6 ms·
I feel naive bayes works pretty well with such text classification tasks. You might want to give it a try.
by priyankt 10y ago
I feel naive bayes works pretty well with such text classification tasks. You might want to give it a try.
- visarga 10y agoVowpal Wabbit is also pretty easy to use and fast. I train and test on 100K text examples in under 1 minute. Works better than random forests and other things.
- denzil_correa 10y agoThe choice of algorithm completely depends on the type of classification task at hand. Naive Bayes would be good for certain types of classifications problems but there could be better ones for another type of text classification problems.
- autokad 10y agonaive bayes does well on small data sets, but would in general do poorly on larger text sets due to its independence assumption (which is horribly wrong in language).