Multi aspects based requirements prioritization for large scale software using deep neural lagrange multiplier

In multi aspect based software (MABS) aspects like business values, benefits, cost, time, penalty and risk on business are evaluated. Requirement prioritization (RP) is contemplated as a portion of Requirements Engineering and is pivotal decision-making activity. In existing RP techniques, MABS aspects are not taken into consideration. The objective of this paper is to develop a new RP method called, Deep Neural Lagrange Multipler-based Multi-aspect Large Scale Software Requirement Prioritization (DLM-MLSRP). The method consists of four different layers, i.e., one input layer, two hidden layers and one output layer. The requirement specification acquired from the customer forms as input to the input layer. The first hidden layer performs requirement selection via Criteria Hypothesis formulation. The second hidden layer performs Pair-wise Assessment by means of Lagrange Multipler Eigen-based function. Finally, the requirement prioritization matrix forms the output layer. The performance of our method is evaluated based on four paramters, first one is RP time, time efficiency of DLM-MLSRP method was found to be 24% and 36% better than that of SRP Tackle and IFS respectively. Second parameter is RP accuracy and our method shows 98.33% accuracy as compared with 96.6% and 93.33% of other two methods, third paprameter is sensitivity and our method shows 0.88 compared to 0.85 and 0.81 of other two methods. The final parameter that we consider is specificity and our results show improvement of specificity of DLM-MLSRP method by 8% and 20% compared to other two methods.