XML in serial publishing: past, present and future

Journal production was one of the first commercial arenas in which markup technologies took hold, and today most major (and many minor) journal publishers have journal content in SGML or XML data formats. However, over the past few years the introduction of XML and its many adjunctive technologies has reshaped the markup landscape. With many publishers having digital workflows that were established before the flowering of XML, and some publishers looking to move to digital workflows for the first time, this article attempts to tour those parts of the XML technology family, and those XML-related activities, that are of most relevance to journal production. Has structured markup made its case? The main promises of structured markup technologies, vendor-neutrality and ease of content reuse, have been palpably realised in journal publishing over the last 15 or so years. Vendor neutrality is (or should be) an unavoidable consequence of using SGML or XML to hold content. Such content conforms to an ISO standard (for SGML), or a W3C Recommendation (for XML), and so is in theory usable by systems which themselves conform to these technical specifications. In practice, journal content often contains some small dependencies on particular systems, for example, hinting to allow a typesetting system to optimize layout, or specific content to ease integration with a particular vendor’s content management system. However, in general once content is marked-up as SGML or XML, the choice of application vendor for tools to process that content widens considerably, especially in the case of XML. Content reuse should be a natural consequence of using structured markup for content. In practice a number of strategic and technical considerations affect how smooth reuse actually is. Strategically, the biggest issue is consistency of content and there is no doubt that even among experienced participants in the journals publishing market, inconsistently marked-up content is widespread. Technically, content must be modelled in a sufficiently abstract way if it is to be re-used easily. Content valid to SGML ∗Copyright © 2003 Griffin Brown Digital Publishing Ltd. A version of this article appears in OCLC Systems & Services Vol. 19 (4), 2003.