Food Supply Chain Risk Perception, Identification, and Rating System Based on Knowledge Graph

Abstract

The present invention discloses a food supply chain risk perception, identification, and rating system based on a knowledge graph, including a food supply chain risk perception module, a risk identification module, a risk rating module, and a response and handling module. Its characteristics are as follows: The food supply chain risk identification module encompasses all classification names of the same food in different food supply chain links; first, it locates the food supply chain level where the currently sampled food name is located, and then it performs fuzzy matching for the classification name and limit value of the currently sampled food name for this food supply chain level. The food supply chain risk rating module's rating for the likelihood and severity of risk occurrence is based on the likelihood and severity of risk occurrence that vary with time and space. The present invention solves the problem that most of the currently constructed food safety knowledge graphs are based on the food classification standards of the industry, without considering the inconsistent classification of the same food in the food supply chain links, resulting in a broken chain in food safety risk identification.

Description

Technical Field

[0001] The present invention relates to the field of food safety, and particularly to a food supply chain risk perception, identification, and rating system based on a knowledge graph. Background Art

[0002] Generally, food safety risk management involves four steps: risk perception, risk identification, risk rating, and risk control.

[0003] In the risk identification step, the food classification is described differently in different links of the food supply chain. For example, for the classification of vegetables, in the planting and breeding link, it is classified according to the edibility of vegetables, and vegetables are divided into 11 categories such as melons, bulb onions (onions and garlic), leafy greens, legumes, solanaceous fruits, aquatic vegetables, etc., and each category is further subdivided into types and varieties; in the production and processing link, it is classified according to the processing technology characteristics of vegetables, and vegetables are divided into 8 categories such as fresh vegetables treated on the surface, frozen vegetables, canned vegetables, dried vegetables, pickled vegetables, etc., and each category is further subdivided into types and varieties; in the commodity circulation and catering consumption links, it is classified according to the edible function characteristics of vegetables, taking the union of the classification in the planting and breeding link and the classification in the production and processing link, that is, vegetables and their vegetable products. Due to the inconsistent description of food classification in different links of the food supply chain, it leads to the problem of difficult accurate risk identification.

[0004] In the risk rating step, it is necessary to calculate the food safety risk level based on the data including food consumption. At this time, due to the inconsistent description of food classification in different links of the food supply chain, it is impossible to obtain the consumption data of food varieties with fine granularity, resulting in the problem of difficult risk rating.

[0005] In the risk control step, it is necessary to conduct risk control according to the risk rating of food. Due to the inconsistent description of food classification in different links of the food supply chain, it ultimately leads to information silos in the management and control of food supply chain risks, resulting in the problem of difficult risk control.

[0006] As an important part of artificial intelligence semantic web technology, a knowledge graph is a technical method for describing the association relationships between all things in the world, aiming to identify, discover, and infer the complex relationships between things and concepts from data, and is a computable model of thing relationships. Currently, there are already methods and devices for risk judgment of food safety incidents based on knowledge graph reasoning.

[0007] However, there are technical deficiencies in actual applications: (1) In terms of food safety risk identification, most of the currently constructed food safety knowledge graphs are based on the food classification standards of the industry, without considering the inconsistent classification of the same food in the food supply chain, resulting in a broken chain problem in food safety risk identification; (2) In terms of food safety risk identification, there is a lack of identification of food safety risk associations from the perspective of supply chain migration and diffusion, and the massive food safety supervision and sampling inspection and risk monitoring data are not fully utilized for risk reasoning of the upstream and downstream of the food supply chain; (3) In terms of food safety risk rating, the risk rating model lacks quantitative analysis of the spatio-temporal dimensions of the likelihood and severity of food hazard factors, resulting in insufficient personalization of the response and disposal plans and unsatisfactory timeliness. Summary of the Invention

[0008] In view of the problems existing in the prior art, the present invention proposes a food supply chain risk perception, identification, and rating system based on a knowledge graph. The first objective is to solve the problem in food safety risk identification: most of the currently constructed food safety knowledge graphs are based on the food classification standards of the industry, without considering the inconsistent classification of the same food in the food supply chain, resulting in a broken chain problem in food safety risk identification; the second objective is to solve the problem in food safety risk identification, that is, there is a lack of identification of food safety risk associations from the perspective of supply chain migration and diffusion, and the massive food safety supervision and sampling inspection and risk monitoring data are not fully utilized for risk reasoning of the upstream and downstream of the food supply chain; the third objective is to solve the problem in food safety risk rating: the risk rating model lacks quantitative analysis of the spatio-temporal dimensions of the likelihood and severity of food hazard factors, resulting in insufficient personalization of the response and disposal plans and unsatisfactory timeliness.

[0009] The present invention proposes the following technical solutions to solve its technical problems:

[0010] A food supply chain risk perception, identification, and rating system based on a knowledge graph, including a food supply chain risk perception module, a food supply chain risk identification module, a food supply chain risk rating module, and a food supply chain response and disposal module;

[0011] The food supply chain risk perception module provides basic data for the food supply chain risk identification module;

[0012] The food supply chain risk identification module includes a food supply chain knowledge graph management sub-module and a food safety knowledge graph reasoning sub-module. The food safety knowledge graph reasoning sub-module obtains basic data from the risk perception module and sends the food safety knowledge reasoning data to the risk rating module;

[0013] The food supply chain risk rating module is a risk rating module based on a knowledge graph. It conducts risk rating according to the food classification, spatio-temporal risk hierarchy, food hazard factor limit values, and food hazard factor detection values provided by the food safety knowledge graph reasoning sub-module, and sends the risk rating results to the food supply chain risk response and handling module;

[0014] The food supply chain response and handling module is a response and handling module based on a knowledge graph. According to the risk rating results of the food supply chain risk rating module, it matches response and handling templates and automatically generates response and handling plans;

[0015] Its characteristics are: the multi-level tree structure based on the food ontology of the food supply chain knowledge graph management sub-module encompasses all classification names of the same food in different food supply chain links; the food safety knowledge graph reasoning sub-module first locates the food supply chain level where the currently sampled food name is located, and then performs fuzzy matching on the classification name and limit value of the currently sampled food name for this food supply chain level; the risk rating of the food supply chain risk rating module for the likelihood and severity of risk occurrence is a rating based on the likelihood and severity of risk occurrence changing with time and space.

[0016] Furthermore, the food supply chain risk perception module provides basic data for the food supply chain risk identification module. This basic data includes: the name of the sampled food, the name of the food hazard factor of the sampled food, the detection value of the food hazard factor of the sampled food, the name of the sampled unit, the name of the nominally sampled production unit, the sampling time, the sampling address, the production time of the sampled food, and the production address of the sampled food; the name of the sampled unit is the unit name of food safety sampling and monitoring, including food production units and food business units; the name of the nominally sampled production unit is the unit name of food production, limited to food production units and not involving food business units.

[0017] Furthermore, the food supply chain knowledge graph management sub-module constructs a food safety knowledge ontology library and provides basic data for the food supply chain risk identification module, the food supply chain risk rating module, and the food supply chain response and handling module in the form of entity, attribute, and relationship triples. Specifically:

[0018] Manage using the knowledge graph ontology library. Extract the top levels of the multiple hierarchical tree structures shared by the food supply chain risk identification module, the food supply chain risk rating module, and the food supply chain response and handling module to establish an ontology. Specifically: establish a food ontology, which is used for the food safety knowledge graph reasoning sub-module to match the level of the food supply chain where the name of the food to be inspected is located with the food sub-classes below the food ontology; establish a food supply chain ontology, which is used for the food supply chain risk rating module to match the name of the inspected unit with the food supply chain sub-classes below the food supply chain ontology; establish a food hazard factor ontology based on the food supply chain; this ontology is used for the food safety knowledge graph reasoning sub-module to match the food classification of the food to be inspected with the food hazard factor sub-classes corresponding to this food classification below the food hazard factor ontology.

[0019] Manage in the form of entities and attributes of the knowledge graph. Construct the upper layer and the lower layer that have a master-slave relationship with the food ontology, the food supply chain ontology, and the food hazard factor ontology. Specifically: establish multiple levels of attributes that have a subordinate relationship with the food ontology and form a multi-level tree structure based on the food ontology. This multi-level tree structure based on the food ontology encompasses all the classification names of the same food in different food supply chain links; the upper layer below this multi-level tree structure ontology is the entity, and the lower layer is the attribute of the upper layer. This multi-level tree structure is used for the risk identification module to match the name of the food to be inspected with a certain attribute in this multi-level tree structure; establish multiple levels of attributes that have a subordinate relationship with the food supply chain ontology and form a multi-level tree structure based on the food supply chain ontology. The upper layer below this multi-level tree structure ontology is the entity, and the lower layer is the attribute of the upper layer. This multi-level tree structure is used for the food supply chain risk rating module to match the name of the inspected unit with a certain attribute in this multi-level tree structure; establish multiple levels of attributes that have a subordinate relationship with the food hazard factor ontology and form a multi-level tree structure based on the food hazard factor ontology. The upper layer below this multi-level tree structure ontology is the entity, and the lower layer is the attribute of the upper layer; this multi-level tree structure is used for the food safety knowledge graph reasoning sub-module to match the food classification of the food to be inspected with a certain attribute corresponding to this food classification in this multi-level tree structure.

[0020] Manage in the form of relationships of the knowledge graph. Specifically: mark the upstream and downstream in the multi-level tree structures of the food ontology, the food supply chain ontology, and the food hazard factor ontology respectively; mark the restricting party and the restricted party in the multi-level tree structures of the food ontology, the food supply chain ontology, and the food hazard factor ontology respectively; mark the supplier and the demander in the multi-level tree structures of the food ontology, the food supply chain ontology, and the food hazard factor ontology respectively; mark the processing party and the processed party in the multi-level tree structures of the food ontology, the food supply chain ontology, and the food hazard factor ontology respectively.

[0021] Furthermore, for the food safety knowledge graph reasoning sub-module of the food supply chain risk identification module, for the name of the food to be inspected, the name of the inspected unit, and the name of the nominally producing unit of the food to be inspected input by the food supply chain risk perception module, first perform knowledge reasoning on the food supply chain link. When the name of the inspected unit in the input data is the same as the name of the nominally producing unit of the food to be inspected, it is considered to belong to the food production and processing link, and this production and processing link is used as the food supply chain level where the food to be inspected is located; when the name of the inspected unit is different from the name of the nominally producing unit of the food to be inspected, it is considered to belong to the food storage, transportation, circulation, and catering consumption links, and the corresponding sub-links are matched according to the representative keywords of the unit name, and this sub-link is used as the food supply chain level where the food to be inspected is located; in the multi-level tree structure based on the food supply chain ontology, search for the food supply chain level where the food to be inspected is located, and then search in the multi-level tree structure based on the food ontology for the food classification corresponding to the name of the food to be inspected at this food supply chain level, and send this food classification to the food supply chain risk rating module; for the name of the hazard factor of the food to be inspected input by the food supply chain risk perception module, search in the multi-level tree structure based on the food hazard factor ontology for the limit value of the hazard factor corresponding to this food node, and send this hazard factor limit value and the detected value of the hazard factor of the food to be inspected to the food supply chain risk rating module.

[0022] Furthermore, for the name of the inspected unit input by the food supply chain risk perception module by the food safety knowledge graph reasoning sub-module of the food supply chain risk identification module, search in the multi-level tree structure based on the food supply chain ontology whether the inspected unit is an upstream supply chain node or a downstream supply chain node. If it is an upstream supply chain node, mark all downstream supply chain nodes corresponding to the name of the nominally producing unit associated with this upstream supply chain node as risk nodes, and send the spatio-temporal risk hierarchy corresponding to this risk node to the food supply chain risk rating module.

[0023] Furthermore, the food supply chain risk rating module performs risk rating for the name of the food to be inspected according to the food classification name, the corresponding hazard factor limit, and the detected value of the hazard factor sent by the food safety knowledge graph reasoning sub-module; and based on whether the inspected food unit is upstream or downstream in the multi-level tree structure of the food supply chain, perform spatio-temporal risk level marking for this inspected food unit. If the inspected food unit is upstream in the multi-level tree structure of the food supply chain, perform spatio-temporal risk marking for all downstream nominally producing units of it. If the inspected food unit is downstream in the multi-level tree structure of the food supply chain, only perform spatio-temporal risk marking for the current food unit.

[0024] Further, the food supply chain risk rating module performs a risk rating for the name of the food sampled according to the limit of the hazard factor and the content of the hazard factor corresponding to the name of the food sampled sent by the food safety knowledge graph reasoning sub-module, specifically as follows:

[0025] This risk rating is jointly completed by the risk occurrence probability calculation sub-module P(X,T), the risk occurrence severity calculation sub-module S(X,T), and the risk level assessment sub-module P(X,T); the risk occurrence probability calculation sub-module comprehensively considers the possibility of food safety risk occurrence based on the detection failure rate and consumption volume; the risk occurrence severity calculation sub-module judges the severity of food risk occurrence according to the degree of food contamination and the harm it causes; the risk level assessment module judges the level of food safety risk occurrence based on the possibility and severity of risk occurrence.

[0026] Further, the risk occurrence probability calculation sub-module comprehensively considers the possibility of food safety risk occurrence based on the detection failure rate and consumption volume, specifically: setting the risk occurrence probability P(X,T) as the product of the failure rate of the hazard factor and the food consumption volume:

[0027] P(X,T)=H(X,T)×F(X,T) (1)

[0028] where H(X,T) is the failure rate distribution of the hazard factor changing with time and space, F(X,T) is the function of food consumption volume changing with time and space, and the basic value of food consumption volume has different settings according to different seasons and regional locations.

[0029] Further, the risk occurrence severity calculation sub-module judges the severity of food risk occurrence according to the degree of food contamination and the harm it causes, specifically: determining the risk occurrence severity S(X,T) by the product of the contamination degree of the hazard factor and the harm it causes:

[0030] S(X,T)=I(X,T)×D(X,T) (2)

[0031] where I(X,T) is called the pollution index distribution of the hazard factor changing with time and space, the hazard factors are divided into chemical hazard factors and microbial hazard factors, D(X,T) represents the harm degree of the hazard factor changing with time and space, and the harm degree function has different settings according to the basic values of different specific time and space scenarios of the hazard factors.

[0032] The pollution index I of the chemical hazard factor at the specific time and space x,t is calculated as follows:

[0033]

[0034] where Mx,t The detected value of the hazard factor representing a specific time and space, L x,t Represents the limit value of the hazard factor for a specific time and space.

[0035] The pollution index I of a single indicator of the microbial contaminant in the specific time and space x,t The calculation formula is:

[0036]

[0037] where, Lmin x,t Represents the lowest limit value of the hazard index for a specific time and space, Lmax x,t Represents the highest limit value of the hazard index for a specific time and space.

[0038] Furthermore, the risk level assessment module determines the level of food safety risk according to the possibility and severity of the risk occurrence, specifically: the food supply chain risk value R(X,T) is set as the product of the possibility P(X,T) of the hazard factor occurrence and the severity S(X,T) of the hazard factor generation:

[0039] R(X,T) = P(X,T) × S(X,T) (5) According to the magnitude of the risk value R(X,T), the level of risk occurrence is divided into four levels.

[0040] Advantages and effects of the present invention

[0041] 1. Based on the multi-level tree structure of the food entity, the present invention encompasses all classification names of the same food in different food supply chain links. Its advantages are as follows: According to the name of the food to be inspected, as well as the name of the unit to be inspected and the name of the nominal production unit to be inspected, it can quickly find out which food supply chain link it is in the multi-level tree structure based on the food supply chain entity. Then, according to the food supply chain link it is in, it can perform fuzzy matching to the food classification name for the current supply chain link in the multi-level tree structure based on the food entity. Then, according to the food classification name, it can accurately find the limit value of its hazard factor. Then, based on the limit value of the hazard factor and the detected value of the hazard factor of the food to be inspected, it can conduct a food safety spatio-temporal risk rating, thereby making the food risk rating more dynamically accurate.

[0042] 2. Before searching for the food classification of the sampled food name in the knowledge graph, the food safety knowledge graph reasoning sub-module of the present invention first finds the supply chain level where the sampled food unit corresponding to the sampled food name is located, and then searches for the classification of the sampled food name corresponding to it in the multi-level tree structure based on the food ontology. This solves the problem in the prior art that due to different supply chain links of the same food resulting in different classification names, there is a disconnection in food safety risk identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a block diagram of the system for food supply chain risk perception, identification, and rating of the present invention;

[0044] Figure 2 It is a flowchart of the food safety knowledge graph reasoning sub-module of the present invention;

[0045] Figure 3 It is a schematic diagram of the ontology library based on the food supply chain of the present invention;

[0046] Figure 4 It is a schematic diagram of the knowledge graph based on the food supply chain of the present invention; DETAILED DESCRIPTION OF THE INVENTION

[0047] Design Principle of the Present Invention

[0048] 1. Innovation Points of the Present Invention:

[0049] One of the innovation points lies in: organically combining three information silos (food classification, food supply chain, food hazard factors) of the prior art through a knowledge graph. First, organically combine food classification and food supply chain: (1) In the establishment of the knowledge graph (food supply chain knowledge graph management sub-module), the food classification common to different industries is combined with the food supply chain links. In the multi-level tree structure based on the food ontology, the same food entity may correspond to the names of multiple supply chains. The entity is at the upper layer, and the names of multiple supply chains are at the lower layer of this entity. Then, taking each supply chain as an entity, each entity is further subdivided into multiple food classifications below, thus organically combining the multi-level tree structure based on the food ontology and the food supply chain; (2) In the risk reasoning process of the knowledge graph (food safety knowledge graph reasoning sub-module), first search for the food supply chain link where the name of the food to be inspected is located. After determining the food supply chain link where the name of the food to be inspected is located, then search in the multi-level tree structure based on the food ontology for the food classification corresponding to this food supply chain link; conversely, if it is not known which link the current food to be inspected is in, it is equivalent to having no search keyword, and without a search keyword, it is impossible to search. Second, organically combine food hazard factors and food supply chain. In different food supply chain links, the limits of food hazard factors for the same food name are different. Before searching for the limit of the hazard factor corresponding to the name of the food to be inspected in the present invention, first search for the food supply chain link where the name of the food to be inspected is located, then search for the food classification according to the food supply chain link, and finally search for the limit of the hazard factor corresponding to this food classification.

[0050] The second innovation point of the present invention lies in: using the upstream and downstream relationships of the knowledge graph for spatio-temporal risk identification and risk rating. If the unit to be inspected is an upstream enterprise, then conduct spatio-temporal risk rating on all associated nominal production enterprises downstream of it.

[0051] The third innovation point of the present invention lies in: conducting spatio-temporal risk rating according to the possibility and severity of the risk. Here, the possibility P(X,T) and severity S(X,T) both change with time and space, rather than being static. The time refers to the sampling time or the production time of the product to be inspected, and the space refers to the sampling location or the production location of the product to be inspected.

[0052] In short, through the food supply chain of the knowledge graph, the present invention accurately finds the food classification, then accurately finds the corresponding limit of the hazard factor through the food classification, and also quickly finds the corresponding relationship between the upstream and downstream in the supply chain through the food supply chain of the knowledge graph. Then, through the corresponding relationship between the upstream and downstream, accurately capture the area affected by the risk, and then conduct a risk rating including spatio-temporal information on the area affected by the risk, making the risk rating more accurate and scientific.

[0053] Based on the above principles, the present invention designs a food supply chain risk perception, identification, and rating system based on a knowledge graph, as Figure 1 shown, which includes a food supply chain risk perception module, a food supply chain risk identification module, a food supply chain risk rating module, and a food supply chain response and handling module;

[0054] The food supply chain risk perception module provides basic data for the food supply chain risk identification module;

[0055] The food supply chain risk identification module includes a food supply chain knowledge graph management sub-module and a food safety knowledge graph reasoning sub-module. The food safety knowledge graph reasoning sub-module obtains basic data from the risk perception module and sends the food safety knowledge reasoning data to the risk rating module;

[0056] The food supply chain risk rating module is a risk rating module based on a knowledge graph. It conducts risk rating according to the food classification, spatio-temporal risk hierarchy, food hazard factor limit values, and food hazard factor detection values provided by the food safety knowledge graph reasoning sub-module, and sends the risk rating results to the food supply chain risk response and handling module;

[0057] The food supply chain response and handling module is a response and handling module based on a knowledge graph. According to the risk rating results of the food supply chain risk rating module, it matches the response and handling templates and automatically generates response and handling plans;

[0058] Its characteristics are: The multi-level tree structure based on the food ontology of the food supply chain knowledge graph management sub-module encompasses all classification names of the same food in different food supply chain links; The food safety knowledge graph reasoning sub-module first finds the food supply chain level where the currently sampled food name is located, and then fuzzily matches the classification name and limit value of the currently sampled food name for this food supply chain level; The risk rating of the food supply chain risk rating module for the likelihood and severity of risk occurrence is based on the likelihood and severity of risk occurrence changing with time and space.

[0059] As Figure 1 shown, the food supply chain risk perception module provides basic data for the food supply chain risk identification module. The basic data includes: the name of the sampled food, the name of the food hazard factor of the sampled food, the detection value of the food hazard factor of the sampled food, the name of the sampled unit, the name of the nominal production unit of the sampled food, the sampling time, the sampling address, the production time of the sampled food, and the production address of the sampled food; The name of the sampled unit is the name of the unit for food safety sampling and monitoring, including food production units and food business units; The name of the nominal production unit of the sampled food is the name of the food production unit, limited to food production units and not involving food business units.

[0060] Furthermore, the food supply chain knowledge graph management sub-module constructs a food safety knowledge ontology library, and provides basic data for the food supply chain risk identification module, the food supply chain risk rating module, and the food supply chain response and handling module in the form of entity, attribute, and relationship triples. Specifically:

[0061] As Figure 3 shown, it is managed with the knowledge graph ontology library. The top levels of multiple hierarchical tree structures shared by the food supply chain risk identification module, the food supply chain risk rating module, and the food supply chain response and handling module are extracted to establish an ontology. Specifically: a food ontology is established, which is used for the food safety knowledge graph reasoning sub-module to match the level of the food supply chain where the name of the food to be inspected is located with the food sub-classes below the food ontology; a food supply chain ontology is established, which is used for the food supply chain risk rating module to match the name of the unit to be inspected with the food supply chain sub-classes below the food supply chain ontology; a food hazard factor ontology based on the food supply chain is established; this ontology is used for the food safety knowledge graph reasoning sub-module to match the food classification of the food to be inspected with the food hazard factor sub-classes corresponding to this food classification below the food hazard factor ontology.

[0062] It is managed in the form of entities and attributes of the knowledge graph, and the upper layer and the lower layer with a master-slave relationship with the food ontology, the food supply chain ontology, and the food hazard factor ontology are constructed. Specifically: multiple attributes with a subordinate relationship to the food ontology are established, and a multi-level tree structure based on the food ontology is formed. This multi-level tree structure based on the food ontology encompasses all classification names of the same food in different food supply chain links; the upper layer below this multi-level tree structure ontology is the entity, and the lower layer is the attribute of the upper layer. This multi-level tree structure is used for the risk identification module to match the name of the food to be inspected with a certain attribute in this multi-level tree structure; multiple attributes with a subordinate relationship to the food supply chain ontology are established, and a multi-level tree structure based on the food supply chain ontology is formed. The upper layer below this multi-level tree structure ontology is the entity, and the lower layer is the attribute of the upper layer. This multi-level tree structure is used for the food supply chain risk rating module to match the name of the unit to be inspected with a certain attribute in this multi-level tree structure; multiple attributes with a subordinate relationship to the food hazard factor ontology are established, and a multi-level tree structure based on the food hazard factor ontology is formed. The upper layer below this multi-level tree structure ontology is the entity, and the lower layer is the attribute of the upper layer; this multi-level tree structure is used for the food safety knowledge graph reasoning sub-module to match the food classification of the food to be inspected with a certain attribute corresponding to this food classification in this multi-level tree structure.

[0063] Manage in the form of relationships in the knowledge graph, specifically: Mark the upstream and downstream in the multi-level tree structures of the food ontology, food supply chain ontology, and food hazard factor ontology respectively; Mark the restricting party and the restricted party in the multi-level tree structures of the food ontology, food supply chain ontology, and food hazard factor ontology respectively; Mark the supplier and the demander in the multi-level tree structures of the food ontology, food supply chain ontology, and food hazard factor ontology respectively; Mark the processing party and the processed party in the multi-level tree structures of the food ontology, food supply chain ontology, and food hazard factor ontology respectively.

[0064] As Figure 1 shown, for the food safety knowledge graph reasoning sub-module of the food supply chain risk identification module, for the name of the food to be inspected, the name of the inspected unit, and the name of the nominally producing unit of the food to be inspected input by the food supply chain risk perception module, first conduct knowledge reasoning on the food supply chain link. When the name of the inspected unit in the input data is the same as the name of the nominally producing unit of the food to be inspected, it is considered to belong to the food production and processing link, and this production and processing link is used as the food supply chain level where the food to be inspected is located; When the name of the inspected unit is different from the name of the nominally producing unit of the food to be inspected, it is considered to belong to the food storage, transportation, circulation, and catering consumption links, and the corresponding sub-links are matched according to the representative keywords of the unit name, and this sub-link is used as the food supply chain level where the food to be inspected is located; In the multi-level tree structure based on the food supply chain ontology, search for the food supply chain level where the food to be inspected is located, and find the food classification corresponding to the name of the food to be inspected at this food supply chain level in the multi-level tree structure based on the food ontology, and send this food classification to the food supply chain risk rating module; For the name of the food hazard factor of the food to be inspected input by the food supply chain risk perception module, in the multi-level tree structure based on the food hazard factor ontology, search for the limit value of the hazard factor corresponding to this food node, and send this hazard factor limit value and the detected value of the food hazard factor of the food to be inspected to the food supply chain risk rating module.

[0065] Furthermore, for the name of the inspected unit input by the food supply chain risk perception module by the food safety knowledge graph reasoning sub-module of the food supply chain risk identification module, in the multi-level tree structure based on the food supply chain ontology, search whether the inspected unit is an upstream supply chain node or a downstream supply chain node. If it is an upstream supply chain node, uniformly mark all downstream supply chain nodes corresponding to this upstream supply chain node associated with the name of the nominally producing unit of the food to be inspected as risk nodes, and send the spatio-temporal risk hierarchy structure corresponding to this risk node to the food supply chain risk rating module.

[0066] Supplementary Note 1:

[0067] The spatio-temporal risk hierarchy refers to the sampling time, sampling address, production time of the sampled product, and production address of the sampled product. This information comes from the risk perception module, which transmits this information to the risk identification module, and the risk identification module then transmits this information to the risk rating module.

[0068] Furthermore, the food supply chain risk rating module conducts a risk rating for the name of the sampled food based on the food classification name sent by the food safety knowledge graph reasoning sub-module, as well as the corresponding hazard factor limits and hazard factor detection values; and based on whether the sampled food unit is upstream or downstream in the multi-level tree structure of the food supply chain, it assigns a spatio-temporal risk level label to the sampled food unit. If the sampled food unit is upstream in the multi-level tree structure of the food supply chain, then all the downstream sampled nominal production units are assigned spatio-temporal risk labels. If the sampled food unit is downstream in the multi-level tree structure of the food supply chain, then only the current food unit is assigned a spatio-temporal risk label.

[0069] Furthermore, the food supply chain risk rating module conducts a risk rating for the name of the sampled food based on the hazard factor limits and hazard factor contents corresponding to the name of the sampled food sent by the food safety knowledge graph reasoning sub-module, specifically as follows:

[0070] This risk rating is jointly completed by the risk occurrence probability calculation sub-module P(X,T), the risk occurrence severity calculation sub-module S(X,T), and the risk level assessment sub-module P(X,T); the risk occurrence probability calculation sub-module comprehensively considers the possibility of food safety risk occurrence based on the detection failure rate and consumption volume; the risk occurrence severity calculation sub-module judges the severity of food risk occurrence based on the degree of food contamination and the harm it causes; the risk level assessment module judges the level of food safety risk occurrence based on the possibility and severity of risk occurrence.

[0071] Furthermore, the risk occurrence probability calculation sub-module comprehensively considers the possibility of food safety risk occurrence based on the detection failure rate and consumption volume, specifically as follows: Set the risk occurrence probability P(X,T) as the product of the failure rate of the hazard factor and the food consumption volume:

[0072] P(X,T) = H(X,T) × F(X,T) (1)

[0073] Wherein, H(X,T) is the failure rate distribution of the hazard factor changing with time and space, and F(X,T) is the function of food consumption volume changing with time and space. The basic value of the food consumption volume has different settings according to different seasons and regional locations.

[0074] Supplementary Note 2:

[0075] The time-varying information is used by the risk rating module for dynamic risk rating. The risk rating is based not only on the limited values of the sampled hazard factors and the detected values of the sampled hazard factors, but also on the time-varying information, that is, it is necessary to consider whether the intelligent response and handling module has eliminated the current risk before conducting the risk rating; the space-varying information is also used for risk rating. The spatial information refers to the sampling location and the sampled production location. If risks occur repeatedly at the same sampling location or the same sampled production location, the severity level is relatively high.

[0076] Furthermore, the risk occurrence severity calculation sub-module determines the severity of food risk occurrence based on the degree of food contamination and the harm it causes. Specifically: the risk occurrence severity S(X,T) is determined by the product of the contamination degree of the hazard factor and the harm it causes:

[0077] S(X,T) = I(X,T) × D(X,T) (2)

[0078] Among them, I(X,T) is called the pollution index distribution of the hazard factor with spatio-temporal changes. The hazard factors are divided into chemical hazard factors and microbial hazard factors. D(X,T) represents the harm degree of the hazard factor with spatio-temporal changes. The harm degree function has different settings according to the specific spatio-temporal scenario base values of different hazard factors.

[0079] The pollution index I of the chemical hazard factor at the specific time and space x,t is calculated as follows:

[0080]

[0081] Among them, M x,t represents the detected value of the hazard factor at the specific time and space, and L x,t represents the limited value of the hazard factor at the specific time and space.

[0082] The pollution index I of a single index of the microbial contaminant at the specific time and space x,t The calculation formula is:

[0083]

[0084] Among them, Lmin x,t represents the lowest limit value of the hazard index at the specific time and space, and Lmax x,t represents the highest limit value of the hazard index at the specific time and space.

[0085] Further, the risk level assessment module determines the level of food safety risk based on the likelihood and severity of the risk occurrence. Specifically, the food supply chain risk value R(X,T) is set as the product of the likelihood P(X,T) of the hazard factor occurrence and the severity S(X,T) of the hazard factor:

[0086] R(X,T) = P(X,T) × S(X,T) (5) According to the magnitude of the risk value R(X,T), the level of risk occurrence is divided into four levels.

[0087] Supplementary Note 3:

[0088] The intelligent response and disposal module for food supply chain risks matches the response and disposal template according to the food classification, associated risk hierarchy, and risk level results of the food supply chain risk at the corresponding link, and automatically generates a response and disposal plan. For details, refer to the invention patent 202210533370.6 Food Safety Risk Response and Disposal Method, Electronic Device, and Storage Medium, which will not be elaborated here.

[0089] The above content is only an example and illustration of the concept of the present invention. Those skilled in the art of this technology can make various modifications, supplements, or use similar methods to replace the specific embodiments described. As long as they do not deviate from the concept of the invention or exceed the scope defined by this claim book, they should fall within the protection scope of the present invention.

Claims

1. A food supply chain risk perception, identification, and rating system based on a knowledge graph, including a risk perception module, a risk identification module, a risk rating module, and a response and handling module; the risk perception module provides basic data for the risk identification module; The risk identification module includes a knowledge graph management sub-module and a food safety knowledge graph reasoning sub-module. The food safety knowledge graph reasoning sub-module obtains basic data from the risk perception module and sends the reasoning data to the risk rating module; The risk rating module conducts risk rating based on the food classification, spatio-temporal risk hierarchy, food hazard factor limit values, and food hazard factor detection values provided by the food safety knowledge graph reasoning sub-module, and sends the risk rating results to the response and handling module; The response and handling module matches the response and handling template according to the risk rating results of the risk rating module and automatically generates a response and handling plan; It is characterized in that: The multi-level tree structure based on the food ontology of the knowledge graph management sub-module encompasses all classification names of the same food in different food supply chain links; the food safety knowledge graph reasoning sub-module first locates the food supply chain level where the currently sampled food name is located, and then performs fuzzy matching on the classification name and limit value of the currently sampled food name for this food supply chain level; the risk rating module's rating of the likelihood and severity of risk occurrence is based on the likelihood and severity of risk occurrence with spatio-temporal changes; The step of first locating the food supply chain level where the currently sampled food name is located is specifically as follows: First, conduct knowledge reasoning on the food supply chain link. When the sampled unit name in the input data is the same as the sampled nominal production unit name, it is considered to belong to the food production and processing link, and this production and processing link is used as the food supply chain level where the sampled food is located; when the sampled unit name is different from the sampled nominal production unit name, it is considered to belong to the food storage, transportation, circulation, and catering consumption links.

2. The food supply chain risk perception, identification, and rating system based on a knowledge graph according to claim 1, wherein: The risk perception module provides basic data for the risk identification module, and this basic data includes: the sampled food name, the sampled food hazard factor name, the sampled food hazard factor detection value, the sampled unit name, the sampled nominal production unit name, the sampling time, the sampling address, the sampled production time, and the sampled production address; the sampled unit name is the unit name of food safety sampling and monitoring, including food production units and food business units; the sampled nominal production unit name is the unit name of food production, limited to food production units and not involving food business units.

3. The food supply chain risk perception, identification, and rating system based on a knowledge graph according to claim 1, characterized in that: The knowledge graph management sub-module constructs a food safety knowledge ontology library and provides basic data for the risk identification module, the risk rating module, and the response and handling module in the form of entity, attribute, and relationship triples. Specifically: Manage using the knowledge graph ontology library. Extract the highest levels of multiple hierarchical tree structures shared by the risk identification module, risk rating module, and response handling module to establish an ontology. Specifically: establish a food ontology, which is used for the food safety knowledge graph inference sub-module to match the level of the food supply chain where the name of the sampled food is located with the food sub-classes below the food ontology; establish a food supply chain ontology, which is used for the risk rating module to match the name of the sampled unit with the food supply chain sub-classes below the food supply chain ontology; establish a food hazard factor ontology based on the food supply chain; this ontology is used for the food safety knowledge graph inference sub-module to match the food classification of the sampled food name with the food hazard factor sub-classes corresponding to this food classification below the food hazard factor ontology. Manage in the form of entities and attributes of the knowledge graph, and construct the upper layer and the lower layer with a master-slave relationship with the food ontology, food supply chain ontology, and food hazard factor ontology. Specifically: establish multiple levels of attributes with a subordinate relationship to the food ontology and form a multi-level tree structure based on the food ontology. This multi-level tree structure based on the food ontology encompasses all classification names of the same food in different food supply chain links; the upper layer below this multi-level tree structure ontology is the entity, and the lower layer is the attribute of the upper layer. This multi-level tree structure is used for the risk identification module to match the name of the sampled food with a certain attribute in this multi-level tree structure; establish multiple levels of attributes with a subordinate relationship to the food supply chain ontology and form a multi-level tree structure based on the food supply chain ontology. The upper layer below this multi-level tree structure ontology is the entity, and the lower layer is the attribute of the upper layer. This multi-level tree structure is used for the risk rating module to match the name of the sampled unit with a certain attribute in this multi-level tree structure; establish multiple levels of attributes with a subordinate relationship to the food hazard factor ontology and form a multi-level tree structure based on the food hazard factor ontology. The upper layer below this multi-level tree structure ontology is the entity, and the lower layer is the attribute of the upper layer; this multi-level tree structure is used for the food safety knowledge graph inference sub-module to match the food classification of the sampled food name with a certain attribute corresponding to this food classification in this multi-level tree structure. Manage in the form of relationships of the knowledge graph. Specifically: mark the upstream and downstream in the multi-level tree structures of the food ontology, food supply chain ontology, and food hazard factor ontology respectively; mark the restricting party and the restricted party in the multi-level tree structures of the food ontology, food supply chain ontology, and food hazard factor ontology respectively; mark the supplier and the demander in the multi-level tree structures of the food ontology, food supply chain ontology, and food hazard factor ontology respectively; mark the processing party and the processed party in the multi-level tree structures of the food ontology, food supply chain ontology, and food hazard factor ontology respectively.

4. The food supply chain risk perception, identification, and rating system based on a knowledge graph according to claim 3, characterized in that: The food safety knowledge graph reasoning sub-module of the risk identification module first conducts knowledge reasoning on the food supply chain links for the name of the food subject to sampling inspection, the name of the sampled unit, and the name of the nominally sampled production unit input by the risk perception module. When the name of the sampled unit in the input data is the same as the name of the nominally sampled production unit, it is considered to belong to the food production and processing link, and this production and processing link is used as the food supply chain level where the food subject to sampling inspection is located; when the name of the sampled unit is different from the name of the nominally sampled production unit, it is considered to belong to the food storage, transportation, circulation, and catering consumption links, and the corresponding sub-links are matched according to the representative keywords of the unit name, and this sub-link is used as the food supply chain level where the food subject to sampling inspection is located; in the multi-level tree structure based on the food supply chain ontology, search for the food supply chain level where the food subject to sampling inspection is located, and then search in the multi-level tree structure based on the food ontology for the food classification corresponding to the name of the food subject to sampling inspection at this food supply chain level, and send this food classification to the risk rating module; For the name of the hazard factor of the food subject to sampling inspection input by the risk perception module, search in the multi-level tree structure based on the food hazard factor ontology for the limit value of the hazard factor corresponding to this food node, and send this hazard factor limit value and the detected value of the hazard factor of the food subject to sampling inspection to the risk rating module.

5. The food supply chain risk perception, identification, and rating system based on a knowledge graph according to claim 3, characterized in that: The food safety knowledge graph reasoning sub-module of the risk identification module searches in the multi-level tree structure based on the food supply chain ontology to determine whether the sampled unit input by the risk perception module is an upstream supply chain node or a downstream supply chain node. If it is an upstream supply chain node, all downstream supply chain nodes corresponding to the nominally sampled production unit associated with this upstream supply chain node are marked as risk nodes, and the spatio-temporal risk hierarchy corresponding to this risk node is sent to the risk rating module.

6. The food supply chain risk perception, identification, and rating system based on a knowledge graph according to claim 5, characterized in that: The risk rating module conducts a risk rating for the name of the food subject to sampling inspection based on the food classification name sent by the food safety knowledge graph reasoning sub-module, as well as the corresponding hazard factor limit and hazard factor detection value; And based on whether the sampled food unit is upstream or downstream in the multi-level tree structure of the food supply chain, conduct spatio-temporal risk level marking for this sampled food unit. If the sampled food unit is upstream in the multi-level tree structure of the food supply chain, all downstream nominally sampled production units of it are subject to spatio-temporal risk marking. If the sampled food unit is downstream in the multi-level tree structure of the food supply chain, only the current food unit is subject to spatio-temporal risk marking.

7. The food supply chain risk perception, identification, and rating system based on a knowledge graph according to claim 6, characterized in that: The risk rating module conducts a risk rating for the name of the food subject to sampling inspection based on the hazard factor limit and hazard factor content corresponding to the name of the food subject to sampling inspection sent by the food safety knowledge graph reasoning sub-module, specifically as follows: This risk rating is jointly completed by the risk occurrence probability calculation sub-module P(X, T), the risk occurrence severity calculation sub-module S(X, T), and the risk level assessment sub-module P(X, T); the risk occurrence probability calculation sub-module comprehensively considers the possibility of food safety risk occurrence based on the detection unqualified rate and consumption volume; the risk occurrence severity calculation sub-module judges the severity of food risk occurrence according to the food pollution degree and the harm degree it causes; the risk level assessment module judges the level of food safety risk occurrence based on the possibility and severity of risk occurrence.

8. The food supply chain risk perception, identification, and rating system based on a knowledge graph according to claim 7, characterized in that: The risk occurrence probability calculation sub-module comprehensively considers the possibility of food safety risk occurrence based on the detection unqualified rate and consumption volume, specifically: setting the risk occurrence probability P(X, T) as the product of the unqualified rate of the hazard factor and the food consumption volume: P(X, T) = H(X, T) × F(X, T) (1) Among them, H(X, T) is the unqualified rate distribution of the hazard factor changing with time and space, F(X, T) is the function of food consumption volume changing with time and space, and the basic value of food consumption volume has different settings according to different seasons and regional locations.

9. The food supply chain risk perception, identification, and rating system based on a knowledge graph according to claim 7, wherein: The risk occurrence severity calculation sub-module judges the severity of food risk occurrence according to the food pollution degree and the harm degree it causes, specifically: the risk occurrence severity S(X, T) is determined by the product of the pollution degree of the hazard factor and the harm degree it causes S(X, T = I(X, T) × D(X, T) (2) Among them, I(X, T) is called the pollution index distribution of the hazard factor changing with time and space. The hazard factors are divided into chemical hazard factors and microbial hazard factors. D(X, T) represents the harm degree of the hazard factor changing with time and space. The harm degree function has different settings according to the basic values of different specific time and space scenarios where the hazard factors are located; The pollution index I of the chemical hazard factors at the specific time and space x，t is calculated as follows: Among them, M x，t represents the detected value of the hazard factor at a specific time and space, and L x，t represents the limit value of the hazard factor at a specific time and space; The pollution index I of a single indicator of microbial contaminants in a specific time and space x，t The calculation formula is as follows: Among them, Lmin x，t represents the minimum limit value of the hazard index for a specific time and space, and Lmax x，t represents the maximum limit value of the hazard index for a specific time and space.

10. The food supply chain risk perception, identification, and rating system based on a knowledge graph according to claim 7, wherein: The risk level assessment module judges the level of food safety risk occurrence based on the possibility and severity of risk occurrence, specifically: setting the food supply chain risk value R(X, T) as the product of the possibility P(X, T) of the hazard factor occurrence and the severity S(X, T) of the hazard factor generated: R(X, T) = P(X, T) × S(X, T) (5) According to the magnitude of the risk value R(X, T), the level of risk occurrence is divided into four levels.

Images