Spam Filtered Algorithm Based on Knowledge Accumulation
Liang Guobiao · Science Technology and Engineering · 2007
New variation of spam leads to exceed range of training set used in current spam filtering algorithms, and lower their performance of filtering illegal emails. To solve this problem, a new algorithm is provided based on knowledge accumulation. Using the laziness of KNN(K nearest neighbors), it automatically expands the training set with the help of new spam, and then uses KNN to classify the unlabelled emails. Experimental results also show the new algorithm with better performance on filtering spam.