Evolutionary neural networks for improving the prediction performance of recommender systems

Seref B., BOSTANCI G. E., GÜZEL M. S.

TURKISH JOURNAL OF ELECTRICAL ENGINEERING AND COMPUTER SCIENCES, cilt.29, sa.1, ss.62-77, 2021 (SCI-Expanded) identifier identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 29 Sayı: 1
  • Basım Tarihi: 2021
  • Doi Numarası: 10.3906/elk-2004-138
  • Dergi Adı: TURKISH JOURNAL OF ELECTRICAL ENGINEERING AND COMPUTER SCIENCES
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Applied Science & Technology Source, Compendex, Computer & Applied Sciences, INSPEC, TR DİZİN (ULAKBİM)
  • Sayfa Sayıları: ss.62-77
  • Anahtar Kelimeler: Recommender systems, prediction, neural networks, genetic algorithms
  • Ankara Üniversitesi Adresli: Evet

Özet

Recommender systems provide recommendations to users using background data such as ratings of users about items and features of items. These systems are used in several areas such as e-commerce, news websites, and article websites. By using recommender systems, customers are provided with relevant data as soon as possible and are able to make good decisions. There are more studies about recommender systems and improving their performance. In this study, prediction performances of neural networks are evaluated and their performances are improved using genetic algorithms. Performances obtained in this study are compared with those of other studies. After that, superiority of this study is shown. While multilayer perceptron, generalized feed-forward network, and coactive neuro fuzzy inference systems were used as neural network algorithms, Movielens 100K and Movielens 1M datasets, which are widely preferred in recommender system studies, were used to train and test the system in the present study. Mean square error and root mean square error were employed as performance metrics. As a result, it was observed that genetic algorithm improves performance of neural networks, and prediction performance of hybrid combination of neural networks and genetic algorithm is superior to prediction performance of recommender systems available in the literature.