Visualizing Bibliographic Metadata - A Virtual (Book) Spine Viewer (16 Oct 2004)
The virtual spine approach has great potential in an increasingly digitized, computer-mediated world of information discovery. At the same time, the author is well aware of the urgent need for a user-centered design process for the viewer application. The design of the virtual spines and viewer will benefit greatly from many iterations of user feedback and newly designed prototypes; the author's work to date is presented here to foster discussion and further work. Among other concerns, the author is particularly interested in user feedback on what virtual spines should look like, how many virtual spines can usefully be displayed in the viewer, how virtual spines should appear when zooming to different levels of detail, and if there is an optimal density of virtual spines. It may also be useful to provide alternate visualizations or views of the virtual spine data. One possibility is a table view, allowing users to sort by any combination of columns to organize the data, and perhaps using Fisheye Menus for scrolling. Spotfire has a terrific table implementation, allowing dynamic queries for selection, sorting and so on, but alas, it's not free or open source.
User feedback is vital, but even without it, there is a long list of ideas and improvements to implement, such as a) displaying axes values in the scatter plot, b) allowing searches on the text associated with displayed virtual spines, c) better indicating when multiple virtual spines map to the same spot in the scatter plot by using slight displacement, often referred to as jitter, and d) allowing users to select more mappings for display characteristics, such as spine color or shape, in addition to the axes.
Some interest has already been expressed for customizable versions of the prototype, and the intention is to create a free, open API to allow maximum configuration possibilities. Three improvements along these lines will be 1) loading the left panel hierarchy in the viewer from a schema-validating XML file, 2) loading data for the scatter plot display from any comma or tab separated text file and 3) allowing an easy way for installation specific spine appearances and mappings.
Explorations with the spine viewer prototype have highlighted the benefits of consistent metadata using controlled vocabularies. To this end, the NSDL is working with partners to provide missing metadata for NSDL resources. INFOMINE15 is to provide LCC subject metadata generated automatically from a resource's metadata and its full text, if available for crawling. The Eisenhower National Clearinghouse (ENC)16 will also be enhancing NSDL metadata to support the filtering of NSDL resources by grade level or content pertaining to educational standards.
The NSDL also aspires to subject classification visualizations that are useful both to end-users and to collection organizers. The NSDL would like richer subject classification schemes to be more accessible, for example using a subset of LCC rather than the limited subset of GEM subjects the NSDL currently uses in some contexts.
In short, the author believes the pairing of information visualization techniques with bibliographic metadata will yield great benefits for information discovery.
Article URL: http://www.dlib.org/dlib/october04/dushay/10dushay.html
Read 58 more articles from D-Lib Magazine sorted by
Next Article: All the News That's Fit to Read