A Study of Recommender Techniques Within the Field of Collaborative Filtering

Recommender systems can be seen everywheretoday, having endless possibilities of implementation. However, operating inthe background, they can easily be passed without notice. Essentially, recommendersystems are algorithms that generate predictions by operating on a certain dataset. Each case of recommendation is environment sensitive and dependent on thecondition of the data at hand. Consequently, it is difficult to foresee whichmethod, or combination of methods, to apply in a particular situation forobtaining desired results. The area of recommender systems that this thesis isdelimited to is Collaborative filtering (CF) and can be split up into threedifferent categories, namely memory based, model based and hybrid algorithms.This thesis implements a CF algorithm for each of these categories and setsfocus on comparing their prediction accuracy and their dependency on the amountof available training data (i.e. as a function of sparsity). The results showthat the model based algorithm clearly performs better than the memory based,both in terms of overall accuracy and sparsity dependency. With an increasingsparsity level, the problem of having users without any ratings is encountered,which greatly impacts the accuracy for the memory based algorithm. A hybridbetween these algorithms resulted in a better accuracy than the model basedalgorithm itself but with an insignificant improvement