I'm interested in learning some AI algorithms that have a practical use in web applications eg. search, product recommendations etc. Obviously since I'm asking this question I am look for some more entry level material.
Any sort of useful stuff on the subject is good - books, blogs, tutorials, anything. My language of choice is c# so anything in that would be awesome but I'm happy to look at examples in other languages.
The book Algorithms of the Intelligent Web by Haralambos Marmanis & Dmitry Babenko (Manning ed. 2009) provide a primer into this kind of things.
This book covers a rather broad spectrum of areas where "smarts" can be applied to web applications. Because of this, many topics receive at a rather superficial treatment, BUT each chapter includes very relevant references for digging further.
Also, code-wise, the working examples found in the book are made available in a code.google.com project (sorry however, mostly java, not C# as suggested in the question)
I cannot think of other AI resources specifically focused on Web applications, but many areas of AI-at-large are relevant the "intelligent web", in particular:
I use AForge.NET in several projects. There are quite simple and fast implementations of genetic algorithms, neural networks, machine learning and so on.
It is an open-source project, so I can recommend it as a starting framework for any project using AI.
Toby Segaran's Programming Collective Intelligence isn't strictly an AI book I guess, but it does cover the kind of topics you're interested in (e.g. product recommendations, predictions, price modelling), and as a newcomer to the field I found it pretty accessible. It does sometimes skip over algorithmic detail in favour of "here's how to use this technique via an existing library" though; also the code samples are written in Python though it's clearly explained and easily translated to C#. It has a strong Web focus because, as the title implies, it deals with mining data from user-provided content, particularly on external sites such as eBay, etc.
There is an excellent series of Google talks that will cover the foundations called Statistical Aspects of Data Mining. It's the same graduate level class that David Meese gives at Stanford.