Emergency network public opinion evolution prediction and early warning method based on SEIR model

Abstract

The invention relates to the technical field of network public opinion evolution, in particular to an emergency network public opinion evolution prediction and early warning method based on an SEIR model, and the method comprises the following steps: S1, individual state division: dividing an individual into a plurality of states; s2, an individual state transition process: finally entering a shift-out state R; s3, establishing a network public opinion information diffusion model: calculating a network public opinion diffusion range and a network public opinion diffusion speed; s4, a network public opinion information diffusion process in the complex social network: simulating a public opinion diffusion process; s5, building an emergency network public opinion early warning model based on SPN: calculating an early warning probability; s6, early warning index risk rating: dividing early warning risk grades; s7, setting an early warning threshold value and a rating rule: starting early warnings of different levels according to the rating rule; according to the invention, the development situation of the network public opinion can be described more accurately.

Description

Technical Field

[0001] The present invention relates to the technical field of network public opinion evolution, and in particular to a method for predicting and warning the evolution of network public opinion of emergencies based on a SEIR model. Background Art

[0002] With the rapid development of information technology, the Internet has become the main channel for the generation, dissemination and diffusion of public opinion. Once emergencies occur, such as natural disasters, public health incidents and social security incidents, they can often quickly attract public attention and form extensive discussions and public opinions on the Internet. The development of such online public opinion is characterized by rapid speed, wide range and large emotional fluctuations. Therefore, how to effectively respond to and warn of online public opinion about emergencies has become a major challenge facing enterprises and related institutions.

[0003] The existing SEIR model-based emergency network public opinion evolution model has achieved certain application results in theory and practice. The model simulates the process of public opinion from initial spread to decline by analogy with the spread of infectious diseases. However, the current model and early warning mechanism have many shortcomings, mainly manifested in the relatively simple prediction of the evolution trend of public opinion and the lack of effective identification and timely response to early warning signals. In addition, the warning threshold setting of the existing mechanism is not reasonable enough and cannot cope well with the changes in public opinion at different stages. At the same time, the automation and intelligence level of the early warning needs to be improved, making it difficult to achieve real-time monitoring and rapid response.

[0004] The present invention aims to solve the above-mentioned problems in the prior art and proposes a method for predicting and warning the evolution of emergency network public opinion based on the SEIR model. By optimizing the early warning mechanism of public opinion evolution, setting reasonable early warning thresholds and response strategies, the accuracy and response speed of the early warning are improved, and real-time monitoring and intelligent processing of the spread of public opinion are realized. Summary of the invention

[0005] The present invention provides a method for predicting and warning the evolution of network public opinion of emergencies based on a SEIR model.

[0006] A method for predicting and warning the evolution of network public opinion of emergencies based on the SEIR model includes the following steps:

[0007] S1, individual state division: the individuals involved in the diffusion of online public opinion information are divided into multiple states, including the uninformed state S, the informed state E, the propagation state I and the removal state R, where;

[0008] The uninformed state S refers to individuals who have not yet been affected by online public opinion;

[0009] The informed state E refers to individuals who understand online public opinion information but do not spread it;

[0010] The dissemination state I is the individual that spreads the network public opinion information;

[0011] The removed state R refers to individuals who have lost interest in online public opinion information and no longer spread it;

[0012] S2, individual state transfer process: in the process of spreading online public opinion about emergencies, individuals in the uninformed state S change to the informed state E after receiving information, individuals in the informed state E choose to spread public opinion and become individuals in the spreading state I, and individuals in the informed state E who do not spread public opinion directly enter the removal state R, and individuals in the spreading state I lose their attention to public opinion after a predetermined period of time and eventually enter the removal state R;

[0013] S3, Establishment of the online public opinion information diffusion model: Based on the SEIR model, the online public opinion diffusion model is constructed. The online public opinion diffusion network is assumed to be static, the group size is set to N, and the differential equation of individual state transfer is established using the mean field method. The proportion of public opinion communicators and informed persons is simulated according to the differential equation, and the scope and speed of online public opinion diffusion are calculated.

[0014] S4, the process of online public opinion information diffusion in complex social networks: The process of public opinion diffusion is simulated through simulation experiments. The initial state is set as 98% of individuals are in an uninformed state and 2% of individuals are in a spreading state. As time goes by, the scope of public opinion diffusion gradually expands, and the proportion of spreading individuals first increases, then decreases, and finally disappears;

[0015] S5, Construction of emergency network public opinion warning model based on SPN: The SEIR-SPN structure model is used to model the evolution and warning of emergency network public opinion, set the key elements and changes of warning, analyze the state transition in the process of public opinion development through Markov chain, and calculate the warning probability according to different states;

[0016] S6, early warning indicator risk rating: The early warning indicators of network public opinion of emergencies are weighted by network analysis method improved by analytic hierarchy process, and the early warning risk levels are divided into Class I, Class II and Class III;

[0017] S7, warning threshold and rating rule setting: according to the simulation results, the warning threshold is set to 0.1, and the warning levels are divided into blue, yellow, orange and red, and different levels of warnings are activated according to the rating rules.

[0018] Optionally, the calculation of the transition probability of the individual state transition process in S2 includes:

[0019] The transition probability from the uninformed state S to the informed state E is expressed as:

[0020]

[0021] Among them, Q i represents the number of people spreading public opinion in the social network, k i represents the number of people that individual i contacts and spreads public opinion information to, and α is the transfer parameter;

[0022] The transition probability from the uninformed state S to the propagation state I is expressed as:

[0023]

[0024] The transition probability from the informed state E to the propagated state I is expressed as:

[0025]

[0026] Among them, b represents the social reinforcement effect, mi represents the number of times individual i contacts the public opinion communicator, and M represents the memory tolerance;

[0027] The transition probability from the informed state E to the exit state R is expressed as:

[0028] P i t (E→R)=1,m>M;

[0029] The transition probability from the propagation state I to the exit state R is expressed as:

[0030]

[0031] Among them, γ is the removal rate.

[0032] Optionally, the differential equation for the individual state transition in S3 is expressed as:

[0033]

[0034] Among them, α, 1-α, β, ω and γ correspond to the transition probabilities between states S→I, S→E, E→I, E→R and I→R respectively. It represents the probability that a node with in-degree k at time t has an edge coming from a propagation node.

[0035] Optionally, the diffusion range of the online public opinion is obtained by calculating the ratio of informed persons, disseminators and movers to all individuals, expressed as:

[0036] z(t)=e(t)+i(t)+r(t)=1-s(t);

[0037] Among them, z(t) is the ratio of informed individuals, disseminators and emigrants to all individuals, i(t) is the ratio of disseminators to all individuals, e(t) is the ratio of informed individuals to all individuals, r(t) is the ratio of emigrants to all individuals, and s(t) is the ratio of uninformed individuals to all individuals.

[0038] Optionally, the calculation of the network public opinion diffusion speed is expressed as:

[0039]

[0040] Δν t = -ΔS(t) = S(t-1) - S(t);

[0041] Where v is the average number of informed individuals added or the number of uninformed individuals reduced at each moment when the system runs to time T, Δν t is the instantaneous diffusion velocity.

[0042] Optionally, the construction of the SPN-based emergency network public opinion warning model in S5 includes:

[0043] S51, define early warning elements: set the early warning key element set P of the network public opinion of the emergency event = {P1, P2, ..., P n}, including original information of public opinion events, harmfulness of public opinion information, authenticity of public opinion information, integration and processing of public opinion information, deviation of situation development, data sentiment analysis, deterioration of netizens' emotional behavior, activation of crisis warning, calming of public opinion and control of the situation;

[0044] S52, set warning transition: define the transition in SPN as an exponential distribution function in Molloy form, expressed as:

[0045]

[0046] T={t1,t2,…,t k};

[0047] Among them, λ t represents the average rate of transition t, the random delay time variable x≥0, and T represents the total number of changes of warning elements;

[0048] S53, Construction of warning input and output relationship: through directed arc element set It represents the input-output relationship within the warning. Each arc is defined as input (triggering warning conditions) or output (warning results). The arc weight function W:F→N+ assigns weights to directed arcs. The weight value is N+={1, 2, …, n}. At the same time, the identification set M:P→N represents the state that occurs, and M1 is the initial identification.

[0049] S54, calculation of transition rate: define λ as the transition rate, and calculate it using a negative exponential distribution function;

[0050] S55, establishment of Markov chain: the SPN model is isomorphic to the Markov chain, the state transition diagram of the SPN model is constructed, and the Markov chain is used to analyze the dynamic changes of state transition during the development of public opinion.

[0051] Optionally, the establishment of the Markov chain in S55 includes:

[0052] S551, define state vector: define all state vectors M i , where M1,M2,...,M 12 They represent different evolution states of public opinion respectively, and the initial definition of the state vector is expressed as:

[0053]

[0054] S552, Steady-state constraints: Steady-state constraints are defined based on the Markov stationary distribution theorem and the Chapman-Kolmogorov equation and are expressed as:

[0055]

[0056] S553, Construction and solution of stationary distribution equation: According to the state vector M1,M2,...,M 12 and the corresponding state transition rates λ1,λ2,...,λ 12 , construct the stationary distribution equation, and calculate the stationary probability of each state, expressed as:

[0057]

[0058] S554, Application of Markov chain to analyze the evolution of public opinion: Utilize the calculation results of the Markov chain steady state to analyze the long-term evolution trend of network public opinion on emergencies.

[0059] Optionally, the rating rules in S7 include:

[0060] If the warning probability P∈[0.1,0.2] or reaches the warning critical surface, the blue warning is activated;

[0061] If the warning probability P∈(0.2,0.3] or reaches the warning critical surface, the yellow warning is activated;

[0062] If the warning probability P∈(0.3,0.4] or reaches the warning critical surface, the orange warning is activated;

[0063] If the warning probability P>0.4 or reaches the warning critical surface, the red warning will be activated.

[0064] Optionally, the warning probability is expressed as:

[0065]

[0066] Among them, P represents the probability of crisis warning P(M 12 ), ψ and l are unknown constants, and i and j = 1, 2, …, 12.

[0067] Beneficial effects of the present invention:

[0068] The present invention expands the application field of the SEIR model by innovatively introducing it into the analysis of the evolution of online public opinion, so that it can not only be used to study the spread of infectious diseases in biomedicine, but also to analyze the propagation of online public opinion information and the diffusion process of emotions. This innovation provides a new analytical perspective and tools, which can more accurately describe the development trend of online public opinion and provide a reliable theoretical basis and data support for the formulation of public opinion warning and response strategies.

[0069] The present invention constructs an emergency network public opinion early warning model based on the SEIR model and designs corresponding early warning rules and model operation paths. The model can dynamically evaluate public opinion risks and predict future development trends according to different stages of public opinion evolution, helping decision makers to issue early warning signals and take timely response measures. This dynamic early warning and real-time response mechanism greatly improves the efficiency of public opinion management, helps relevant departments to quickly adjust response strategies, and avoids the social impact caused by the spread of public opinion. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0071] Figure 1 A schematic diagram of an individual state transfer process according to an embodiment of the present invention;

[0072] Figure 2 It is a schematic diagram of the individual state transfer disassembly of an embodiment of the present invention;

[0073] Figure 3 A schematic diagram of the change of individual density in different states over time during the diffusion of network public opinion information in an embodiment of the present invention;

[0074] Figure 4 This is a schematic diagram of the SPN model for early warning of public opinion on sudden incidents according to an embodiment of the present invention;

[0075] Figure 5 A schematic diagram of a Markov chain isomorphic to the SPN model for early warning of online public opinion according to an embodiment of the present invention;

[0076] Figure 6 The figure is a flowchart of the prediction and early warning method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0077] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0078] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0079] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0080] like Figure 1-Figure 6 As shown, a method for predicting and warning the evolution of network public opinion of emergencies based on the SEIR model includes the following steps:

[0081] Step 1: Individual status division:

[0082] The online public opinion information that is constantly spreading and spreading in the Internet space does not exist in isolation. It is a combination of a certain event or topic and other events or topics derived from it. Due to the openness of the Internet space and the diversity of roles that individuals can play in the process of information flow, individuals are no longer passive recipients of information as in the traditional mass media system. They can more actively participate in the entire process of information flow.

[0083] Using the compartment model to divide individual states, individuals can usually be divided into four states during the spread of online public opinion information:

[0084] The first is the uninformed state (S), where they are temporarily not affected by the spread of online public opinion information and lack knowledge of online public opinion information, but are influenced by other related individuals in their social network and have the possibility of state transfer;

[0085] The second is the informed state (E). Individuals in this state have a certain understanding of online public opinion information, but due to their personal interests or other factors, they will not continue to spread it temporarily or even during the entire process of the spread of online public opinion information.

[0086] The third is the dissemination state (I). These individuals are the main promoters of the spread of online public opinion information. Under their influence, individuals in the uninformed state may be able to get rid of the uninformed state and enter the informed state or dissemination state;

[0087] The fourth is the removal state (R), in which individuals lose interest in the dissemination of a certain online public opinion information and are immune to related information.

[0088] Step 2: Individual state transfer process:

[0089] As the online public opinion of emergencies spreads and ferments, netizens will have different emotional reactions. The various emotional reactions will be affected by a variety of subjective and objective factors and undergo probability conversion. This kind of emotional conversion behavior of netizens conforms to the disease transmission characteristics of the SEIR model in the field of public health. According to the above-mentioned division of individual states and the characteristics of the SEIR model, the propagation path of public opinion in the crowd is S→E→I→R. The unknown S becomes the propagator I at a rate of α, and becomes the lurker E at a rate of 1-α. If the lurker E chooses to spread public opinion information, it will become the propagator I at a rate of β, otherwise it will become the immune R at a rate of ω, and the propagator I will become the immune R at a rate of γ. The propagation path is as follows: Figure 1 shown.

[0090] In the process of individual state transfer, this study sets the following assumptions:

[0091] 1. The network of online public opinion information diffusion is static. The statistical characteristics and topological structure of the network composed of individuals and their connections remain unchanged throughout the process of online public opinion information diffusion;

[0092] 2. After an individual publishes or forwards online public opinion information, his or her neighbors will be affected in the same way. In other words, there is no situation where the individual cannot receive the information due to absence or other reasons.

[0093] 3. Direct contact can only occur between individuals that are connected by edges, and individuals that are not connected by edges can only have indirect contact through the paths between them;

[0094] 4. Whether an individual's state transition occurs is closely related to his position in the network and the states of all individuals in his attention network;

[0095] 5. During the diffusion of online public opinion information, individuals in the removal state will not be affected by other factors and return to the three previous lag states of ignorance, awareness and dissemination.

[0096] Step 3: Establishment of online public opinion information diffusion model:

[0097] Based on the assumptions of information diffusion and state transition, a network public opinion diffusion model is established. Let the group size be N, and {1, 2, 3, ..., i, ..., n} be all its elements. These N individuals and the connections between them constitute a complex social network G (N, E). The edges connecting the nodes in the network G are all directed, that is, in the process of network public opinion information diffusion, individuals are only affected by their "following individuals" and can pass this influence to their "fan individuals". The in-degree of individual i in the network is represented by ki, which is the number of all the individuals followed by i. The in-degree distribution P (k) represents the proportion of individuals with in-degree k in all individuals.

[0098] Assume that the number of individuals in the uninformed state, informed state, propagation state and removal state at time t are S(t), E(t), I(t) and R(t) respectively, and S(t)+E(t)+I(t)+R(t)=N, then the proportion of individuals in each state in the group at that time can be expressed as s(t), e(t), i(t) and r(t), and s(t)+e(t)+i(t)+r(t)=1. When assuming that individuals with the same degree have the same behavior, the mean field method can be used to establish the following differential equations:

[0099]

[0100] Among them, the parameters α, 1-α, β, ω and γ correspond to the transition probabilities between states S→I, S→E, E→I, E→R, and I→R, respectively. It represents the probability that a node with in-degree k at time t has an edge coming from a propagation node.

[0101] like Figure 2 As shown, individuals in state S and E will only transition to another state when they come into contact with infected person I; individuals in state I will transition to state R with a certain probability, but the removed individual R will not transition to other states. The transition probabilities between the four states are calculated as follows:

[0102]

[0103] P i t(E→R)=1,m>M(5)

[0104]

[0105] In this model, Q i represents the number of people spreading public opinion in this social network; k i represents the number of people that individual i can contact and spread public opinion information to; m i represents the number of times individual i contacts the public opinion communicator; M represents the memory tolerance. i ≤M, the probability of individual i becoming a public opinion disseminator increases with m i increases with the increase of m i >M, individual i no longer moves to the public opinion dissemination state; b represents the social reinforcement effect, and the larger its value, the greater the transition from m i The greater the probability of transitioning to a public opinion disseminator;

[0106] According to the individual state transfer model in the diffusion of online public opinion information, an indicator that can measure the degree of diffusion of online public opinion information can be obtained through simple mathematical changes. i(t) is determined as the ratio of public opinion communicators to all individuals, and z(t) is the ratio of informed persons, communicators and removers to all individuals, which represents the diffusion range of online public opinion information. Its mathematical form is:

[0107] z(t)=e(t)+i(t)+r(t)=1-s(t)(7)

[0108] v is the average number of informed individuals added or the number of uninformed individuals reduced at each moment when the system runs to time T. Its mathematical form is:

[0109]

[0110] Δν t is the instantaneous diffusion velocity, and its mathematical form is:

[0111] Δν t = -ΔS(t) = S(t-1) - S(t)(9)

[0112] Step 4: The diffusion process of online public opinion information in complex social networks:

[0113] From a macroscopic perspective, the process of online public opinion information diffusion is manifested as the gradual expansion of the coverage of online public opinion information; from a microscopic perspective, this process is manifested as the continuous transfer of individual states. Based on the previously established online public opinion diffusion model, the regularity of state transfer in the process of online public opinion diffusion is analyzed. At t = 0, the individual ratios s(t), e(t), i(t) and r(t) of the uninformed state, informed state, propagation state and removal state in the network space are 0.98, 0.00, 0.02 and 0.00 respectively. The statistical characteristics of the individual network used in the simulation experiment and the main model parameters are given.

[0114] Table 1 Main parameters and explanations of the public opinion propagation model

[0115]

[0116]

[0117] Using the model parameters given in Table 1, taking the average of the results of multiple experiments, we can get Figure 3 The simulation results are shown in Figure 1. It can be seen that the process of network public opinion information diffusion ends when the number of individuals in the four states no longer changes. In the process of public opinion information diffusion, the number of unknown individuals in the network continues to decline, and eventually accounts for only 0% of the group; the number of exiting individuals continues to grow, and eventually reaches 99.89% of the total number of individuals; the number of individuals in the latent state and the propagation state experience a process of first growth and then decline, and the proportion of these two types of individuals eventually reaches 0% and 0.11%, respectively, and their peaks appear at t=8 and t=11, respectively, and the proportions at the peak moments are 51.33% and 35.64%, respectively. Therefore, public opinion information can be diffused over a large area in a relatively short period of time.

[0118] Step 5: Construction of emergency network public opinion warning model based on SPN:

[0119] The SEIR-SPN structural model is used to analyze the evolution of the network public opinion system of emergencies and the cross-influence between indicators in the early warning system. The transition in SPN is defined as an exponential distribution function in the Molloy form, as shown in formula (10):

[0120]

[0121] Among them, the parameter λ t represents the average rate of transition t, a random delay time variable x≥0.

[0122] SPN is generally defined as a directed graph containing six elements: SPN = (P, T, F, W, M, λ), where:

[0123] ①P={P1,P2,…,P n} is a finite set of key early warning factors, n>0 represents the total number of places;

[0124] ②T i ={t1, t2, ..., t k} represents the total number of changes in early warning factors;

[0125] ③ It is a set of transition input and output directed arc elements, representing the triggering warning conditions and warning results respectively;

[0126] ④W:F→N+ is the arc weight function, which assigns weights to directed arcs, N+={1, 2,…, n};

[0127] ⑤M:P→N is a set of identifiers, indicating possible states, and M1 is the initial identifier;

[0128] ⑥λ={λ1,λ2,…,λ n} is the set of transition firing rates, which is associated with the time transition. The time transition obeys a negative exponential distribution, and λ represents the parameter of the distribution function.

[0129] Emergency network public opinion SPN early warning model (such as Figure 4 The model consists of 9 network public opinion warning elements and 12 element transitions. The symbol definitions in the SPN structure model are shown in the following table:

[0130] Table 2 Symbol definitions in the SPN early warning model for emergency network public opinion

[0131]

[0132]

[0133] By using the SPN model of emergency network public opinion warning and the Markov chain isomorphism, the dynamic changes of the emergency network public opinion evolution system are analyzed. The initial mark is determined by the SPN model: M1 = (1, 1, 1, 0, 0, 0, 0, 0), and according to the triggering rules of SPN, the following reachable mark is established, see formula (11).

[0134]

[0135] Based on the reachable set, a Markov chain isomorphic to SPN is constructed, such as Figure 5 As shown in the figure, the directed arcs in the figure represent the transition process of the early warning SPN model from one state to another, and the label on each arc represents the average implementation rate λ of the corresponding transition of the arc. j .

[0136] Let P(M i)(i=1,2,…,12) is the M of the crisis warning SPN model in the stable state i Probability, then the steady-state probability P = (P(M1), P(M2), ..., P(M 12 ), according to the Markov stationary distribution theorem and the Chapman-Kolmogorov equation, we can get equation (12).

[0137]

[0138] Depend on Figure 5 The Markov chain and equation (12) shown above give the relationship between the state probabilities of the sudden incident network public opinion evolution system (13).

[0139]

[0140] Model operation path and warning activation rules:

[0141] Based on the construction of the above-mentioned SEIR-SPN emergency network public opinion evolution prediction and early warning model, the actual operation path and process of the model are designed according to the development process of the emergency network public opinion system.

[0142] Step 6: Early warning indicator risk rating:

[0143] Based on the weights of the early warning indicators of network public opinion of emergencies, the risk levels of early warning indicators are divided into Class I, Class II and Class III, and the severity is evaluated. The network analysis method improved by the hierarchical analysis method is used to calculate the weights and perform descending processing to obtain the risk rating of early warning indicators, see Table 3.

[0144] Table 3 Emergency network public opinion warning rating indicators

[0145]

[0146]

[0147] Step 7: Establish the SPN warning structure model:

[0148] By crawling Internet data, we can mine some open data of actual cases and obtain λ1, …, λ 12 There are 12 average implementation rate parameters in total to obtain the system status warning probability under specific circumstances:

[0149]

[0150] Where: P represents the probability of crisis warning P(M 12 ), ψ and l are unknown constants, and i and j = 1, 2, …, 12.

[0151] At the same time, the warning threshold is set to 0.1, and the warning rating rules are set as follows:

[0152] If the warning probability P∈[0.1,0.2] or reaches the warning critical surface, the blue warning is activated;

[0153] If the warning probability P∈(0.2,0.3] or reaches the warning critical surface, the yellow warning is activated;

[0154] If the warning probability P∈(0.3,0.4] or reaches the warning critical surface, the orange warning is activated;

[0155] If the warning probability P>0.4 or reaches the warning critical surface, the red warning will be activated.

[0156] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0157] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for predicting and warning the evolution of network public opinion of emergencies based on the SEIR model, characterized in that: The following steps are involved: S1, individual state division: the individuals involved in the diffusion of online public opinion information are divided into multiple states, including the uninformed state S, the informed state E, the propagation state I and the removal state R, where; The uninformed state S refers to individuals who have not yet been affected by online public opinion; The informed state E refers to individuals who understand online public opinion information but do not spread it; The dissemination state I is the individual that spreads the network public opinion information; The removed state R refers to individuals who have lost interest in online public opinion information and no longer spread it; S2, individual state transfer process: in the process of spreading online public opinion about emergencies, individuals in the uninformed state S change to the informed state E after receiving information, individuals in the informed state E choose to spread public opinion and become individuals in the spreading state I, and individuals in the informed state E who do not spread public opinion directly enter the removal state R, and individuals in the spreading state I lose their attention to public opinion after a predetermined period of time and eventually enter the removal state R; S3, Establishment of the online public opinion information diffusion model: Based on the SEIR model, the online public opinion diffusion model is constructed. The online public opinion diffusion network is assumed to be static, the group size is set to N, and the differential equation of individual state transfer is established using the mean field method. The proportion of public opinion communicators and informed persons is simulated according to the differential equation, and the scope and speed of online public opinion diffusion are calculated. S4, the process of online public opinion information diffusion in complex social networks: The process of public opinion diffusion is simulated through simulation experiments. The initial state is set as 98% of individuals are in an uninformed state and 2% of individuals are in a spreading state. As time goes by, the scope of public opinion diffusion gradually expands, and the proportion of spreading individuals first increases, then decreases, and finally disappears; S5, Construction of emergency network public opinion warning model based on SPN: The SEIR-SPN structure model is used to model the evolution and warning of emergency network public opinion, set the key elements and changes of warning, analyze the state transition in the process of public opinion development through Markov chain, and calculate the warning probability according to different states; S6, early warning indicator risk rating: The early warning indicators of network public opinion of emergencies are weighted by network analysis method improved by analytic hierarchy process, and the early warning risk levels are divided into Class I, Class II and Class III; S7, warning threshold and rating rule setting: according to the simulation results, the warning threshold is set to 0.1, and the warning levels are divided into blue, yellow, orange and red, and different levels of warnings are activated according to the rating rules.

2. According to the SEIR model-based method for predicting and warning the evolution of network public opinion of emergencies according to claim 1, it is characterized in that: The calculation of the transition probability of the individual state transition process in S2 includes: The transition probability from the uninformed state S to the informed state E is expressed as: Among them, Q i represents the number of people spreading public opinion in the social network, k i represents the number of people that individual i contacts and spreads public opinion information to, and α is the transfer parameter; The transition probability from the uninformed state S to the propagation state I is expressed as: The transition probability from the informed state E to the propagated state I is expressed as: Among them, b represents the social reinforcement effect, mi represents the number of times individual i contacts the public opinion communicator, and M represents the memory tolerance; The transition probability from the informed state E to the exit state R is expressed as: P i t (E→R)=1,m>M; The transition probability from the propagation state I to the exit state R is expressed as: Among them, γ is the removal rate.

3. According to claim 2, a method for predicting and warning the evolution of network public opinion of emergencies based on the SEIR model is characterized in that: The differential equation for the individual state transition in S3 is expressed as: Among them, α, 1-α, β, ω and γ correspond to the transition probabilities between states S→I, S→E, E→I, E→R and I→R respectively. It represents the probability that a node with in-degree k at time t has an edge coming from a propagation node.

4. According to claim 3, a method for predicting and warning the evolution of network public opinion of emergencies based on the SEIR model is characterized in that: The diffusion range of online public opinion is obtained by calculating the ratio of informed persons, disseminators and movers to all individuals, expressed as: z(t)=e(t)+i(t)+r(t)=1-s(t); Among them, z(t) is the ratio of informed individuals, disseminators and emigrants to all individuals, i(t) is the ratio of disseminators to all individuals, e(t) is the ratio of informed individuals to all individuals, r(t) is the ratio of emigrants to all individuals, and s(t) is the ratio of uninformed individuals to all individuals.

5. According to claim 4, a method for predicting and warning the evolution of network public opinion of emergencies based on the SEIR model is characterized in that: The calculation of the network public opinion diffusion speed is expressed as: Δν t =-ΔS(t)=S(t-1)-S(t); Where v is the average number of informed individuals added or the number of uninformed individuals reduced at each moment when the system runs to time T, Δν t is the instantaneous diffusion velocity.

6. According to claim 5, a method for predicting and warning the evolution of network public opinion of emergencies based on the SEIR model is characterized in that: The construction of the SPN-based emergency network public opinion early warning model in S5 includes: S51, define early warning elements: set the early warning key element set P of the network public opinion of the emergency event = {P1, P2, ..., P n }, including original information of public opinion events, harmfulness of public opinion information, authenticity of public opinion information, integration and processing of public opinion information, deviation of situation development, data sentiment analysis, deterioration of netizens' emotional behavior, activation of crisis warning, calming of public opinion and control of the situation; S52, set warning transition: define the transition in SPN as an exponential distribution function in Molloy form, expressed as: T={t1,t2,…,t k }; Among them, λ t represents the average rate of transition t, the random delay time variable x≥0, and T represents the total number of changes of warning elements; S53, Construction of warning input and output relationship: through directed arc element set It represents the input-output relationship in the warning. Each arc is defined as input or output. The arc weight function W:F→N+ assigns weights to the directed arcs. The weight value is N+={1, 2, …, n}. At the same time, the identification set M:P→N represents the state that occurs, and M1 is the initial identification. S54, calculation of transition rate: define λ as the transition rate, and calculate it using a negative exponential distribution function; S55, establishment of Markov chain: the SPN model is isomorphic to the Markov chain, the state transition diagram of the SPN model is constructed, and the Markov chain is used to analyze the dynamic changes of state transition during the development of public opinion.

7. According to claim 6, a method for predicting and warning the evolution of network public opinion of emergencies based on the SEIR model is characterized in that: The establishment of the Markov chain in S55 includes: S551, define state vector: define all state vectors M i , where M1,M2,...,M 12 They represent different evolution states of public opinion respectively, and the initial definition of the state vector is expressed as: S552, Steady-state constraints: Steady-state constraints are defined based on the Markov stationary distribution theorem and the Chapman-Kolmogorov equation and are expressed as: S553, Construction and solution of stationary distribution equation: According to the state vector M1,M2,...,M 12 and the corresponding state transition rates λ1,λ2,...,λ 12 , construct the stationary distribution equation, and calculate the stationary probability of each state, expressed as: S554, Application of Markov chain to analyze the evolution of public opinion: Utilize the calculation results of the Markov chain steady state to analyze the long-term evolution trend of network public opinion on emergencies.

8. According to claim 7, a method for predicting and warning the evolution of network public opinion of emergencies based on the SEIR model is characterized in that: The rating rules in S7 include: If the warning probability P∈[0.1,0.2] or reaches the warning critical surface, the blue warning is activated; If the warning probability P∈(0.2,0.3] or reaches the warning critical surface, the yellow warning is activated; If the warning probability P∈(0.3,0.4] or reaches the warning critical surface, the orange warning is activated; If the warning probability P>0.4 or reaches the warning critical surface, the red warning will be activated.

9. The method for predicting and warning the evolution of network public opinion of emergencies based on the SEIR model according to claim 8 is characterized in that: The warning probability is expressed as: Among them, P represents the crisis warning probability P(M 12 ), ψ and l are unknown constants, and i and j = 1, 2, …, 12.

Images