Which Semantic Web? Catherine C. Marshall Microsoft Corporation 1 Microsoft Way Redmond, WA 98052 1 (425) 705 - 9057 [email protected] Frank M. Shipman Department of Computer Science Texas A&M University College Station, TX 77843-3112 1 (979) 862 - 3216 [email protected] ABSTRACT Through scenarios in the popular press and technical papers in the research literature, the promise of the Semantic Web has raised a number of different expectations. These expectations can be traced to three different perspectives on the Semantic Web. The Semantic Web is portrayed as: (1) a universal library, to be readily accessed and used by humans in a variety of information use contexts; (2) the backdrop for the work of computational agents completing sophisticated activities on behalf of their human counterparts; and (3) a method for federating particular knowledge bases and databases to perform anticipated tasks for humans and their agents. Each of these perspectives has both theoretical and pragmatic entailments, and a wealth of past experiences to guide and temper our expectations. In this paper, we examine all three perspectives from rhetorical, theoretical, and pragmatic viewpoints with an eye toward possible outcomes as Semantic Web efforts move forward. Categories and Subject Descriptors H.5.4 [Information Interfaces and Presentation]: Hypertext/Hypermedia – architectures, theory, navigation. General Terms Design, Experimentation, Standardization, Theory. Keywords Semantic Web, Hypertext, Digital Libraries, Knowledge Representation, Knowledge Acquisition, Information Systems. 1. INTRODUCTION Over the past decade, the Web has grown from what many perceived as an improved Gopher interface to become the new medium of communication. It would have been hard to predict such a transition; the doubts that many researchers had about this outcome turned out to be misplaced. So when we read about the Semantic Web as the next era of the Web, we are less critical of the claims – we do not want to make the same mistake twice. Yet it seems prudent to examine the future of the Semantic Web more carefully with an eye toward differing perspectives on and expectations of its use, as well as theoretical and pragmatic considerations that will affect its evolution. The Semantic Web is the outgrowth of many diverse desires and influences, all aimed at making better use of the Web as it stands. The anxiety over the apparent disorder of this new world of digital documents – how one makes sense of new genres, new technologies, and new uses and modes of publishing and organizing materials – is one such influence [24]. A second comes from the field of Artificial Intelligence, with its maturing sense of the kinds of computation that can take place given formal representations – what kinds of problems are tractable to the methods that have been developed over the past 30 years (see for example [27]). Finally, there is a utopian desire to offload the burden of information overload and the complexity of everyday life onto the computer, using the vast resources that have accumulated on the Web as a backdrop to help us in our everyday activities and to address the most normal of problems [5]. All three of these desires and influences are readily justified, given the scope and depth of the information on the Web; we now must ask ourselves which of them are realistic? How can we set appropriate expectations for the reach of the Semantic Web? From the W3C’s inception, there was a perceived need to bring order to the loosely connected networks of digital documents that made up the Web. Although this order was to be realized by consortium’s development of standards, it would also reflect the order that libraries have and the Web does not – a consistent structure by which people can access materials. More recently, we can see evidence that this view of the Semantic Web is still widely held in the Hypertext and World-Wide Web communities [8]; Scenario 1 in [29], an information access scenario in which the retrieval is aided by semantic metadata, is a good example. A second perspective for the Semantic Web is one of a globally distributed knowledge base. This perspective on the Semantic Web was put forth early in the Web’s development by Berners-Lee, who began his efforts with the aim of eventually creating networked knowledge ontologies [3]. Berners-Lee has gone on to describe the Semantic Web as being able to learn from the experience of Cyc [23], creating an infrastructure for knowledge acquisition, representation, and utilization across diverse use contexts [4]. In scenarios reminiscent of Apple’s Knowledge Navigator vision from the mid 1980’s [1], this global knowledge base will be used by personal agents to collect and reason about information, assisting people with tasks common to everyday life. A third perspective on the Semantic Web is as infrastructure for the coordinated sharing of data and knowledge. In this vision, Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. HT ’03, August 26-30, 2003, Nottingham, United Kingdom. Copyright 2003 ACM 1-58113-704-4/03/0008…$5.00. 57developers create a distributed knowledge or data base for their particular domain-oriented applications. The representation language, the communication protocols, and the access control and authentication are handled by the Semantic Web. This perspective is similar to Bieber and Kacmar’s efforts to add computation to hypertext [6], and Halasz’s exhortation to this effect in his influential Seven Issues paper [17]. These three perspectives lead to very different expectations of what the Semantic Web will bring to the Web as we know it today. Some of these expectations consider only the technical feasibility and do not consider the social and cognitive implications of the approach, much like Xerox’s 1970 vision of a paperless office [36]. Other expectations ignore the difficulty of scaling knowledge-based systems to reason across domains, like Apple’s Knowledge Navigator, or are overly optimistic that common sense results from the representation of a