TY  - JOUR
T1  - Improved Efficiency of Social Recommender Systems Using ANN and GA
AU - Hosseini, Seyed Mohammad AU - Safaei, Ali Asghar 
JO  - Journal of Engineering and Applied Sciences
VL  - 11
IS  - 10
SP  - 2213
EP  - 2221
PY  - 2016
DA  - 2001/08/19
SN  - 1816-949x
DO  - jeasci.2016.2213.2221
UR  - https://makhillpublications.co/view-article.php?doi=jeasci.2016.2213.2221
KW  - Social recommender systems
KW  -artificial neural network
KW  -implicit relationships
KW  -feature selection
KW  -genetic algorithm
AB  - Currently, the rapid growth of internet and huge volume of informationrequire systems which are able to recommend the most appropriate products and services to the user. Using data mining tools, recommender systems can make appropriate suggestions to choose from large amounts of data. Traditional recommender systems, particularly collaborative filtering,used similarity criteria to selectsimilar neighbors for the active user and made suggestions based on estimate of their feedbacks. Similarity criteria considerably influence performance of these systems. Therefore, it is challenging to select the proper criteria. The main objective of this study is to develop a SRS which can make appropriate suggestions by feedbacks of other people and trust to them without having to find similar users. ANNs are known as one of the common methods used in thesesystems to explore relationships and trust. In fact, ANNs explore implicit relationships (trust) between trusted target users to make accurate suggestions. To increase efficiency of ANN in the suggested method, GA is used as feature selection method to find optimal set of features. By implementing and comparingthis hybrid algorithm with other similar algorithms, the results indicate error reduction (mean absolute error and root mean square error) in making suggestions to users.
ER  -