Shirin Noakhah | University Technology Malaysia (original) (raw)
Uploads
Papers by Shirin Noakhah
Abstract To achieve the high performance system without any failure, we should provide the high r... more Abstract
To achieve the high performance system without any failure, we should provide the high
reliability level of software. Soft computing models for software reliability prediction suffer
from low accuracy during predicting the number of faults. Moreover, the models have some
problems like no solid mathematical foundation for analysis, being trapped in local minima,
and convergence problem. This paper introduces Imperialist Competitive Algorithm (ICA)
to overcome the weaknesses of previous models and improve the efficiency of training
process of Multi-Layer Perceptron (MLP) neural network. Therefore, the network can
predict the number of faults precisely. The results show that the proposed predicting model
is more efficient than the existing techniques in prediction performance
The World Wide Web has brought an enormous improvement in the lives of people, during the last co... more The World Wide Web has brought an enormous improvement in the lives of people, during the last couple of decades. Nowadays, most companies and businesses exploit Ecommerce to sell their products and services, to discover the market trend and to analyze their competitor's activities. Opinion spams are those fake and untruth opinions which target companies or products to fame or defame them. However, to the best of our knowledge, previous works never considered both behavioural and linguistic features simultaneously. In this paper, we propose an iterative algorithm to detect fake reviews, review spammers and group of spammers at the same time. To accomplish this goal, we propose a new graph structure which considers all the features and entity relationships between reviews and reviewers. Experimental results prove that our algorithm outperforms all the other baseline approaches in terms of accuracy.
Abstract To achieve the high performance system without any failure, we should provide the high r... more Abstract
To achieve the high performance system without any failure, we should provide the high
reliability level of software. Soft computing models for software reliability prediction suffer
from low accuracy during predicting the number of faults. Moreover, the models have some
problems like no solid mathematical foundation for analysis, being trapped in local minima,
and convergence problem. This paper introduces Imperialist Competitive Algorithm (ICA)
to overcome the weaknesses of previous models and improve the efficiency of training
process of Multi-Layer Perceptron (MLP) neural network. Therefore, the network can
predict the number of faults precisely. The results show that the proposed predicting model
is more efficient than the existing techniques in prediction performance
The World Wide Web has brought an enormous improvement in the lives of people, during the last co... more The World Wide Web has brought an enormous improvement in the lives of people, during the last couple of decades. Nowadays, most companies and businesses exploit Ecommerce to sell their products and services, to discover the market trend and to analyze their competitor's activities. Opinion spams are those fake and untruth opinions which target companies or products to fame or defame them. However, to the best of our knowledge, previous works never considered both behavioural and linguistic features simultaneously. In this paper, we propose an iterative algorithm to detect fake reviews, review spammers and group of spammers at the same time. To accomplish this goal, we propose a new graph structure which considers all the features and entity relationships between reviews and reviewers. Experimental results prove that our algorithm outperforms all the other baseline approaches in terms of accuracy.