A New Framework for Web-based Contributory Encyclopedias

The current graph structure of the Web, which has been shown by recent
studies to be remarkably weakly connected, suggests in practical terms
the continued need for some centralizing mechanism for knowledge
organization, like traditional encyclopedias. However, such a mechanism
should extend current models of on-line encyclopedias to allow
self-selecting communities of people to easily author and contribute
articles to a common database. Furthermore, those articles should be
automatically indexed and interconnected in useful ways, and users
should be provided with sophisticated search mechanisms for locating
related collections of articles in the database. This is what Pierre
Levy calls a cosmopedia, an intentional knowledge space which is
"deterritorialized," yet can be easily located and used for developing a
"collective intelligence." Such a system is referred to generically as a
contributory encyclopedia. Our framework for contributory encyclopedias
is called a Communipedia. It is a generic framework that allows the
creation of any number of distinct compilations of information. The
paper proposed for ACH/ALLC 2001 will detail motivations, design and
implementation of this framework for creating and managing Web-based
contributory encyclopedias.

The encyclopedia represents an early advance in the organization and
distribution of knowledge. Traditional encyclopedias - whether in print
form or, more recently, on CD-ROM - provide highly organized systems of
information, but they are static and are often seen as elitist.
Historically, encyclopedias have come to be seen as not simply forms of
knowledge organization, but as social processes which seek to define
which knowledge is valid and important. The editorial processes used to
create encyclopedias are usually closed to all but a select editorial
board, and both traditional and CD-ROM-based encyclopedias are expensive
to produce and own.

The World-Wide Web (WWW) addresses one aspect of this problem, in that
it represents a revolutionary step in democratizing the organization and
dissemination of knowledge. It has allowed people all over the world to
author and share information in a distributed and independent fashion.
Recent studies have shown, however, that most of the WWW is poorly
connected. This makes it difficult for users to find information in
large parts of the WWW. It also makes it less likely that others will
find the information that users intend to share. Overly simple indexing
and searching mechanisms employed within individual Web sites compound
the problem. One of the main virtues of the WWW is that it enables
distributed and autonomous information sharing, but this situation makes
clear that there is also a need for centralizing mechanisms for
knowledge organization and dissemination that are available to all.
WWW-based encyclopedias fill this role to a limited extent. On-line
encyclopedias can be dynamic and made accessible globally at a lower
cost than traditional and CD-ROM-based encyclopedia, but as with
traditional encyclopedias, contributions are restricted to an elite
group and their enabling technologies are usually not in the public
domain. Furthermore, the WWW-based encyclopedias do not yet take
advantage of more recent technologies, such as XML, for facilitating
better authoring, indexing and hyperlinking.

The focus of this project is to reconceptualize the form and function of
the encyclopedia so as to create a new form which will provide better
support for the organization of historical knowledge and on-line access
to that knowledge than existing WWW-based encyclopedias. This new form
seeks to move the encyclopedia from a static, linear, and elite type of
knowledge organizing to one which is contributory, multi-dimensional,
analyzable, and extensible. The design of this new form is geared toward
use by average users. We call this new form a Communipedia. The data
model supports the representation of composite documents and will be
amenable to both rich forms of hyperlinking and database indexing for
efficient searches.

An instance of the Communipedia framework is contributory in the sense
that any member of a self-selecting community will be able to submit
articles. It will be possible for new articles and revisions of existing
articles to be added to the Communipedia at any time. The framework is
designed to accept articles, provide sophisticated on-the-fly indexing
and hyperlinking of incoming articles, and store them in an XML database
which will support a sophisticated query mechanism. A complementary
design goal has been to allow sophisticated authoring processes for
articles and is at the same time easy to use.

The Communipedia framework is extensible in that it is designed to allow
any user to conveniently add features to its technical design. One
application of this aspect of the framework is the development by users
of new tools for analyzing a collection or new document type definitions
(DTDs) to represent new types of entries.

The basic element of any collection managed by a Communipedia is the
document. All documents exist as first class objects in the database
server and have unique identity. Documents may be one of three
semantically-related types: article, annotation, or map. All documents
are semi-structured data objects modeled in an XML-based schema. Various
built-in semantic relationships between these document types are
supported.

A document of type article corresponds generally to what is thought of
as an encyclopedia article, but also includes multimedia assets such as
images, audio and video. Each article, unlike traditional encyclopedia
articles, however, may be composite in the object-oriented sense,
containing other articles. Allowing composite articles will allow
authors to reuse existing articles in composing new ones.

A document of type annotation is associated with a document of article
type. An annotation may contain a natural language commentary on the
associated article for human consumption, or it may be used to represent
a workflow specification and log for the automated editorial workflow
processing of that article. A special case of the former is an
annotation consisting of only a link to another, already existing,
document. Each article document may be associated with zero or more
annotation documents.

Documents of type map serve as a higher-level form of knowledge
organization than database index structures by representing specific
semantic relationships and roles between articles. These are based on
the concept of the topic map. Each map has a user specified type, such
as "is a", "authored by", or "is related to." The Communipedia system
will have a set of built-in map types; however, users and other system
components will also be able to define their own map types. Each
document (i.e. article or map) which participates in a map has a role. A
map of type `authored by," for example, implies the participation of a
document which represents the author and the document authored by that
person. Since Communipedia may be linked to others for networked
searches, each map must have a specific scope which defines the
collection in which it exists. Map documents are intended to enable more
sophisticated types of browsing and searching through a collection. Maps
can be both the subjects and objects of queries and keyword searches
since they are first class database objects.

An instance of the Communipedia can be multi-dimensional in several
respects. On an article level, a Communipedia goes beyond the linearity
of traditional encyclopedias to provide dynamic hyperlinking between
articles tailored to the query being performed.

On an editorial level, the framework supports optional editorial layers
of articles. Using this feature, a Communipedia could be organized, for
example, into one or more collections of articles developed by
associated editorial boards, while other collections in the same
Communipedia might consist of articles contributed by users independent
of an editorial board. This project envisions a key scenario in which
this feature is critical. The ultimate goal of this work is the creation
of an Encyclopedia Africana through the use of both an editorial board
consisting of traditional, trained scholars along with independent
contributions from communities or individuals. Using this approach, an
entry on African American inventors written by trained historians, for
example, might be linked to contributions of related information by
independent scholars or individuals. Such a contributor might add
information about experiencing the introduction of an invention, an
encounter with the inventor, or a recording or image of the inventor
they own, or the results of their own research.

Other layers can be added dynamically to hold alternate versions or
perspectives of articles, or running commentaries on one or more other
layers. On an access level, Communipedias also support optional
networked interconnection with other Communipedias, using a
server-to-server, information exchange protocol to create so-called Web
syndicates at the server level.

The Communipedia is designed as an extensible, component-based system
consisting of: user interfaces for authors, editors and browsers, an XML
database server, an XML-based workflow management subsystem, and a
peer-to-peer communication system for interconnecting multiple
Communipedia. In addition to keyword search, general query expressions
are supported using an XML-based query language, as well as hyperlink
browsing. The workflow manager automates the routing of documents as
part of the editorial process selected by users of a Communipedia.
Workflows are configurable using specifications written in the WfMC
extension to XML.

This project makes contributions both to the technical aspects of
hypermedia systems and their impact on social systems for knowledge
organization and dissemination. African American Studies is the initial
context for its evaluation. New approaches to organizing, searching and
managing on-line encyclopedias using XML-based technologies have been
developed, which are, in turn, serving as the basis for new approaches
to pedagogy, collaboration, research, and writing in African American
studies and other disciplines. The Communipedia will serve as a
framework for examining a broad range of research issues, including
information needs and use models in marginal communities, comparative
studies of traditional versus vernacular historiography, technology
appropriation and collaborative design in marginal communities, and
universal access to information.