Authors:
Diogo Silva
1
;
2
;
Davi Silva da Cruz
1
;
Diego Corrêa da Silva
2
;
João Dias de Almeida
2
and
Frederico Durão
2
Affiliations:
1
Federal Institute of Maranhão - IFMA, Coelho Neto, Brazil
;
2
Department of Computing, Federal University of Bahia - UFBA, Salvador, Brazil
Keyword(s):
Recommender Systems, Clustering, Markov Chains, Assessment Questionnaire, Long-Tail.
Abstract:
The primary goal of this paper is to develop recommendation models that guide users to niche but highly relevant items in the long tail. Two major clustering techniques and representing matrices through graphs are explored for this. The first technique adopts Markov chains to calculate similarities of the nodes of a user-item graph. The second technique applies clustering to the set of items in a dataset. The results show that it is possible to improve the accuracy of the recommendations even by focusing on less popular items, in this case, niche products that form the long tail. The recall in some cases improved by about 27.9%, while the popularity of recommended items has declined. In addition, the recommendations to contain more diversified items indicate better exploitation of the long tail. Finally, an online experiment was conducted using an evaluation questionnaire with the employees of the HomeCenter store, providing the dataset. The aim is to analyze the performance of the p
roposed algorithms directly with the users. The results showed that the evaluators preferred the proposed algorithms, demonstrating the proposed approaches’ effectiveness.
(More)