Research topic for Data Mining+Collaborative Filtering+NLP

I am thinking of MS as a safety boat in case the ship that I am sailing in sinks :)
Moreover the pattern of MS is also gonna change.. So I feel that it is always better to have the score in hand.

I am planning to do an MS in computer science

I havent done any major projects in computer science..
To get into the top colleges [with aid] I feel that they would be looking at my inclination towards research..
I am planning of doing some project this year [and publish a paper.. maybe]

I am thinking of doing a project in collaborative filtering (involving NLP, Datamining)

Looking for some research topics..
Literature survey documented below

Searching in
ACM library, Google, Google Scholar, and a collection of search tools by Todd Veldhuizen.

Need an ACM account.. will finish the topic selection by sunday
[update - 5th Jan]
hmm.. friends say that I am becoming unstable day by day.. some even coined a noun for me - lahari baba.. the fickle one.. and here is another case of this behaviour..
I am thinking of a topic that involves [Artificial Intelligence - > [ANN or Genetic Algorithms] and design optimisation]..
One topic that I thought of is "Multi objective design optimisation of Steam turbine bucket and wheel using genetic algorithms" (This topic is research oriented, aligned with my work at office and also requires less effort)..
Now I need to look up what genetic algorithms are in detail.. I have a fair idea about them.. They operate on the principle of natural selection.. picking up the best qualities of the previous generation, generating new offspring with these qualities and then replacing the least fittest of the current population with the newly generated offspring...
I need to look at how I need to plug in the GA into the already existing program structure..