A similarity metric designed to speed up, using hardware, the recommender systems k-nearest neighbors algorithm

• DOI: 10.1016/j.knosys.2013.06.010
• Publication Date: 2013-07-16T17:30:41Z
• Authors: Jesús Bobadilla; Fernando Ortega; Antonio Hernando; Guillermo Glez-de...
• Keywords: Similarity (geometry); Speedup

A significant number of recommender systems utilize the k-nearest neighbor (kNN) algorithm as the collaborative filtering core. This algorithm is simple; it utilizes updated data and facilitates the explanations of recommendations. Its greatest inconveniences are the amount of execution time that is required and the non-scalable nature of the algorithm. The algorithm is based on the repetitive execution of the selected similarity metric. In this paper, an innovative similarity metric is presented: HwSimilarity. This metric attains high-quality recommendations that are similar to those provided by the best existing metrics and can be processed by employing low-cost hardware circuits. This paper examines the key design concepts and recommendation-quality results of the metric. The hardware design, cost of implementation, and improvements achieved during execution are also explored.