The explosion of networking technology has led to the ubiquitous Internet and the large-scale accessibility of diverse information sources. Although there is much debate on the accuracy of Internet measurements, a study [32] done in 1995 conservatively estimates that there are ten million hosts and twenty million users on the Internet with numbers doubling roughly every year. The current situation of numerous incompatible Internet protocols, the large number of users seeking information, and a wide variety of hosts providing information has resulted in user frustration and the inefficient utilization of resources. These issues have not been adequately resolved by current Internet service tools. Information consumers, those seeking information, are left with the daunting tasks of contending with data overload, resolving information heterogeneity, learning different information access methods, and compiling the information obtained. The development of the World Wide Web and the proliferation of easy-to-use navigational tools (e.g., Netscape, Internet Explorer, Mosaic, etc.) have gone far to facilitate the availability of online information but still leave much of the searching and assembly of information up to the user. Users are spending approximately 79% of online time browsing [13], therefore it is estimated that much bandwidth today is wasted due to inefficient searching and ineffective assembly of result data.
A critical challenge presented by the explosive evolution of Internet
is the intelligent and scalable interoperation of data, of heterogeneous
semantics and representations, among the growing number of
autonomous information sources. Among the most
desirable services are: (1) the location of relevant information sources to
answer a consumer's query request, and (2) the integration of information
gathered from multiple information sources
or through different information browsers or GUI tools.
In this paper we report our experience in
building an experimental cooperative information system,
called Diorama, capable of
providing transparent and customizable information access across
multiple heterogeneous and autonomous information sources.
The theoretical model and architecture of Diorama
are based on previous work on
DIOM [17,19,21],
which features the support of the USECA properties throughout the development
of distributed query services.
Diorama demonstrates the systematic design and the flexible
cooperation of the DIOM distributed query mediation services,
and the practical applicability of the
DIOM approach [19,21] towards
an intelligent and scalable solution for interconnecting
information consumers with information producers.
The main contribution of this paper is the practical demonstration of the DIOM information mediation methodology [21] through the design and implementation of an experimental query mediation toolkit in Diorama. We describe Diorama by a walk-through example query to illustrate its ability to automate the linking of information consumers (users searching for information) with information producers (hosts providing information). The query mediation toolkit includes an interactive query tracing facility to support user-driven processing. Diorama also contains a generic wrapper library to support the creation and inclusion of many kinds of information source wrappers.
The remainder of the paper is organized as follows. Section 2 outlines the DIOM System Architecture. Section 3 outlines a sample application. Section 4 describe the implementation. Section 5 summarizes related work and Section 6 concludes the paper.