R is good at computing, not controlling, So For web applications, I suggest you use other language (such as PHP) to handle the data flow while call R for computing.
the problem with CF is that its very hard to extend and needs strong technical support. if you have plan to use web app own then you may proceed but for larger scale you have to spend most of your time to debug and to make things object oriented because CF has weak OO support. I like to tell you that use php or ASP for web application. but if you really want to develop CF than CFEclipse and DreamWeaver are nice editors. beside this I suggest you to download developer version source engine like open blue dragon for your work. all the best.
If you're considering Java as a basis for web applications, then there are a few projects around that will allow you to call R from within Java, one being JRI.
Whatever the language, you will definitely want to consider carefully how you will bring something like this into a web application. User driven actions will likely be comparatively expensive and won't enjoy caching tricks found with storage resources. If the actions in reality make an enumerable set, then you might just consider storing their result and refreshing it periodically.
Thaks everyone for your input and thanks Deepak for your suggestions on cold fusion.
I just did additional research on the area and found that phython can interphase with R. From what I have gathered rpy2 is a simple and efficient access to R from Python (just what Stefan mentioned) - sounds like a winner. However, I do have an additional question / concern: rpy2 seems not to support windows as an OS....Is this true? and if so, what are my options with windows, phython and R?