PennLabs

Subject Maps

About

This project is building new tools for browsing and discovering library resources, using conceptual maps based on Library of Congress subject headings. The aim is to provide more effective subject-based discovery that takes maximum advantage of the investments libraries have made in subject cataloging.

The tools cluster and link library item information using existing subject metadata cataloged according to the LCSH subject ontology, letting users "browse the shelves" along multiple topical dimensions, using ordinary text-based web pages. The online "maps" feature side-by-side display of subject terms and items described by those terms, and clustered display of items with related subject terms. These let users quickly see what's available under different catagories, and quickly navigate to promising items and topics. The maps are automatically adjusted to the characteristics of the collection being browsed, and can also be adapted to local communities and alternative subject ontologies. They can be applied to small-scale or large-scale collections.

Applications

Online Books Page collection (25,000+ items, 17,000+ subject headings) -- in production now
Franklin / Penn Library collection (3 million+ items; 1 million+ subject headings) -- demo available only for limited times

Comparisons with other subject browsing approaches

Traditional OPACs (like our own Franklin) all too often present a "worm's eye view" of subject browsing, with only alphabetical browsing and very limited support for finding related subjects. Subject terms, relationships, and items are on different screens, requiring a lot of clicking around to see what's available under particular categories and category sets.
Facet-based catalogs (like the new NC State / Endeca catalog) offer substantially richer browsing environments, and with their relevancy ranking features can be especially useful for casual research. But they don't take advantage of the full range of explicit and implicit subject relationships afforded in professionally managed ontologies like LCSH. We believe that the relationships we find and derive from such systems are very important for in-depth research, and hope to bring them out in our map-based interfaces, while providing interactivity and exploration possibilities similar to those that make catalogs like NC State's appealing.
Social tagging systems (such as our own PennTags) allow people to create shared collections of resources and describe them in their own terms, and can be especially useful for current awareness-- finding out what peers are doing and looking at now. Yet it's unclear how well they work to describe large-scale research collections, where there may be millions of relevant subjects (far more than could fit into a comprehensible tagcloud) that need to be described consistently, and where researchers may need to dig deep to find materials not recently looked at by any other scholar or tagger. There may be some interesting interactions possible between tagging systems and subject maps.
Full text collection searches (such as in Google Book Search) allow one to drill down into books with unprecedented detail. They may prove to be indispensable for many kinds of research, including obscure topics that may not be well represented in book cataloging. At the same time, they may also produce too many results, with inadequate priority and organization, for other topics. Subject headings assigned by professional librarians may be provide much higher signal-to-noise for finding useful research materials on many topics. We hope that tools like the ones we are building in thie project will let researchers make the most of those headings.

Selected articles and presentations on subject maps

Mapping the library future: Subject navigation for today's and tomorrow's library catalogs (2008). Slides from a recent ALA presentation on subject maps.
New Maps of the Library: Building Better Subject Discovery Tools Using Library of Congress Subject Headings (2006). A white paper, and slides from an older presentation, describing the rationale, design, and construction of subject maps..

Planned Improvements and activities

Additional heuristics for relating subject terms
Smarter subject clustering display
Online tools for quick tweaking of subject aliases and relationships (possibly tied to log analysis tools)
Adding facet-based filtering for maps and results
Adding subject keyword search option
Better techniques for presenting major subject "hubs" (e.g. "United States", "Fiction", "History")
Evaluation of map-based subject browsing and other subject search techniques
Combining authority and bibliiographic records from multiple sources
Evaluating using tags as leadin terms to subject maps

Contact

For more information on this project, contact John Mark Ockerbloom.