4 ms·
Document Classification for Newspaper Articles (2009) [pdf]
- PaulHoule 12y agoTraining sets too small, film at 11.
- mark_l_watson 12y agoI was not surprised that they didn't use larger training sets. They are comparing three different approaches, one of which maximum entropy they had trouble running without reducing the number of features used.
- PaulHoule 12y agoWith the training set sizes they are using it is no surprise they did not get good results. They could have done better with them had they used some kind of dimensional reduction. However, just from seeing a lot of systems that wind up on the wrong side of the "commercialization valley of death" I wouldn't start this kind of project unless I had a bigger training set and some latitude to adjust the categories to produce something more learnable.
- beshrkayali 12y agoI don't know a lot about this topic as I started getting into it recently, but I think the purpose of this is more of a study than an actual solution ready for commercialization.