Research on User Clustering Algorithm Based on Improved ROCK Algorithm
Rong Wang, Aimin Sun · IOP Conference Series Materials Science and Engineering · 2020
Abstract This paper analyzes the merits and demerits of the ROCK algorithm firstly, then indicates that the merits of Rock is easy to cluster categorical database such as Mushroom, and the demerits is that the similarity formula sim of this algorithm depends on the intuition of domain experts. The improved ROCK gets the result of clustering by calculating the similarity using Jaccard coefficient based on the reason which is that the bigger figure of the similarity is, the more alike the object is. Then we can implement personality recommendation according to the result.