similar background of member

Hi,
I need to build a web system which can find out the member with similar profile/buying behaviour.
I have searched the web but it seem that it belongs to the field of data mining and the stuff are quite complex. Some datamining ebook have provided a lot of algorithm that confuse me so much!
Any idea for me to start? Should i find some tut to read first?
Thx

Data mining will be complex. No two ways about that.
What you should do is step back and ask yourself, "What would be the easiest way to achieve this?" A startingly easy way is to simplyask the members.
To take an example that we've all seen through online ads, think about a dating service. They ask the members to define their own category by selecting from a list: man seeking woman, woman seeking man, and then some more interesting combinations. When signing up, they're sure to ask the member whether they prefer smokers or non-smokers, drinkers or non-drinkers, and so on.
This sort of knowledge would be extremely hard to learn through data mining, but extremely easy to learn by asking the members. And asking questions of members is a common step in the sign-up process, so there's no reason to avoid this step.
So that's my idea as a way to start. Think about the easiest way to achieve your goals. Don't take the hard route if there's an easier way.