Consider a Sharium

Gary Marchionini

University of North Carolina at Chapel Hill

DRAFT 1.1

Digital libraries (DL) are emerging in a variety of forms and venues around the world (e.g., see April 1998 Communications of the ACM). As the logical extensions and augmentations of physical libraries in the electronic information society, DLs extend existing resources and services quickly and broadly and eventually will augment libraries by enable new kinds of human problem solving and expression. Like all innovations, digital libraries will go through phases that emulate, replicate, extend, and finally augment existing solutions. This presentation aims to encourage the DL research community to first look beyond the collections of libraries to other library services and roles that digital libraries may emulate, and then to ways that digital libraries might yield new kinds of services beyond today’s physical libraries. To illustrate one trajectory toward these goals, the concept of a sharium is introduced to describe digital libraries that combine elements of learning communities, scientific collaboratories, and special libraries to distribute the load of solving information problems. The sharium is used to consider extending current research more fully in the public services area of libraries and propose leveraging technology to create new levels of human participation and contribution.

A new term may be helpful at this point in time to jar us from the somewhat constraining connotations of the terms "digital" and "library." A sharium is all the things that a library is today, but also adds strong sharing components. First, creating facilities and tools to support ephemeral sharing of time and expertise during the collaborative problem solving that takes place in physical libraries is a grand challenge for extension and amplification in digital libraries. The second type of sharing that must be supported is not typical in physical libraries and allows individuals to contribute physical objects and materials to the collection. This physical sharing suggests one possible augmentation effect of DLs.

Extend DL research and development to reference, referral, and education.

The DL research and development community has identified many important research problems and is working with practitioners and businesses to create and maintain digital collections. The most common research and development issues addressed in the DL literature include: object acquisition and digitization; development of indexing and search procedures and tools; delivery of digital objects instantly and globally; management of intellectual property rights; security and authority; interoperability within and across DLs; and managing the long-term integration of analog and digital materials. Although these are critical problems for DLs, they center on the content of libraries. Clearly, libraries exist to aggregate, preserve, and serve information manifested in artifacts (whether analog or digital) and our efforts to transition to DLs must address these basic needs.

What is conspicuously absent from these issues is attention to a second primary function of libraries, what is often termed public services. The main public service function in most libraries is reference, the process of aiding patrons with informational problem solving. With few exceptions, most DLs provide a search engine, a query or browse-based user interface, and possibly a frequently asked question (FAQ) list and expect information seekers to help themselves. This is akin to eating out of vending machines--we should at least be able to get to the level of fast-food drive through and aim toward the full-service restaurant! With the exception of work on intelligent reference agents, this is an important aspect of physical libraries that is yet to be addressed by the DL R & D community. This has been recognized by many in the DL community (e.g., Paepcke, 1996 http://www.dlib.org/dlib/may96/stanford/05paepcke.html).

It is easy to understand why these problems have not been addressed—they are exceedingly difficult to solve and this was known long before the WWW caused everyone to be an online searcher. Studies of the search interviews that intermediaries conduct with patrons before conducting online searches (e.g., Auster & Lawton, 1984) and investigations of professional intermediaries actually conducting searches in online databases (e.g., Fidel, 1984) demonstrate the conceptual challenges of reference services. More than a decade of research at MIT on the Intrex and Conit systems to automate online searching (e.g., Marcus, 1983) illustrates the complexities of online interactions and the need for user support in using online services. Search front-ends such as Grateful Med for medical literature and various interfaces for end-user searching of commercial services gave end users access to online information but even these systems included training, manuals, and online help to support fairly sophisticated end users such as medical and legal students and practitioners. The long series of studies of online public access catalogs (OPACS) demonstrate the wide range of problems undergraduates and other end users have with even fairly recent systems (e.g., Borgman, 1996).

WWW search services have addressed the problem by applying thirty years of information retrieval research to allow users to enter "natural language" queries that return ranked lists of hits from the entire universe of material. Unfortunately, people tend to enter very few terms (one or two) in their queries and not limit their search. It is certain that the general population’s facility with these systems will improve over time with increased familiarity but it seems clear that even with powerful tools, people are often unable to solve their information problems on their own.

Another approach is to offload the complexities of the search process to software processes. The Michigan DL project envisions user interface agents as one of three types of autonomous, intelligent agents that will serve to carry a user profile and query through the architecture to find information (e.g., http://www.cnri.reston.va.us/home/dlib/July95/07birmingham.html). Although this is a provocative line of development, no working systems have yet emerged. More significantly, the information-seeking process is most often embedded in a larger problem solving space that involves learning and decision making—at the very least, users must still be able to articulate queries, let alone define good profiles and manage their agents.

User interfaces to search systems can make query specification and browsing easier for information seekers. Shneiderman, Byrd & Croft have identified a taxonomy of search features that should be included in search interfaces (http://www.dlib.org/dlib/january97/retrieval/01shneiderman.html ) but given that even professional intermediaries have difficulty helping users specify their true information needs, improved query specification interfaces are only part of the solution.

Clearly, hiring millions of professional intermediaries to serve the world’s population is not a viable solution. Ideally, people should be able to solve most of their own information problems. Systems should help us do so, and when we cannot solve a problem ourselves, other humans must be involved. This is fundamentally a communication and apportionment process.

This is especially important given that email and other communications functions are among the most popular and useful applications of the Internet. Libraries serve as meeting places where groups can physically get together (e.g., public libraries often have meeting rooms for community groups and academic libraries have study rooms and commons where students and faculty can meet). The ancient Library of Alexandria was as much a university and gathering place for scholars as a collection of scrolls. Libraries are spaces where people can come together to collaboratively solve information problems. People are drawn to these spaces by the material resources assembled to meet the common needs of a community of practice. The virtual community aspect of DLs was one of the main foci of the UCLA-NSF Social Aspects of Digital Libraries Workshop (http://www-lis.gseis.ucla.edu/DL/).

A promising thread of DL research is to find ways to bring people together to help each other, especially when individual information-seeking actions are unsuccessful. A sharium will provide facilities and tools to allow community members to share their time and expertise. To be scalable, this approach must go beyond the specialized online reference services that link information seekers to professional librarians (e.g., Abels, 1996; Lankes, 1998) to link any individuals or groups on a just-in-time basis. Newsgroups (USENET News) and electronic bulletin boards, and more recently, chat rooms allow communities of practice to form as support question answering. Ackerman’s AnswerGarden system (1990, 1996; http://www.ics.uci.edu/~ackerman/docs/cscw96.ag2/cscw96.ag2.html) is the best example of a hybrid solution that combines FAQ and email question answering (with the human-generated answers automatically added to the evolving FAQ). Finding ways to organize and coordinate services ranging from ad hoc question asking to extended discussions is a DL challenge. Key technical problems include:

transitioning information seekers from self-directed search to reference and community assistance;
developing layers of assistance from fully automated (e.g., FAQs), through community assistance (e.g., posting a question to a newsgroup), to professional assistance (e.g., online with a reference librarian), including hybrid solutions; and
creating interfaces that do not overwhelm or frustrate information seekers as they transition through the various service layers.

Considering DLs in this communication sense seems exactly the right thing to do given that much of the crucial knowledge we need and can best understand is contained in other people’s heads--only formalized manifestations generalized to facilitate communication are externalized into publications of various kinds. Interacting with other people who have the knowledge we need personalizes and customizes that knowledge and makes it much more accessible and useful. Additionally, the interaction itself provides metacognitive benefits and aids in finding the needed information in our own heads as we act to articulate, reflect, and examine feedback. It seems essential that if DLs are to truly evolve to be as useful as physical libraries they must strongly address communication capabilities in general and the reference problem in particular.

Augment physical library constraints on contributions.

DLs are first extensions of physical libraries and eventually will be augmentations of libraries (Marchionini, 1998; http://ils.unc.edu/~march/digital_library_R_and_D.html). It is clear that digital and physical libraries will co-exist for the foreseeable future and just as the information technology industry finds ways to integrate digital and analog solutions (e.g., Xlibris; CrossPad; Ariel; fax and voice with desktop/mobile computing; July 1993 issue of CACM on augmented environments), the library community will investigate ways to integrate digital and physical services and in cases where integration is not feasible, to articulate complementarity to information seekers. DLs certainly broaden and hasten access to digitized resources and we will discover the consequences of these extensions as well as new library capabilities made possible by the digital medium. One capability that DLs offer is the potential to allow patrons to become active contributors to the collection.

Unless the donor is a significant personage, most librarians shudder at donations to the collection. These materials must be sorted, tagged, cataloged, shelved, and preserved; all expensive activities. Thus, donations of books and journals to physical libraries are often more a burden than help and understandably few libraries actively promote patron giving of such materials. A sharium, on the other hand will solicit and welcome patron contributions of physical artifacts and/or their digital representations.

The key technical research issues for such an environment then relate to contribution mechanisms that allow people to easily digitize and/or transport objects. Other research issues include:

managing content authority and quality;
insuring access (metadata formats; preservation and persistence).

Unlike a physical library, where the burden of collection development is on the librarian, in a sharium, the community assumes some of the responsibility for quality control. The examples of collaborative filtering (e.g., Goldberg et al., 1992; Shardanand & Maes, 1995) illustrate practical means by which communities may categorize and rate information objects. Collaborative ratings only provide one aspect of quality and other approaches must be taken as well. These include well-known methods of peer/panel review, and personal annotations made by contributors or users. We can envision tiers of ratings where objects may exist in multiple "bins" of quality, topic, and authority.

Contributors may wish to maintain ownership of objects while providing access to the community. Policies and tools for preserving objects and links and templates for assigning metadata must be added to the digital librarian and patron toolkits.

Supporting both ephemeral and physical sharing requires significant technical developments but fundamental research on human motivations and behaviors in shared environments is also needed. We know that people are motivated to share, whether for philanthropic or self-preservation motives. Public libraries serve to share expensive resources across a community and the Linux community (http://www.linux.org/)

represents one of the most successful digital shared libraries of user-contributed objects. Whether people are more willing to share ephemeral or physical objects, what critical mass is necessary to develop a collection, and whether people will share in more general, ad hoc areas remain open questions.

Learning and Sharing

Another way to think about the sharium concept is from the learning mission of public and academic libraries. Libraries serve as centers for self-directed learning. The great Carnegie libraries were created in large measure to allow those who could not go to universities the opportunity to freely learn on their own. We have argued that DLs will bring formal (school-based), informal (self-directed), and professional learning closer together (Marchionini & Mauer, 1995) and the sharium can act as an open school where individual, self-directed learning that libraries have always facilitated can be extended to collaborative, self-directed learning unconstrained by distance and time. The learning page at the Library of Congress (http://memory.loc.gov/ammem/ndlpedu/) provides excellent primary materials for students and teachers, the University of Michigan DL Teaching and Learning Project (see Soloway, DLIB) (http://mydl.soe.umich.edu/index.html) guides students to do inquiry-based learning with Internet tools and resources. In both cases, these DL resources can be used for formal, school-based learning or self-directed learning. One example of a learning community project that includes the contribution aspects of a sharium is the Baltimore Learning Community (http://www.learn.umd.edu), a DL of multimedia materials augmented by lesson-construction and presentation tools for middle school science and social studies. It is important to note that this project is focused on teachers first rather than students. The collection of materials not only grows as a result of the curricular needs of the teachers involved, but the modules that teachers create may themselves be contributed back to the DL. At present, approximately 100 modules have been created and teacher-defined procedures for contributing and reviewing modules are under development. As long as the community remains small (fifteen teachers), the contribution process is not problematic, but we are looking for scalable solutions that do not burden the submitter or DL managers.

As the DL R & D community continues to mature, we must seek new challenges that leverage information technology to not only do all the critical things that physical libraries do, but create new products and services that advance the missions and impacts of libraries and the communities of practice they serve. Consider the sharium as one placeholder token for the DLs unhindered by the constraints of atoms.

Other Models of Sharing (to be added)

Linux Community

OCLC

Oxford English Dictionary and Dictionary of American Regional English

MUDS/MOOS and cyberspace democracy: what are the rules of interaction in a sharium?

Leveraging Technology to Share Collective Intelligence
[proposal details lay out strategy]

References

Abels, E. G. (1996). The e-mail reference interview. RQ, 35, 345-358.

Ackerman, M.S. & McDonald, D. W. (1996). AnswerGarden 2: Merging organizational memory with collaborative help. Proceedings of the ACM Conference on Computer-Supported Cooperative Work (Boston, November 16-20, 1996). New York: ACM, 97-105.

Anderson, R.H., Bikson, T., Law, S., & Mitchell, B. (1995). Universal access to e-mail: Feasibility and societal implications. Santa Monica, CA: Rand.

Auster, E. & Lawton, S. (1984). Search interview techniques and information gain as antecedents of user satisfaction with online bibliographic retrieval. Journal of the American Society for Information Science, 35(2), 90-103.

Borgman, C. (1996). Why are online catalogs still hard to use? . Journal of the American Society for Information Science, 47(7), 493-503.

Fidel, R. (1984). Online searching styles: case-study-based model of searching behavior. Journal of the American Society for Information Science, 35(4), 211-221.

Goldberg, D., Nichols, D., Oki, B., & Terry, D. (1992). Using collaborative filtering to weave an information tapestry. Communications of the ACM, 35(12), 61-70.

Lankes, R. D. (1998). The virtual reference desk: Building expertise into information systems. Paper to be presented at the 1998 annual meeting of the American Society for Information Science (Pittsburgh, October 26-29, 1998).

Marchionini, G. & Maurer, H. (1995). The roles of digital libraries in teaching and learning. Communications of the ACM, 38(4), 67-75.

Marcus, R. (1983). An experimental comparison of the effectiveness of computers and humans as search intermediaries. Journal of the American Society for Information Science, 34(6), 381-404.

Shardanand, U. & Maes, P. (1995). Social information filtering: Algorithms for automating "Word of Mouth." Proceedings of CHI ’95 (Denver, May 7-11, 1995). NY: ACM Press, 210-217.