purple motes

Web citations make connections well. Web citations combine widely varying citation texts with a standard address (URL). The authority for a standard address can remap it and associate other addresses with it. Anyone can create citations and addresses. Freely available search engines find all citations and rank them according to various algorithms. The result is an inclusive, decentralized environment for making relevant connections among digital works.

While scholarly citations typically indicate works with relatively high knowledge value, the traditional scholarly citation apparatus makes connections much less effectively. Consider this bibliography. It's unlinked, not-marked-up, plain text. Because (standardized) bibliographic control is a very difficult problem, and many different reference formats exists, a reference text string provides a weak mechanisms for searching for works making the same reference. Moreover, tools for ranking the relevance or importance of web documents that include a common reference string probably don't recognize unlinked references.

Some services analyze citation links. The Science Citation Index is an expensive service that analyzes citations from a closed universe of sources ("3,700 of the world's leading scholarly science and technical journals covering more than 100 disciplines"). Evidently, that was too narrow of a universe, because "Also available is Science Citation Index Expanded™, which covers more than 5,800 journals." Google Scholar recognizes citations in papers posted to SSRN, arXiv, and other scholarly paper repositories. But it doesn't include plain-text citations in blog posts, syllabuses, and bibliographies posted on the web. More significantly, Google Scholar doesn't provide a mechanism for end-users to create a value-added reference link. Google Scholar adds whatever value it chooses to whatever plain-text references it recognizes.

I could link a book reference to the relevant WorldCat entry. That would provide some additional information about the book, allow others to download conveniently the citation, and show libraries that hold the book near me. But a person who followed the link probably would be more interested in libraries that hold the book near her. Moreover, searching for links to a WorldCat entry doesn't seem to be helpful for uncovering related work or interested persons: I've found no such links in this example and others. In addition, I couldn't create a link to WorldCat by adding to it a book that is not already in it. WorldCat's control over book records and addresses significantly limits the attractiveness to new enterprises of building services using these data.

I could link a book reference to the relevant LibraryThing entry. That would provide some additional information about the book, provide multiple source options for purchasing the book, provide recommendations for related books, and links to persons who have recorded that book in their LibraryThing library. The LibraryThing entry includes a space for conversations, but these seem to be casual, unfocused, and sparse. More significantly, LibraryThing focuses on collections of multiple-interest books, not on books as work-specific links. These are somewhat different structures of conversation.

I could link an article citation to the relevant CiteULike entry. The CiteULike entry provides links a link to the article if it's available online. The entry also provides rich, formated bibliographic information. Why embed such information in my bibliography rather than link to it, when such links provide considerable additional value? Like LibraryThing, CiteULike links persons who have the same article in their libraries. Just as for LibraryThing, this structure is awkward for relating work-defined collections of references.

In the future I might link book references to Open Library entries. Open Library aims to be an open, extensible catalog for all books. It intends to associate with a catalog entry opportunities to buy, borrow, and download the book. Services could easily be written that use Open Library data to export citations in multiple formats. Independently developed search engines could use links to Open Library records, as well as Open Library record contents, to add considerable value to document relevance ranking. Perhaps Open Library will become one catalog to enable effective citation links, not through bibliographic control, but by becoming a popular reference address space.

Citation and catalog software have not given much attention to hypertext citation links. The Functional Requirements for Bibliographic Records defines user tasks as find, identify, select, and obtain. Why isn't "link reference" a recognized user task? As far as I can tell, Endnote, a popular tool for managing citations and generating bibliographies, provides no support for embedding hypertext references to independent, value-added, web-based bibliographic entries. Perhaps Zotero, a new citation management tool, will make such links a central aspect of citation management. Better networked scholarly citations would undoubtedly spur faster and broader development of knowledge.

4 Comments