Research And Implementation Of Software Next Release Problem Based On Search Technology

With the rapid development of science and technology and coding technology,the development trend of software industry has become irresistible.The role of requirement optimization selection in the software development process has become more and more important.Because the users' requirements present the trends of diversity and variability,the scale and complexity of software development have also been greatly improved,which causes the best set of user requirements is difficult to be accurately obtained,and it is also the primary prerequisite for completing the software system development successfully.When multiple users are involved in a software development project,each user has a different requirement for software.With the increasing users' requirements,how to select a set of the best combination of requirements to meet multiple users as much as possible,which has become a difficulty to be solved first by software developers.This kind of problem is called the software next release problem(NRP),and it is a typical NP-hard optimization problem.Therefore,it is important to select and improve the intelligent optimization algorithm to solve the the problem of requirement optimization selection,which has important theoretical significance and application prospect.In this thesis,the software next release problem is studied by us,decompose the problem into two kinds with users' requirements having dependency relationship or no,and build the corresponding problem model.The dragonfly algorithm is used as the main optimization algorithm to solve the problem,and a series of improvements are carried out on the basis of this algorithm to improve the accuracy and efficiency of optimization.The following work has been completed:(1)In order to sovle the NRP problem without dependence between users,we improve dragonfly algorithm by adding dynamic parameter setting and random walk strategy,and present a Improved Discrete Dragonfly Algorithm to solve this problem,the simulation results show that the proposed Improved Discrete Dragonfly Algorithm has some advantages in solving this problem.(2)For the NRP problem with dependence between users,we improve the Dragonfly Algorithm with adding neighborhood search and local search based on Artificial Bee Colony Algorithm,and present a fusion algorithm,which has good convergence speed and global optimization ability.The results show that the proposed fusion algorithm in this thesis is more effective in solving the NRP problem.(3)Based on the Matlab platform,we design next release problem requirement optimization verification system on the basis of the above experimental research results,and carry out the demonstrating experiments.