There have been many different algorithms proposed in the past to improve the predictionsgenerated regarding the ratings given to items. As we learned in this thesis, the algorithmsproposed so far would mostly form one of the three categories of recommendation systems, viz,Content-based filtering recommendation systems; Collaborative Filtering basedrecommendation systems or Knowledge-based recommendation systems. We also learned thatof the three types of recommendation systems, except for certain specific contexts, the secondkind mentioned above proves to show better results. A common algorithm that we set out toimprove is based on Collaborative Filtering technique and is a neighborhood model basedimplementation. This algorithm, as we examined in detail in chapter 2, is built on Pearson‟scorrelation coefficient. A user‟s potential rating for a movie will be determined based upon theratings of all other movies by those users whose ratings have a strong correlation with themovie in question. It is seen to work well but there was obvious scope of improvement whichbecame the motivation to develop the ClusCor algorithm as discussed