Research And Development Of Rapid Vote-Counting System Based On Software Product Family

As a new vote-counting way, the rapid vote-counting system, which combines paper ballots and digital image processing technology perfectly, can solve the problems that the traditional manual vote-counting way has, containing low correctness and poor timeliness. However, the changing market requirement cannot be quickly responded by the rapid vote-counting system based on the traditional software development mode. And the shortcomings are as followings:low development efficiency, long development lifecycle, hard to guarantee the quality, poor maintainability and practicality.In order to solve the above problems, the idea of software product family is introduced to guide the development of rapid vote-counting system. The major contributions of this paper are as follows:1) The modeling approach of rapid vote-counting software product family is researchedThe workflow of rapid vote-counting software and the detailed description of its key technology are given. And then the domain scope of rapid vote-counting software is defined. Finally, the improved feature and use-case integration variability modeling approach is applied to the domain of rapid vote-counting system, by which the domain use case model and feature model are obtained.2) The framework of rapid vote-counting software product family is designedFirstly, the base objects of rapid vote-counting domain are extracted, and a framework design approach of software product family is proposed based on base objects. Secondly, the proposed approach is applied to the rapid vote-counting domain to obtain its framework, which is realized with the design pattern. Finally, the software configuration process of rapid vote-counting domain is proposed. 3) The module planning of software product family based on function requirement vector is proposedBased on text clustering technology, a module planning of software product family based on function requirement vector is proposed. Firstly, the concept of function requirement vector is given and the construction process of function requirement vector is improved. Secondly, based on the given metrics of the vector’s similarity and the clustering quality factor, the clustering analysis of function requirement vector is carried out. Thirdly, through the constructed module extraction matrix, the common and special modules of software product family can be extracted. Finally, the approach is applied to rapid vote-counting domain to prove its feasibility and effectiveness.Finally, with the reusable assets produced in domain project, the application practice of rapid vote-counting software product family is conducted, which proves the advantages of the software development method base on software product family, improving the software quality, shortening the software lifecycle and reducing development costs.