We will see much more activity connecting user environments and bibliographic resources. I am thinking of citation managers, reading lists, social bookmarking sites (see citulike and unalog) and RSS feeds. Some of these may be specifically supported by the library (e.g. a citation manager service), some may be developed within an academic or scholarly context (e.g. Zotero, citulike, ...), and some may be general network services. People have multiple ways of creating personal and shared collections of data and links.

They are also an example of an increasingly important aspect of our bibliographic apparatus - we have discovery or 'rendezvous' experiences outside the library resource, where it would be good to be able to link back into a library service for fulfillment, or indeed into other services. As we expose more data to search engines, this provides another example. We don't have robust, general ways of doing this across resource types.

In this context I was very interested to read a report from work done at the University of Minnesota on the ability to resolve references in the RefWorks collections of graduate students and others. Here is the abstract:

Introduction. Digital library users collect, enhance and manage their online reference collections to facilitate their research tasks. These personal collections, therefore, are likely to reflect users' interests, and are representative of their profile. Understanding these collections offers great opportunities for developing personalized digital library services, such as reference recommender systems.

Method. We recruited subjects by individual e-mails to the users of RefWorks - a web-based personal reference management tool installed for use at the University of Minnesota. To participate, subjects needed to give their consent and share their references with us. 96 subjects participated, majority (65) of who were graduate students, resulting into 30,336 references. Based on the type of the reference, these were stratified into one of the three valid identifying IDs - DOI, ISBN, or URL. Multiple reference resolvers (CrossRef, WorldCat) were used to enhance the overall resolvability of these collections.

Analysis. Descriptive statistics and simple graphics analysis were used to describe the dataset.

Results. Over 90% of the total references in users' personal collections could possibly have a valid ID (DOI, ISBN, URL), and therefore, are potentially resolvable. However, only about 17% of the references in these collections had a valid ID, and fewer than 11% actually resolved successfully. Using a combination of reference resolvers, the total resolvability of the references in these collections was enhanced from under 11% to over 41%.

Conclusions. Users' personal reference collections have a tremendous potential of building, supporting, and enhancing personalized digital library services, such as reference recommender systems.
[Resolvability of References in Users' Personal Collections]

Related entries:

Comments: 0