Elsevier

Food Control

Volume 99, May 2019, Pages 48-56
Food Control

An improved ISM method based on GRA for hierarchical analyzing the influencing factors of food safety

https://doi.org/10.1016/j.foodcont.2018.12.020Get rights and content

Highlights

  • An improved GRA-ISM method is proposed.

  • The multi-hierarchical structure model of different factors affecting food safety is established.

  • The validity and the practicability of the proposed method are verified by food safety inspection data.

  • The proposed method can guide relevant departments to strengthen supervision and urge enterprises to work safely.

Abstract

Food safety is closely related to economic development and daily life of people. The valid food safety and risk warning contribute to the social health and sustainable development. However, the inspection data of food safety is characterized by high complexity, high dimension and non-linearity. Therefore, an improved interpretative structural modeling (ISM) method based on the grey relational analysis (GRA) (GRA-ISM) is proposed to hierarchical analyze influencing factors of food safety. The correlation coefficient between the influencing factors is calculated by the GRA. Then the ISM is used for stratifying and establishing the multi-hierarchical structure of influencing factors of food safety. Finally, the infant formula data and the sterilized milk data of food safety in China are hierarchical analyzed by the GRA-ISM. The multi-hierarchical structure model of different factors affecting food safety can be obtained. The Student's t-test (t-test) is used to verify the validity of the threshold selection and the result of the GRA-ISM. Meanwhile, through the early warning analysis of the major factors, the proposed method can guide relevant departments to strengthen supervision and urge enterprises to work safely.

Introduction

Food safety affects people's health, which has profound social significance. In recent years, food safety accidents have happened frequently all around the world (Chiou et al., 2015, Lam et al., 2013). The outbreak of toxic eggs in Europe in 2017 has been reported in nearly 20 countries and regions, including the Netherlands, Belgium and Germany (Liu, 2017). And a series of food safety incidents have happened in China, such as Sudan red (Liu, Hei, He, & Li, 2011), melamine (Xiu & Klein, 2010) and gutter oil (Lu & Wu, 2014). The accidents happened have hampered social stability and economic development. The national food unqualified situation report of 2012–2017 is issued by the China State Bureau of Technical Supervision as shown in Fig. 1, which shows that the number and proportion of unqualified food batches in China are on the rise from 2012 to 2017 with a small fluctuation. Hence, it is critical to find the major factors affecting food safety and food safety problems in advance.

By analyzing the influencing factors of food safety, the source control of major factors can be realized. Consider the grim situation of dairy food safety in China (Yan, 2012), a new interpretative structural modeling (ISM) method based on the grey rational analysis (GRA) (GRA-ISM) for hierarchical analyzing the influencing factor of food safety is proposed in this paper. Firstly, the correlation coefficient between influencing factors is calculated by the GRA, and the strength and weakness of the mutual relations between the factors is quantified simultaneously. Then the ISM is used to stratify the influencing factors of food safety, and the main influencing factors are obtained. Finally, the infant formula data and the sterilized milk data of food safety in China are hierarchical analyzed by the GRA-ISM. Though the Student's t-test (t-test), the validity of the threshold selection and the result of the GRA-ISM is verified. Meanwhile, the experimental results show that the GRA-ISM can analyze the factors affecting food safety better.

The organizational structure of this paper is as follows. The current research of the ISM method and the GRA are introduced in section2. Section 3 is the detailed description of the GRA-ISM method. Section 4 is the case study for hierarchical analyzing the influencing factors of the infant formula and the sterilized milk. Section 5 is the discussion of the experiment results and the proposed method. Finally, the conclusion is given in Section 6.

Section snippets

Relate work

Due to food safety incidents occur frequently, food safety research has become the focus of the global research (Perrot, Trelea, Baudrit, Trystram, & Bourgine, 2011). At present, the Artificial Neural Network (ANN) (Geng, Shang, Han, & Zhong, 2019) and the Bayesian Network (BN) (Greco, Landoni, Biondi-Zoccai, D'Ascenzo, & Zangrillo, 2016) are the most mature research methods on food safety. The ANN is an effective computing model and widely used in classification of food species (da Silva,

The grey relational analysis (GRA)

Correlation coefficient can measure the degree of linear correlation between two random factors (Puth, Neuhäuser, & Ruxton, 2014). Generally, the more consistent the change tendency of the two variables, the higher the degree of correlation between the two variables. On the contrary, the degree is low. The correlation coefficient matrix is established by using the GRA.

Let the reference sequence be x0  = { x0 (1), x0 (2), …,x0(n)}, where n is the number of samples, and the comparison sequence is

Case study: the influencing factors analysis of food safety

This paper analyzes the influencing factors of food safety. The used data is from a food inspection agency in a province in China. Firstly, the GRA is used to establish the correlation coefficient matrix of each factor. Then the ISM is used to stratify the influencing factors, and the multilevel hierarchical structure of the factors affecting the food safety is established. The inspection data of infant formula and sterilized milk are used in this paper.

Discussion

First, The GRA-ISM is proposed. The correlation coefficient matrix of the non-linear data is calculated by using the GRA. Then major factors affecting food safety can be found based on the ISM. By focusing on major factors, the proposed method can guide relevant departments to strengthen supervision and urge enterprises to work safely.

Second, this proposed method is used for effectively analyzing the major factors affecting the safety of infant formula and sterilized milk. The t-test is used to

Conclusion

The improved GRA-ISM method for hierarchical analyzing the influencing factor of food safety is proposed in this paper. Taking the inspection data of infant formula and sterilized milk as examples, the correlation coefficient matrix of the factors is calculated by using the GRA. And then the ISM method is used to stratify the influencing factors and establish the multi-hierarchical structure of the influencing factors of infant formula and sterilized milk. The t-test is used to verify the

Acknowledgement

This work is partly financial supported by National Key Research and Development Program of China (2017YFC1601800), National Natural Science Foundation of China (61673046 and 61374166), and Fundamental Research Funds for the Central Universities (XK1802-4).

References (35)

Cited by (43)

  • Automated food safety early warning system in the dairy supply chain using machine learning

    2022, Food Control
    Citation Excerpt :

    An analysis of food safety incidents concluded that a wider recognition of the environment in which food is being produced is needed to ensure that a similar incident will not occur in the future (Costa et al., 2017; Kleter & Marvin, 2009; Maeda et al., 2005; Marvin & Kleter, 2014; Marvin et al., 2009). Approaches and procedures towards such proactive systems have been developed, including systems for the early detection of food safety issues or food fraud in (social) media using text mining (Kate et al., 2014) and ontologies (Luijckx et al., 2016; Van de Brug et al., 2014), system approaches to integrate data of impacting drivers of change and food safety monitoring data using Bayesian networks (BNs) (H.J. Marvin & Bouzembrak, 2020a; H.J.P. Marvin & Bouzembrak, 2020b; Marvin et al., 2016, 2020; Bouzembrak & Marvin, 2019), artificial neural networks (ANNs) (Lin et al., 2019), anticipation systems for food safety issues using autoregression-based screening tools (Verhaelen et al., 2018), support vector machines (SVMs) (Zhang, 2020), and unsupervised machine learning applied on a media corpus (i.e., the Europe Media Monitor Medical Information System) to rapidly detect specific food fraud incidents in the media (Rortais, Barrucci, Ercolano, Linge, Anna, et al., 2021a, 2021b). Although these systems have shown great potential, successful future warning of food safety risk has been difficult to demonstrate.

  • Early warning and control of food safety risk using an improved AHC-RBF neural network integrating AHP-EW

    2021, Journal of Food Engineering
    Citation Excerpt :

    Specifically, Han et al. (2019a) developed a multiple grey relational model to forecast the food quality safety. Lin et al. (2019) adopted a grey relational analysis model integrating the interpretative structural modeling to analyze the influencing factors of food safety. Bouzembak et al. established a Bayesian network model to analyze and forecast chemical hazards of food safety risks (Bouzembrak and Marvin, 2019) and the food fraud type (Bouzembrak and Marvin, 2016).

  • Debottlenecking cogeneration systems under process variations: Multi-dimensional bottleneck tree analysis with neural network ensemble

    2021, Energy
    Citation Excerpt :

    Grey relational analysis (GRA), on the other hand, has been used to address complex problems that contain incomplete information and the convoluted interrelationship between a given list of criteria [61]. It is one of the most widely used grey system theory [62], which has extended its utility to various fields such as the manufacturing industry [63], logistics companies [64], food security [65], etc. Recently, integration between TOPSIS and GRA (or GRA-based TOPSIS) have been proposed to overcome the accuracy issues on the approximation of the geometric relationship between a given solution to the respective PIS and NIS [38].

View all citing articles on Scopus
View full text