Reference Linking for Journal Articles: Promise, Progress and Perils

The increased availability of electronic journals on the Web has not only improved access and convenience for readers, but has occasioned many improvements in how journal literature is found and used. In many online services, the full text of journal articles is directly searchable, multiple formats of articles are available, and direct e-mail links to authors are given. Internal hyperlinks within articles improve navigation through different sections and among text, tables, footnotes, and references. Users can tailor their default displays, increase font size, and magnify illustrations. Undoubtedly, however, the most significant innovation of all for the scholarly community is the implementation of reference linking.

Reference linking is one of the important added values to emerge from the rise of electronic scholarly publishing. Indeed, such technologies are a key raison d'etre for the move to electronic publishing. Reference linking has existed in one form or another for some time now, and the technology is advancing rapidly in providing increased functionality for the user and increased feasibility for the producer. However, it is important to grasp the significant obstacles to developing complex applications in the Internet environment where libraries are only one of the players. This column describes the state of the art in reference linking and suggests why librarians should care enough both to understand the technology and to influence the shape of its application.

Versions of Reference Linking

Reference linking, sometimes also called citation linking, can be defined as the ability to go directly from a citation to the work cited, or to additional information about the cited work. The source citation might be an entry in an A&I (abstracting and indexing) database, in an e-mail sent from one colleague to another, or anywhere else you can find a clickable link, but in this discussion we will focus on citations embedded within a journal article, generally in the "References" section at the end of the article.

In its simplest form, reference linking is effected by including URLs in the text of citations. This is often found in informal papers self-published on the Web, and occasionally in electronic journals such as D-Lib Magazine. This is not a very robust approach, however, as URLs are notoriously unstable. It also doesn't scale, as URLs generally have to be discovered and supplied manually. The most viable reference linking therefore makes use of identifiers (names) instead of URLs (locations).

Elaborate systems of name-based reference linking have been developed in medicine and astronomy. At the heart of each of these systems is an A&I database, PubMed, and the Astrophysics Data System (ADS). Journal articles indexed in these databases are assigned identifiers, the PubMed ID (PMID) and BibCode respectively. The identifier retrieves the index record and the index record contains a link to the full text of the article if it is available online. Reference linking for any particular journal article is accomplished by first ascertaining the identifiers for each reference cited, which can be done by a batch look-up against the PubMed database for PMIDs, or derived algorithmically from the bibliographic data for BibCodes. The identifier for the reference can then be queried in the database to retrieve the index record for that article, which in turn contains a link to the full text of the article.

This method of reference linking is in some ways a closed system; it works only for articles included in the core A&I database and for users who have access to that A&I database. However, in these particular cases, the systems are effective, because PubMed and ADS give comprehensive coverage of their disciplines, and because both of these databases are freely available to everyone.

Similar closed systems of reference linking are less successful when centered upon databases with more arbitrary content or restricted access. Early publisher attempts to offer reference linking within their own online journal databases were...