Browse Books

Reviewer: George R. Mayforth

The premise behind this book is that, for the World Wide Web to take a step forward in usefulness, its content must take on new characteristics. One of these is the inclusion of semantic descriptions of content, which in turn will allow computer programs to be more intelligent about finding and handling information. This will make human searches for information more efficient and effective. It will also allow for increased complexity of queries, for example, facilitating a query in finding correlation among different sets of data. Talks delivered at a conference in 1999 are aggregated in the book. It presents an overview of the state of development of the semantic Web in three areas: languages and ontologies, knowledge support, and dynamic aspects. For someone unfamiliar with this field, reading the various papers is somewhat like assembling a jigsaw puzzle: as one proceeds, the big picture emerges from the details one gathers along the way. The semantic Web will be achieved by embedding information about content areas along with the content itself. This is done using ontologies, which are formal descriptions of topic areas. The presence of ontological information permits computer programs to interpret the content in a much richer way than would be possible by simply gathering keywords. A number of ontology languages are presented in the book. Most are extensions of Extensible Markup Language (XML), the Web protocol that extends Hypertext Markup Language (HTML) by adding data element definitions. The importance of XML as a Web standard is thereby highlighted. Working with existing Web standards helps support evolutionary enhancement to the environment. Languages discussed include the ontology inference layer (OIL), the DARPA agent markup language ontology (DAML-ONT), simple HTML ontology extensions (SHOE), and unified problem solving modeling language (UPML). A key benefit of the semantic Web is improved knowledge support. The second section of the book covers research into storing and querying information. As one might expect, ontologies are the fundamental way to organize knowledge. A knowledge base must be able to deal with multiple ontologies in order to provide useful access to the data within their respective data collections. Further, knowledge systems must be capable of gathering data from many sources, and of interpreting it to satisfy queries. This will require two things: a degree of standardization of the ways ontological data is maintained, and maintenance of ontological data by site owners. The papers presented here show efforts to realize the first requirement; the second task hasnt really started yet. In addition to querying, semantic information can lead to computer assistance in carrying out tasks. For example, a context- and semantically-aware personal digital assistant (PDA) could provide helpful information to its owner as she or he pursues work or entertainment, such as how to get to a location, where to park an automobile, the best approach to a task, and so on. Software agents could become more effective in carrying out their tasks by being able to interpret the data they encounter, which could lead to being able to automatically handle unusual situations. Given the current state of the Web, it is clear that building a semantic aspect into it is now in the research stage. This book provides a good snapshot of the state of this research from about a year ago: this is a fast moving area, so developments will happen quickly. Overall, the text is clear and readable. There are a few spots where the English is somewhat idiomatic, but these are not a problem. The overleaf describes this as the first handbook for the semantic Web. I disagree with that assessment. This is a collection of papers on the theoretical underpinnings of a possible new evolution for the Web, along with discussions of some experiments that have been done using these theories. Although the concepts are interesting, and their widespread use would increase the power of the Web, its less clear how rapidly the ideas shown here will be adopted and incorporated by major software vendors, by the owners of the billions of existing Web pages, and by the server infrastructure that supports them. The book provides many references and Web links to additional information on the topics it covers. These are welcome. Online Computing Reviews Service

Reviewer: Fred J. Damerau

Edited versions of papers presented at a seminar held in March of 2000 are presented in this book. It demonstrates the unevenness and redundancy common to collections of this kind. The foreword, by Tim Berners of World Wide Web fame, is based on a talk he gave in 1997, advocating what is now called the semantic Web. In reading the main chapters of the book, readers would be advised to keep in mind one of Berners' comments, made near the end of the foreword: "Now we can imagine the world of people with active machines forming part of the infrastructure... I think this will take a long time." Space precludes including a detailed discussion, or even description, of each chapter of the book. The text begins with an introduction to what the semantic Web is, namely that it facilitates knowledge management, Web commerce, and electronic business, and to what will make it possible, namely languages, ontologies, and tools. The book contains five chapters on languages and ontologies, five on knowledge support, and four on dynamic aspects. The first section includes a chapter on the simple HTML ontology extensions (SHOE) language for interoperability; a chapter on the US Defense Advanced Research Projects Agency (DARPA) Agent Markup Language (DAML) for ontologies (DAML-ONT), which is a language for describing Web content; a chapter on the ontology interchange language (OIL) and resource description language (RDF), schema languages which are compared to Extensible Markup Language (XML); and a chapter on the unified problem-solving method development language (UPML), an architecture for heuristic reasoning components. This section also includes a very readable chapter on ontologies in general, by McGuiness, which should perhaps have come first in this section. Section 2 includes chapters on Sesame, an architecture for storing and querying RDF data and schemas, and on a prototype system built by Boeing from existing components; a chapter advocating "a new generation knowledge base, resilient hyper-knowledge base (RHKB) that would be more adaptable to change and reuse;" a chapter on InfoQuilt, which "provides a conceptual framework that allows users to access data from a multitude of diverse autonomous resources and provide tools to help them analyze the data ...;" and a description of an application of semantic portal (SEAL) to the construction of a Web site. The final section includes a chapter on semantic gadgets, describing a fanciful visit to a museum; a chapter on the GLUE language, an extension to Hypertext Markup Language (HTML) that, among other things, permits procedural attachment; a chapter on the desirability of external, as opposed to inline markup, as a means for enabling reuse of markup; and a description of an active ontology and for modeling constraints to characterize plans and processes, and to represent and reason about plans. The Web today is comprised of millions of pages. It is clear that few of these will be retrofitted with the additional markup envisioned by these proposals. New Web pages could be built with these capabilities if there were some advantage to the builder in doing so. Web page builders conform to HTML standards because they want browsers to display their content in a particular way. It is possible to believe that a community with a common purpose (the military, perhaps) might build pages with this more complicated markup in order to interact in a more automatic way, as noted in chapter 15. A group of medical institutions might adopt the same ontology. (Mapping between ontologies, discussed in chapter 5, to the extent that it is feasible, is a way to enlarge a community without requiring massive change by any party.) However, the enrichment of most of the Web for most ordinary users, for ordinary, mundane purposes (chapter 12) seems very far off. In the conclusion of chapter 11, on SEAL, the authors note, "We have shown through our experience that there are many big open issues that have hardly been dealt with so far," like where to put relevant concepts (into the ontology or into the knowledge base__?__). Issues like this are fundamental, and a consensus needs to be reached for real progress to occur. Although the papers are now a bit dated, this book still provides an introduction to ways of thinking about how a more capable World Wide Web might be achieved, and to some of the issues that must be resolved. Online Computing Reviews Service