hi all,
i am doing a project where in i get an input from a consulting company containing the feedback on different representatives in the concern. i would like to create a program that can segregate these comments automatically into different buckets depending upon the keywords in the comments.

for eg:

the rep. was very good - should be bucketed into "GOOD"
the rep. was very bad - should be bucketed into "BAD - needs improvement"
and something like
the rep. was good but was not able to help me - into ?

could someone help me out with this problem?

Thanks a lot in advance....

you could easily implement a naive bayes classifier. You will have to train it with a sampel set of data first. It will need a rather large (arbitrary) set of data to be trained on for each category it is going to classify. Your sets of data would be previously submitted comments. You would have to manually tell classify it at first and run tests periodically to see if it is at an acceptable level of accuracy.