Food safety risk assessment is an important content and key link in food safety management， and scientific rationality and fairness are the key features of a good evaluation system. The traditional arithmetic weighted average method and geometric weighted average method cannot reflect the decisive effect of high-scoring items on the total item， and the fuzzy evaluation method cannot meet the strictly monotonic paradigm requirements. Based on the Euclidian vector modulus length， this paper puts forward the concept of generalized vector modulus length， which is used as a method of score mapping from sub-items to total item. This method can meet the strict monotony paradigm requirement. By selecting different parameter value ranges， not only the risk index but also the safety index can be mapped reasonably. Verification by actual data shows that it can help build an excellent evaluation system. After rigorous mathematical derivation， it is proved that the weighted average method， geometric average method， fuzzy evaluation method， etc. are all examples of this proposed mapping method under special parameters. The generalized vector modulus method is a unified mapping form of all these evaluation methods， and has a wide range of adaptability.