A novel collaborative filtering based recommendation system using exponential grasshopper algorithm.

Autor: Singh, Viomesh Kumar, Sabharwal, Sangeeta, Gabrani, Goldie
Zdroj: Evolutionary Intelligence; Apr2023, Vol. 16 Issue 2, p621-631, 11p
Abstrakt: Collaborative filtering recommends items on the principle of similarity. The effectiveness of the similarity is highly dependent on the performance of the clustering algorithm. This paper presents a new clustering-based collaborative filtering method for efficient item recommendation. The new approach introduces a new variant of the grasshopper optimization algorithm to determine optimal similarity patterns. Nine nature-inspired algorithms have been considered and tested against seventeen standard benchmark functions in terms of mean fitness value and the Friedman test to validate the proposed variant experimentally. Further, the efficacy of the new clustering-based collaborative filtering method is tested on the publicly available MovieLens dataset and evaluated in terms of mean absolute error, precision, and recall over the different cluster settings. The obtained results are compared against nine clustering-based collaborative filtering methods. Experiments affirm that the proposed method is a permissive technique for the recommendation on large-scale datasets. [ABSTRACT FROM AUTHOR]
Databáze: Complementary Index
Externí odkaz: