From cslt Wiki
Revision as of 02:11, 30 September 2013 by Zhaomy
do something to improve the performance of huilan's classifier: 1.remove stopwords; 2.improve the algorithm of computing cos distance(speed up); 3.retain 2-best classes(eliminate class error).
Target for next week:
modify the training script: make a index table [word,[filename,tfidf], ... , ]