investigati n has been c nducted n tw well kn wn similarity-based learning appr aches t text categ rizati n. This includes the k-nearest nei hbor (kNN) classifier and the Rocchio classifier. After identifyin the weakness and stren th of each technique, we

! " ! # " !

$ % % && # '() *+ $ , "

- . . . / 0 ! . . 1

+ 2! " ! # !

! ') $$ "

- 3. . / 04 &. . 1

3 ! ! & % % 1 % ! #5& ! ! 6 . ' ! ! ! & 7 5$$8 !! 9 !! . 3 # % 1 !! ! 4 % ! % !! $$ 5& ! !! &# # ! ! 5$$ 9 !! ! ! 6 & !.

3 6 # ! !#! ! &

% # *5 %! 3 ! 9 !5 :); %! ! !. ' ! ! ! % $$ 5& ! ! 5$$ 9 !! .

' 6 7' 8 ! !1 !! & ! . ' ! 6 !! ! # ! ! ! !! < =. ' # ! #6 6 # &# ! & ! . 3 ! ! & ! # ! !1. ' !! 6 6 ! ! 4 . ' 6 ! ! 6 ! ! & ! !! # ! #.

' # 6 ! ! !! ! 6 !1 6 . ' ! ! $ > # ! 7$ 8 & & ! !! ! < = ! ' !! ! <(= ! 9 ! <?= !! ! <:= $ $ % 1 <@= 5$$ !! ! <: )= A 7 A !8 <; B= 9 !! ! < * = . , # ! ! ! % & ! # # ! ! ! ! 1 # & ! . & ! ! ! ! ! # 6 < =.