Customer Reviews

Adaptive Information: Improving Business Through Semantic Interoperability, Grid Computing, and Enterprise Integration (Wiley Series in Systems Engineering and Management)

5 star:		(4)
4 star:		(1)
3 star:		(2)
2 star:		(1)
1 star:		(0)

 

 

You'd think that a book attempting to explain semantic technologies would make an extra effort to be as clear and practical as possible so as not to make the subject even more confusing.

I think this book would have been better if it was written by only one author. Part of the book was written clearly and was obviously directed toward a reader who was not familiar with semantic (and other, related) technologies. This author clearly laid out various ways that semantic technology is (and can be) used. Information provided about current products and companies using this technology was also very informative.

Other parts of the book, however, were written as if the reader is just as familiar with the technology as the author. It seems as though one of the authors was more concerned about how they sounded (in writing) to themselves (and possibly their colleagues) than they were in attempting to give practical knowledge about the subject to an uninformed reader. It's difficult enough to understand the subject of semantic technologies without having to fight through densely-written sentences and unnecessarily vague buzz words (like "follow-on query"??).

I was interested in learning exactly HOW this technology enables interoperability "with little or no human intervention" but found no difinitive, practical explanation of how this is achieved. Yes, I understand that ontologies and inference engines are involved, but how do they work together to remove the human part of the middleware equation? How is the "mapping" involved in semantic technologies different from the "mapping" that is involed in today's EAI solutions? I thought that this is the most exiting and unique selling proposition of this technology but I finished this book unsatisfied with how this was explained. Perhaps this selling point is more of a vision than a reality?

How is the meaning in natural language related to the meaning in computational machines? The answer to this question is of course of immense importance to the information age, and the lack of an answer has resulted in vast financial commitments from businesses that depend heavily on information technology. The sharing of information between machines has been hampered by the lack of common understanding between these machines, since each one of them has its own way of formatting or conceptualizing the data. Even though some of these machines should definitely be characterized as `intelligent', there is of yet no machine that can communicate or share information with any arbitrarily selected machine in a manner that is independent of the nature of this information. There are some machines that are capable of interacting with many other machines in this manner, but at some point when confronted with a particular machine, they are unable to converse meaningfully without some amount of human intervention. This intervention must take place because the machines do not understanding the meaning of each other's data. They cannot conceptualize it without the assistance of a human, who must then give the appropriate (semantic) translation between the data patterns of the respective machines.

The authors of this book approach these questions in terms of what they have called `semantic interoperability.' This notion is supposed to settle the difficulties of meaning and definition that occur not only in natural language but also in communications between software applications. The context of words in digital systems is dependent on both domain cues and local cues, just as is the case for natural languages, they authors say. Data context for example influences the interpretation of data, and therefore it's meaning may change if viewed from another context or perspective. `Data semantics' then is the meaning of data, and will change as the context changes. Therefore, a successful data processing system will need to make the data semantics explicit. The authors discuss various approaches to the understanding of semantics, such as pattern analysis, schema mappings, and abductive logic. Of all the approaches discussed, the authors seem to favor the one based on abductive logic, referring to it as the least developed approach but one that shows the greatest promise in going beyond rule-based digital systems. However, the authors are incorrect in stating that this approach is not very well developed, as there are a few highly resilient systems, used primarily in bioinformatics and telecommunications, that make heavy use of abductive reasoning.

The conflicts that can occur between data on different machines motivate the authors to consider various approaches in dealing with these conflicts that does not involve customized code. After reviewing the specific types of conflicts that can occur, such as those due to data type, labeling, naming, and domain, the authors review some of the different `semantic solution patterns', for dealing with them. These include the `machine learning pattern' that is based on statistical analysis and reasoning patterns from artificial intelligence and is used to discover semantics within instance data; the `third-party reference pattern' which uses a thesaurus or ontology having a shared meaning across sources and targets; the `model-based mapping pattern' which uses well-defined metadata about context instead of mappings between data structures; and the `inference pattern' that requires a formalism to describe the semantic relationships in the system.

The authors obviously believe that these approaches are not entirely satisfactory, or they would not have written this book. Most of the book therefore is devoted to their solutions for solving semantic conflicts. Central to their approach is the role of metadata, which is viewed by the authors as forming a hierarchy with six layers: instance data (essentially "raw data"), syntactic metadata (needed to process data), structural metadata (which gives form and structure to units of data), referent metadata (to provide linkages between different data models), domain metadata (forms a "conceptual domain ontology" to provide a reference point on which all metadata can be understood), and rules (which constrain the semantics of metadata specifications).

One might call the authors approach a version of ontological engineering, the latter term being used currently to describe efforts to make data understandable in different contexts. In this regard, the authors consider four different types of ontology, namely interface (models essentially the API), process (applying to time-dependent processes), information (specification of a collection of concepts for a given scope), and policy (specification of rules of usage). Of particular interest in their discussion is that of the transformation of ontologies, which allows the moving of data from one model to another.

The approach that is advocated in this book is that of semantic interoperability, which the authors view as a `multimodal' solution since it can apply to different modes of operation and is not tied to a particular technology. A successful semantic architecture however is not a "plug-in" to be incorporated easily into standard middleware. Instead, it is an engine that is dynamic and adaptive, and responds real-time to changes by generating its own instruction sets. At first glance this engine would seem be one that is enormously complex, but the authors break it down into its modes of operation, in order to clarify just how it would function. Crucial to their conception of semantic interoperability is that it allows the discovery and utilization of new information without the intervention of a human. If such a capability can indeed be realized, it would definitely be a major advance and would justify the expense of its operation. The authors discuss in fair detail just how semantic interoperability would work in a business enterprise, at least schematically. To alleviate any skepticism on part of the reader of the reality of their ideas, they devote a large amount of space in the book to (nonproprietary and proprietary) case studies of implementation of semantic interoperability. Although these studies to not encapsulate all of the author's ideas, they do serve to convince the reader that these ideas should be taken seriously.

This is the best book on Semantic Interoperability and Ontologies to date. Full of real solutions, atlternatives and product recommendations.

This book will not disappoint. For serious semantic web practitioners ( novice or pro ), this book is an invaluable asset. The text provides a very detailed and expansive commentary on the current 'state-of-art' in Semantic Technology, rich in case studies the authors hit their mark squarely.

Ever read a book that spends hours developing background, story, history, etc and it never really ends? Like you are waiting for another volume? This book is exactly that. It spends pages and pages talking (not explaining) semantics and there is NO punchline. It is purely a theory book with no examples or applicable case studies. It explains how data is rigid and existing technologies are no good, etc. How middleware is tightly coupled and how it's awful. (when really Middleware is like 50% perl scripts) It also has the tendency to use technobabble run on sentences using words like thesauri, axioms, formalism, quasiformal, frictionless, etc.

The concepts are great, but 375 pages of TALKING and not one page on HOW TO IMPLEMENT is pretty unhelpful.

By gripe is this... EXPLAIN HOW TO USE SEMANTICS and use examples and applicable case studies. Anyone can write 375 pages about how it SHOULD BE and why new technology is needed but not many people can actually explain how to invent something new or actually produce a WAY to do something. This book is for academics and philosophic thinker types but it's also decent if you want to ingrain the concept of better design. The problem is... 375 pages to tell you that current data schemas and design arent great is a bit excessive. PLUS, this EXACT concept is drilled into your head all through grad and ungrad. Why again????

This book was used in my graduate class and I really dread reading the chapters each week.

As a student of SOA & adaptive business networks (composite channels), this book fills a void in my knowledge base. The book is worth buying just for the chapter on Capability Cases, a pre-use case methodology.

I bought this book to learn about Enterprise Integration. It is a good introductary book on the topic

Infrastructure, as a term is quite familiar, and deals with information about the tools and the hardware. Infostructure, is existing in content and is yet to get the global acceptability and appropriability. I see this book as an exclusive text on infostructure, as well as the middleware (other than the hardware, software, etc.)

I found this book thought provoking. It applies various methods of disseminaing the message (that would be otherwise lost in the medium).

Most books lack the skill to present the outsider's experience in the Semantic Web. In this regard Pollock and Hodgson get a high mark for covering in a very unique way experiences of the workplace, under the caption: 'Insider Insight.'

In the preface is the scope of this term: "These are 'guest spots' where industry lumniaries voice their thoughts about an important topic of interest. Thee sidebars are inserted throughout the text wherever an 'insight' overlaps with the content being discussed in the main text. In this way, the reader should enjoy a diversity of ideas surrounding many topic of interest in the book."

The three Listmania (Amazon's book store catalog, prepared by store's visitors) and listed on this page contribute in reflectng the relationship, depth and coverage of this book.

[...]