this engagement to look at legal, technical and organizational factors

Summary and Discussion

The prototype visual directory was shown to the maintainers of the collections, who all seemed to appreciate the purpose of the directory and agreed that it may drive traffic to their image collection should they provide an RSS feed. If this directory was placed on an Engineering portal such as the HE Academy Engineering Subject Centre website it would give Engineers an overview of what types of images were available. This however would only be successful if enough feeds were to be aggregated. We were unsuccessful in encouraging the production of any RSS feed, during the timescale of this study.

There is no technical problem with this approach of creating a visual directory based on aggregating RSS feeds that have images as enclosures. However, we did not find that the widespread use of RSS aggregators for news stories or music was of as great a benefit as might be hoped. While RSS aggregation software is widely available, it tends not to be ideally suited to building an image directory. The issues we encountered when using generic RSS aggregation software to create an image directory included the length of time the directory took to read-in live feeds and how the enclosures were displayed (i.e. hidden behind a "play" icon). We do not doubt that such issues could be overcome. Furthermore, we do not think that any technical approach that relies on implementation by the image provider could involve significantly less effort on the part of the image provider (or for that matter, on the part of the directory provider).

The functionality that could be offered by a visual directory based on static feeds is limited, however we see a clear line of development that could allow a more dynamic service where, for example, the images displayed for each collection are selected to match a query entered by the user[1]. However, given their failure to produce even static RSS feeds we have doubts about the wisdom of going back to any of these image collections with additional requirements once they had created a feed, and therefore we have doubts about the feasibility of incrementally developing a more dynamic service.

While there was some evidence that legal issues were perceived as a barrier to building image collections relevant to engineering education, and while any legal concerns on the part of the end users of these images are beyond the scope of this work, we found no evidence of legal problems inhibiting this approach to syndicating metadata and images.

There is evidence from both the reviewers of the collections and the collection providers of a cultural willingness to share images and recognition of the benefits of sharing. There is some mismatch between what is available for sharing and what is desired, but (while clearly of utmost importance in the general scheme) that is incidental to the specific focus of this work. Sharing of metadata and thumbnails through RSS was well received with representatives of four of the nine collections being that they would be willing in theory to provide an RSS feed for their collection.

We have come to the conclusion that the failure to elicit any RSS feed is due to organisational, that is to say cultural, barriers to technical development. We think the root of this is in the fragmentation of expertise and responsibility and the lack of engagement with machine to machine interoperability. The individuals with an interest in sharing and using images for engineering education largely have an attitude that if they put an image onto the web then it is available, an others can use it as they please. While they are open other suggestion for how images might be shared via machine to machine interoperability they do not have the technical expertise to appreciate the full implications of such approaches or for that matter to implement them. The technical staff with the necessary expertise to do this seem to be in short supply.

There are striking parallels between our experience here and those of the PerX project based at ICBL, which developed a pilot service that provides subject resource discovery across a series of repositories of interest to the engineering learning and research communities by cross-searching repositories using OAI-PMH, Z39.50 and non-standard protocols. This pilot was used as a test-bed to explore the practical issues that would be encountered when considering the possibility of full scale subject resource discovery services.

The PerX project encountered many challenges to sharing metadata, including communication with the data provider. They were often unable to communicate with the individual responsible for providing the data, which mirrors this case study in that four of the image collections provided no response whatsoever to a request for information. They also found that where communication was established with repository owners and a willingness to become a data provider was expressed there was no obvious progress towards implementing the necessary technical interoperability. PerX was successful in gathering metadata from many repositories, often where an existing mechanism existed at the repository but it found that 'while a minority of OAI-PMH services are professionally set up and maintained, many others are relatively immature, inadequately tested and are not well supported' [2]. Furthermore, successfully harvesting metadata on a single occasion did not, in the experience of the PerX project, provide any indication that routine automatic harvesting would be possible in the future.

In our view these findings from PerX act as warning for what can happen if a technical interoperability solution is attempted without the necessary organisational commitment on the part of the data provider: Paul Miller referred to this as 'political or human interoperability' [3]. An element of long term commitment is required to maintain interoperability, which, on the evidence of those image collections which became closed to subscribers only or were put offline by hackers, was not always present in the image collections in this study. If a feed were to be provided, it would require the service provider (aggregator host) to test the feed to ensure it was compatible with the aggregator. There would also have to be regular checks to ensure that when a feed was updated that it remained compatible.

A way forward?

The main issues arising in this case study were cultural as opposed to technical or legal issues. The collections indicated that they were willing to share metadata and images, however a lack of understand about what this requires technically is apparent. Most of the collections shared images by placing them on their own website, which means sharing is only possible if the user happens to find the website and if they can ascertain from the terms and conditions whether they can legally use the image. The collections currently have no provisions for machine to machine interfaces such as OAI-PMH or RSS feeds. These m2m interfaces could drive more traffic to their website from Engineering portals such as the Engineering Subject Centre, Intute or a cross search service like PerX. By increasing traffic to each collection the potential for sharing would be raised.

We believe that the concept of a visual directory has potential, however this potential needs to be demonstrated without the image collection owners being required to undertake technical work in creating an RSS feed. This means working with what is already available and using web-crawling and/or screen scraping technology to harvest selected images (in much the same way that Google Image Search acquires data). We speculate that a successful image directory will hopefully be something that new and developing projects will want to be listed in, and given time they may provide RSS feeds, hopefully to the requirements of the image directory. At the same time, advocacy work and policy directives should be used to ensure that new and developing image collections provide suitable interoperability and recognise the benefits of maintaining them.