Abstract: Objective To construct a food safety risk evaluation model to evaluate the food safety risk of meat products in Guangzhou based on several data sources,including food safety risk surveillance data in Guangzhou City in 2020,regular and irregular food sampling data from the Bureau of Market Supervision and Inspection,and mined internet opinion data. Methods The indicators were screened using correlation analysis and discriminant analysis,and the weights of each indicator were determined by a combination of hierarchical analysis and entropy method. On this basis,a fuzzy comprehensive evaluation method was employed to make a final assessment for the food safety of meat products in Guangzhou City in 2020. Results In 2020,among 55 meat product samples monitored by the Guangzhou food safety risk surveillance,the detection rate of sodium pentachlorophenol was 43.64%,but the food safety index of agricultural and veterinary drug residues and the target hazard quotient of heavy metals were both less than 1. In the samples from the Guangdong Provincial Market Supervision Administration,the unqualified rate of meat products was 1.85%. Based on monitoring,supervision,and public opinion data,the food safety risk index of meat products in Guangzhou City in 2020 was 0.247,which was at a high risk level. Conclusion The multi-source data-based food safety risk evaluation system constructed in our study will not only solve the current problem of data non-circulation among food safety departments,but also evaluate food safety from multiple dimensions,thus improve the efficiency of food safety risk assessment and provide a scientific decision basis for the identification and targeting of potential food safety risks.

Key words: Food safety, Discriminant analysis, Analytical hierarchy process, Entropy method, Fuzzy comprehensive evaluation method, Risk assessment