Resolution to persistence.

Byline: Tracey Caldwell

Resolution to persistence

It's a minefield out there for information professionals trying to keep up with linking and cross-reference technologies on the internet. Tracey Caldwell takes you through some of the key issues surrounding link standards and initiatives like OpenURL, DOIs and CrossRef

A large part of an academic librarian's job is providing library users with access to the scholarly texts that the university has paid for. Many hours work goes into maintaining links to electronic texts at content providers, publishers and agencies.

Link resolvers have smoothed part of this process, but still depend on the maintenance of a database of links. Also, link resolvers rely on the librarian to let all their content providers know if and when there is a change to the library URL a following a security update, for example.

Contractual and access issues, among others, have slowed progress to the shared information environment nirvana, but there are encouraging advances in technologies and standards that mean librarians should increasingly be able to link their users to highly relevant resources more easily.

Linking technology works both externally in accessing resources, and inbound, so librarians need to get involved in making their metadata a query target. Links can be dynamic -- made in real-time to metadata-tagged targets -- or they may be made to a database of links, as in the DOI model (see Jargon Buster).

As libraries become involved in setting up repositories for open access to research papers, link standards such as OpenURL can step in and act as a persistent identifier. The ability to point reliably to different versions of copy on the web helps the process of reviewing and distributing work enormously.

Linking technologies are the natural partner of search facilities. They provide another layer of intelligence available to librarians trying to access the most relevant resource, especially where many copies of the same resource exist in different locations.

The OpenURL standard was first developed as a tool for the academic community, to help librarians locate the most appropriate resource, taking account of their access privileges. Context remains a key aspect of a new version of OpenURL that has recently been released. This will target new types of content, such as patents and dissertations, as well as being able to handle multiple author names.

Open URLs, link resolvers and Edina

An OpenURL looks like a long web address, and has two parts to it. The BaseURL indicates which URL resolver the user is linking with, while the rest of the URL contains metadata, or metadata access, for the text for which the OpenURL is provided. The OpenURL targets the library's OpenURL resolver, which then offers context-sensitive services based on that metadata.

A resource is identified within the OpenURL by its assigned Digital Object Identifier (DOI), or by encoded metadata -- for example, title, author, journal title, location -- or by a combination of both.

A major drawback of a locally-held commercial link resolver is that it is not much use until a librarian has spent time configuring it, so that the database at its heart has all the necessary access information. Tim Stickland, software engineer at Edina, estimates this can amount to four months' work in setting up the systems and 20% of a librarian's time maintaining it.

Edina, based at Edinburgh University Data Library, is a JISC-funded national data centre. It wanted to use OpenURL to link to resources and began to build a database of institutional URL resolvers. The idea of being a national URL resolver routing service arose and Edina decided to create a shared infrastructure that everyone can tap into.

Previously, UK universities had to tell all their content providers what resolver they were using and update them if there was any change to this, or to any other part of the resolver URL. Edina is acting as a middleman, collecting each university's resolver information, updating it and making it available to all the content providers. That way, if there is a change, the university only needs to notify Edina, not each content provider.

Stickland says Edina tried to interest content providers in the router service, to provide better access to content, but it was not successful. Only IOP Ingenta saw the value of the router. "Had the content providers decided this was a good idea we would now be in a position where you could search around easily. It would be fantastic as universities would not be spending thousands on URL resolvers," he says.

Part of the problem could be that the waters are being muddied as traditional content providers have moved into providing link resolvers. Edina and JISC are aware that their low-cost initiatives may, at first sight, present a threat to the bottom line of commercial enterprises. However, Stickland says: "We are just making linking better so that everyone will spend money on it. We think it will enlarge the market."

Ovid LinkSolver and OCLC OORR

Ovid is one content provider that considers its linking technology a core competency. All its hosted content is bundled with linking facilities backed by its LinkSolver OpenURL link resolver. Ovid has recently extended its linking services with Database Link Packages, a service that provides its bibliographic database customers with automatic links to more than 700 Open Access journals, and optional links to valuable internet resources from relevant bibliographic databases.

One-stop links to paid-for and open access resources is the ideal. Stickland points out that one of the goals of the Router is to allow OpenURL links to be added to free access services, where it would not otherwise be possible. Such services do not identify users or their institutions, and so cannot find the local resolver that they should link to.

Ideally, the Edina initiative would be expanded internationally, but access and identification issues seem to preclude this for the time being. The UK is relatively unusual in having the UK-wide Athens system of verification. Other countries have a patchwork of systems. It would be a huge task to incorporate all the access systems into an international database.

In an international initiative to enhance linking, the OCLC OpenURL Resolver Registry (OORR) provides a global database of OpenURL Resolver links and metadata, which can be made available to a range of referers -- all the people who are creating target OpenURL links for their resources whether these are closed or open, free or fee based.

Referers can use information in the OORR registry to determine whether their OpenURL links should be based on IP address or institution affiliation, and the appropriateness of the resolver to the particular item. For example, if a resolver only supports journal and article metadata requests, the referer would not want to generate a link to that resolver if the target item is a book.

In a similar model to the Edina database, the OORR will benefit libraries by allowing them to register their resolver information in one place so that their licensed content may be accessed from a wider array of information providers without the librarian administrator needing to register multiple additional times.

Cross publisher linking

Library-to-publisher linking is just one issue. Arguably of equal importance is the issue of cross publisher linking, facilitating citation links critical to researchers. To address this, CrossRef was set up to enable cross publisher links -- publishers which sign up directly with CrossRef agree to make articles available centrally instead of setting up lots of individual contracts with each other.

Link persistence is a key issue as it is essential that links continue to point to the appropriate resource, at least for an agreed length of time. DOIs aim to do this by providing a centralised database of links that are maintained, so CrossRef publishes a DOI for each article that is attached to the current URL. The DOI initiative has recently been given a boost by US publisher Bowker's announcement that it will turn ISBNs into DOIs if requested by the publisher.

Another key trend is the increasing granularity of DOIs. They do not simply refer to an article title or author, but may also refer to chapter headings and are being extended to conference proceedings and research papers.

Amy Brand, CrossRef's director of business development, says: "We currently have 1.3 million chapter-level DOIs from about 30,000 books and conference proceedings volumes. These include several large references works. For instance, OUP's Oxford Dictionary of National Biography uses over 65,000 DOIs in one volume, with DOIs assigned at the level of dictionary entries." However, CrossRef does not take into account access rights of an individual or institution and a link resolver is still needed.

CrossRef is also working with Google to enable persistent and appropriate links to scholarly research via Google in a search pilot that includes 43 CrossRef member publishers. Academic librarians have often railed at the use of Google to locate academic resources when they could make far more sophisticated search and locate facilities available to find what their library users want.

According to Brand, CrossRef and Google are working together to ensure that publishers' priorities are reflected in Google Scholar a with links to the published version of the article appearing first in the results set and the DOI being used wherever possible to direct crawling and to link the user to the full text.

CrossRef has made fast progress towards it goals, but it faces many issues. US-based Diana Bittern, Ovid director of software product management, and a linking technologies expert, has been heavily involved with CrossRef and the DOI initiative. She says: "CrossRef is an excellent initiative that attempts to corral a very large, disparate 'herd of cats' (publishers) into a coherent system of reference linking. The idea of persistence in a DOI is admirable, and for a database that contains over 15 million records, the overall reliability is good.

"However, CrossRef is a young organisation that involves over 300 member publishers of varying sizes and technical maturity, and has limited methods of policing its policies." She points out that when a publisher deposits DOIs, there is little validation on the quality or completeness of the data. When a publisher transfers ownership of a journal collection, the new publisher may wish to establish ownership of the DOIs and may deposit new DOIs for the content, resulting in duplicate DOIs for the same article.

Bittern says: "If the publisher identifies the problem and re-deposits the DOI, the link may be restored, but often, there is limited information available about if and when a broken DOI will be restored. In all fairness to CrossRef, imperfection is a reality of linking, and CrossRef performs an admirable service to publishers and the world of scholarly researchers by attempting to set standards." She adds that member publishers can help by imposing stricter quality measures on their own DOI registration operation.

Rachel Bruce, programme director at JISC, comments: "There are indications that DOI usage is increasing as costs diminish. For persistent identification JISC recommends using http URIs [a generic name for web identifiers such as URLs] but you have to guarantee persistence for 10 to 15 years."

Many libraries have implemented URL resolvers or subscribed to resolver services. However, the linear chain between library, its link resolver and the paid-for content, is only one simplified model among a host of linking possibilities. Libraries may wish to access internet resources, open access resources and shared resources held by sister libraries or consortiums. A number of technologies and standards exist to meet these needs and many overlap.

Search and linking technologies may well ultimately merge. SRU and SRW technologies (see Jargon Buster) focus on search issues, while OpenURL is an access protocol, but it is easy to view OpenURL as a search syntax, with a relationship to SRU.

Search and retrieval of open access resources, by definition, does not have to concern itself with access issues. The OAI Protocol for Metadata Harvesting (OAI PMH) is not a search protocol, but there is a clear relationship between SRU/SRW and OAI-PMH.

Globalisation of initiatives and the continued co-operation of DOI registration agencies will hasten persistence. More granularity, so that DOIs are attached to chapters and sections of text, will allow more intelligent searching and linking citations. Embargo support will increase access to free content embargoed by publishers.

The goal of persistent links to relevant copy will benefit everyone and there is no doubt that everyone, from publisher to librarian, from researcher to technology solutions provider is working to that goal. l