Food safety risk identification based on big data

　　The risk factors in food mainly include chemical factors such as pesticides, veterinary drugs, pollutants, biological factors such as microorganisms, biological toxins, parasites, physical factors and other factors. The use of big data technology to statistically analyze past food safety data is one of the effective ways for food enterprises to identify food safety risks. At present, Food Partner Network has established a big data cloud platform for food safety information services, and this paper mainly introduces the basic principles and methods of food safety risk identification based on big data.
　　1. Food safety big data content for risk identification
　　1.1 Food standard regulations and regulatory developments
　　According to China's food safety law, the formulation and revision of food standard regulations must be based on risk assessment. The formulation and revision of food standards and regulations, and the changes in each edition of food standards and regulations, reflect the adjustment of China's food safety situation, so food standard regulations and regulatory dynamics are the priority factors that enterprises need to pay attention to in food safety risk identification. For example, GB 2762-2017 removes the limit requirement for rare earth elements in tea, indicating that after scientific assessment, rare earth elements in tea are no longer required to control, and rare earths will no longer be the main risk point in tea.
　　1.2 Food safety incidents
　　Due to the development of media and the Internet, especially the rise of mobile Internet and self-media in recent years, food safety incidents will spread rapidly and widely in public opinion, which has a huge impact on the food industry, such as the plastic seaweed rumors in 2017 brought huge losses to the seaweed industry, and the false reports of apple medicine bags in previous years caused Yantai fruit farmers to suffer heavy losses. The accumulation of information on the Internet also provides convenience for the identification of food safety risks, and the use of public opinion monitoring and analysis tools to review the main public opinion and hot public opinion of the food industry in the past few years, and analyze its main trends and popularity, etc., is helpful to find food safety risks that the media and consumers are concerned about, so as to achieve risk identification. For example, using the public opinion monitoring and collection tool independently developed by the Food Partner Network, searching for food safety reports on rice, wheat, corn, soybeans and other grains in recent years, and conducting data statistics according to risk factors such as moldy insects, transgenic organisms, illegal additions, excessive additives, and excessive pollutants, you can identify the categories that are most likely to produce food safety risks in food, as well as the corresponding risk factors, regions and times.
　　1.3 Food safety sampling inspection and early warning notification
　　Since its establishment, the former CFDA has standardized the supervision and sampling of food, and the CFDA and CFDA at all levels have published food safety sampling inspection information on an open and transparent basis. According to the statistics of the sampling inspection and analysis system of the food partner network, since 2015, China has released more than one million pieces of food sampling information, and the statistical analysis of the unqualified data in these sampling results can identify the main reasons for non-conformity in different categories of food, and mainly the areas prone to non-conformity. For example, since 2015, China has sampled more than 12,000 batches of infant formula food, and the main unqualified factor is the unqualified selenium content in the quality index. This reminds infant formula food enterprises to control the content of selenium as a critical control point in production, and strengthen its monitoring to avoid such unqualified appearance.
　　1.4 Food case information
　　The tenfold compensation clause of the food safety law has led to the activation of professional anti-counterfeiters. The SPC established the China Judgment Document Network to collect and sort out the judgment documents of people's courts at all levels, including professional anti-counterfeiting cases related to food safety. Statistical analysis of the categories involved in these cases, the reasons for prosecution, and whether they have won the lawsuit and obtained tenfold compensation can help food companies identify the risk of being professionally cracked down. For example, in recent years, most complaints about olive blends and oils claiming a certain characteristic ingredient but not indicating the specific content have been compensated, which reminds relevant food companies to pay attention to the compliance of food labeling and prevent such risks.
　　1.5 Food administrative penalty information
　　The information on food administrative penalties published by the Food and Drug Administration at all levels is often easily ignored by other enterprises. At present, there are many places where the public penalty information will explain the reasons and legal basis of the punishment in detail, and the summary research and analysis of this information will also help enterprises identify certain risks.
　　2. Steps to use big data for food safety risk identification
　　2.1 Identify the product objects to be analyzed
　　The first is to clarify the object that needs to be analyzed, whether it is a type of product or a specific product, whether to identify all the risks of the product, or to evaluate the level of a specific risk at a specific time or specific place, only by clarifying the object that needs to be analyzed can we collect targeted data in order to accurately identify the risk.
　　2.2 Collect the required basic data
　　Collect the relevant information of the formulation and revision of all standards and regulations at a specific time and in a specific region, food safety incidents, past food safety cases, food sampling inspection early warning notifications, food jurisprudence and administrative punishment information related to the assessment object, with the help of professional search engines or third-party data collection service agencies.　　2.3 Disassemble the data for structural organization
　　The collected data is disassembled to form structured data for easy statistics. For example, for a specific food, the main indicator requirements of the revision of the standard in recent years, the amount of food safety news reported by various risk factors, and the number of unqualified reasons for various non-conformities can be counted to form structured and quantifiable data. Moreover, different data can be weighted according to the different characteristics of different products, so as to establish a risk model more objectively.
　　2.4 Comprehensive data analysis to identify risks
　　On the basis of data structure, comprehensive analysis of data from various sources is carried out to identify the main risks in food. Provide a basis for follow-up risk prevention and control.
　　Based on the above data and steps, we can preliminarily identify possible risks in food and provide a theoretical basis for the formulation of food regulatory control measures and plans. At the same time, based on years of data accumulation, Food Partner Network can provide you with customized risk identification report preparation services.