[Retracted] Combining Latent Factor Model for Dynamic Recommendations in Community Question Answering Forums
Algorithm 3
Incremental learning.
Input: A feature vector v′ from user latent profile lp′ Output: Output matrix O′ representing similar users in similar groups
Procedure of K-Means:
(1)
First, calculate the means of the existing clusters and apply k-means clustering algorithm to cluster the new data points: .
(2)
Calculate the mean of each cluster ci from in the existing clusters O, where distance di between new data point and the mean of that cluster is minimum using and .
(3)
If the dij is minimum, then make that point the part of that cluster cj and update the centroid by recomputing the mean of that cluster as .
(4)
If the distance is not found to be minimum, then make a new cluster of that point .
(5)
Recompute the cluster center to get the cluster with the minimum distance. Objective function is defined as .
(6)
If the distance dc between any two clusters is below than threshold, merge the clusters.
