*2.3. Media Data*

Social media has become the main way for users to obtain and share food information [27,28]. According to a statistical report on China's Internet Network Information Center, the number of Internet users in China has reached 854 million, and 88.8% [29]. Social networks have gradually become the mainstream platform for disseminating information, a constant strem of videos, news, and other types of data [30]. Purcell et al. [31] found that two-thirds of Internet users ge<sup>t</sup> their news from Facebook and share news through social media. Through research into the generation, and promotion of social events on the Internet, the mode and characteristics of information transmission can be discovered, which provides support for practical application scenarios. In 2009, Google successfully predicted the spread of the H1N1 virus based on query data in its search engine and brought the public valuable time to prevent an outbreak [32]. Combining the real-time advantage of big data with the conventional and available advantages of traditional data will enable effective response to the transmission of public health events such as COVID-19 [33]. Singh et al. discovered supply chain managemen<sup>t</sup> problems by using Twitter data to improve supply chain managemen<sup>t</sup> in food industry [34].

The public participates in the discussion of events by different media, and it expresses clear opinions and attitudes in the form of public opinion [35]. The report (https:// baijiahao. baidu.com/s?id=1617643364060321280&wfr=spider&for=pc (accessed on 2 September 2021)) shows that food safety and food rumors were first among hot food events in 2018, and the topic has become one of the prime targerts for media rumors, and social media's intensification of rumors can create a widespread crisis [35]. By analyzing and understanding the trend of food incidents based on social media data, regulators can formulate timely countermeasures like enhancing public awareness through science education and shaping public opinion [36–38]. However, the field of media data has its challenges that need to be overcome.

• **Multi-source heterogeneous data fusion.** Media data on Twitter, Facebook, YouTube and various information portals have complex formats and multiple sources, and there is a lack of technology to identify relevant data in one sources and link it to others. In the absence of a "fusion technology" for multi-source heterogeneous data further study is urgently needed to address social media rumors and their negative impact on public security.


### **3. Application of Big Data in Food Industry**

The food supply chain is from farm to table, where the main links are planting and breeding, storage, processing, circulation (transportation) and consumption [39]. The discovery of value information on original data needs to go through a continuous cycle: of "discrete data—integrated data—knowledge understanding—mechanism extraction— application effect analysis", from which the potential value of data sets can be mined. The processing system of big data application in food industry is shown in Figure 3. It is composed of five modules: big data collection of food industry, big data processing and fusion, big data mining and analysis, big data view and big data security. Each module is closely connected, and its functions are briefly described as follows.


important research topic. This module provides security technical support for all big data processing to ensure the safety, reliability, and controllability of data.

### *3.1. Social Co-Governance in the Food Industry*

Social co-governance in the food industry provides a feasible solution to the issues of food security and food quality by using public wisdom [52]. Social co-governance is usually based on crowdsourcing to cooperate with consumers or experts to create value [53]. The failure rate of new food product development exceeds 40%, and the failure of new products usually affects the continued operation of small and medium-sized enterprises [3]. Large food companies have tried to collect consumer preference data through crowdsourcing, and have decided the direction of product development based on an analysis of big data. The Danone company encouraged consumers to vote for a creamy dessert flavor, and the 400,000 participants in 2006 more than doubled to 900,000 in 2011. Lay's used the wisdom of crowds to develop more than 245,825 flavors of potato chip [54]. Procter & Gamble, Starbucks and Unilever sought better product design based on collective intelligence [55]. Employees often have a wealth of heterogeneous expertise, and companies can gain insight from their workforce to help improve economic performance. In addition, crowdfunding is another form of social co-governance, sharing business risks and alleviating capital pressure through mutual assistance [56]. Social co-governance has grea<sup>t</sup> potential for food security. Combining the mobile data of consumer groups with food shelf life, the intelligent control of food inventory can be realized to prevent food spoilage and waste [57]. Social co-governance can also be applied to monitoring foodborne diseases [32], identfying contaminated products, reducing the risk rate of food rumors and enhancing food safety [58].

**Figure 3.** The processing model of big data in food.

Although social co-governance can enable food enterprises to obtain consumer demand information through diversified channels, it is still difficult to obtain effective information in time due to the limitation of enterprise resources [53]. In addition, there is a lack of an incentive or fair evaluation method in the food industry to convince consumers to participate.

### *3.2. Exploit Consumption Markets*

There is a huge amount of food-related data both inside and outside the food supply chain, and the collection and analysis can promote enterprises to expand their markets [59]: (1) By collecting commodity and retail information for analysis, they can appraise the market situation, grasp the business dynamics of their competitors, and define the market positioning of products, thereby grasping market opportunity. (2) Collecting consumer information (purchase lists and channels, commodity preferences, usage cycle, family information, working condition, values) will establish a customer database that can give enterprises portraits of their customers that reveal their preferences, consumption tendency, value orientation and commodity reputation. With this information, enterprises can develop efficient marketing strategies and develop trust, so they can continue to compete effectively. (3) Data clustering analysis of consumers' food evaluations (advantages and disadvantages, quality, nutritional value) from social platforms such as Facebook, Twitter and Sina Weibo, allows enterprises to anticipate potential problems and optimize the quality of goods and services.
