Abstract: It is an important problem that how to ascertain the potential risks of the requirement analysis phases and reduce its opposite forces as low as possible in requirement analysis. We proposed an evaluation model of software requirement analysis risk, which combines Fuzzy Logic with BP neural network based on index system of risk assessment of requirement analysis. At last, the results of experiment demonstrate the feasibility and validity of the approach.

Key words: Requirement analysis, Risk, Fuzzy logic, BP neural network