Building Intelligent Web Sites with JavaScript

If you think that DHTML and JavaScript are simply means of achieving gimmicky effects on web pages, then Nigel Ford's latest book will immediately raise your expectations.

Yes, JavaScript can do all sorts of interesting things: image rollovers, pop-up windows, customised time-of-day messages, using cookies to provide different images on different visits, and dynamically generating the content of one frame from another. For example, the image rollover effect works, at its most basic, by assigning different filename values to the "src" property of an "img" object.



It does this using a statement such as "onMouseOver image1.src=image1B.gif". But it can do much more. JavaScript of course is not the same thing as Java. It's not a fully object-oriented language, capable of implementing stand alone applications, but an object-based language that must be embedded into web pages. However, Ford demonstrates that JavaScript is actually very powerful. Forget that it contains the word "script": it's a genuine programming language.

This is not a systematic reference work, with sections offering small illustrative examples. Nor does he dwell on fragments of code for slick effects. Instead, he leads us through the development of a small number of large applications. This is more of a teaching text in the sense that it is orientated towards readers with particular interests and needs. Since Ford teaches in the Department of Information Studies at Sheffield University,these interests are assumed to lie in the areas of expert systems and information retrieval.

To me at least, it seems unfortunate that the publishers have over-hyped what's on offer with techie wording on the front cover - "Mecklermedia Web Developer.Com Guide to Building Intelligent Web Sites with JavaScript: a complete resource for creating websites that think!" This obscures the fact that the book is really about artificial intelligence techniques, with examples and illustrations coded in JavaScript, and that it has genuine academic merit.

After three useful introductory chapters to JavaScript, organised as input ('Obtaining Information from Readers'), process ('Processing the Information to Add Value') and output ('Communicating the Results to Your Readers') the book gets down to its main business. It first creates a backward-chaining rule-based advisor for course entry requirements, and then introduces mechanisms for providing "why I need to know" and "how I concluded" explanations.

It then moves on to consider semantic networks and the use of stored knowledge to drive the Alta Vista search engine. For example, it can be programmed so that a query about Great Britain will automatically retrieve matching information about England, Scotland, Wales and Ireland that does not include any mention of Great Britain. Rules and networks are then combined into a larger knowledge base with a Bayesian probabilistic expert system.

Ford also demonstrates natural language interfaces,ranging from an Eliza-type application with fuzzy matching through to a natural language information retrieval interface based on an augmented transition network. Problem solving is in there too, finding routes through networks, the river-crossing problem, Towers of Hanoi, and game playing in noughts and crosses (tic-tac-toe).

He concludes with three shorter chapters that return to more generalist matters. One is about using JavaScript with NC4 Dynamic HTML, concentrating on cascading style sheets and layers. The second is a useful introduction to server-side JavaScript for providing access to files and databases. The third is a list of URLs for JavaScript tutorials and other resources. There is also a concise JavaScript reference section.

However, Ford's most novel and impressive contribution is in the central chapters on Artificial Intelligence. This is serious programming, but can it really be intended for non-programmers as claimed in the introduction?

At Huddersfield University we actually teach JavaScript as a first programming language for undergraduate Multimedia students, and also as a forerunner to Java for Higher National Diploma computing students, but I doubt we would expect anything like this of them in the first year. I also have doubts that JavaScript is the most appropriate or efficient means of implementing the kinds of application explained in the book. But it's a wonderful introduction to expert systems and artificial intelligence for those who are into this sort of thing.