English to Persian (Farsi) Tech/Engineering Computers (general)

In order to define what items to consider in the prediction step we apply a defuzzification method. In particular, every item is assigned to k clusters with different degrees of membership. Prediction is evaluated keeping into account t1 1 items, which have the highest degree of membership at the cluster 1, t2 items, which have the highest degree of membership at cluster 2, and so on. t1,t2. . . . tk are chosen in a proportional way, e.g., if k = 3 and t = 10 and membership of item i is (0.7,0.2,0.1) we consider top-7 items from cluster 1, top-2 items from cluster 2 and top-1 item from cluster 3.