Classification of Twitter users into 4 different categories

_Mahn_

Hi,

I'm a newbie to Rapidminer. I really appreciate it if anyone can help me resolve the question.

I have a dataset with two attributes: Twitter usernames and bios. I manually sorted 1000 usernames into four categories according to the bios. I would like to find a good way to classify these usernames in Rapidminer. But there is no algorithm that has high accuracy. 

What could I do with my data to improve the accuracy? I want Rapidminer to help me to classify the remaining usernames.

Thanks for your help!

Best regards,
Mahn