Information propagation analysis method based on three-dimensional network information propagation kinetic model

Abstract

The invention provides an information dissemination analysis method and system based on a three-dimensional network information dissemination dynamics model, and the method comprises the steps: representing multiple elements affecting information dissemination as a social network layer, an environment layer, a psychological layer and a viewpoint layer in a super network, and constructing a coupling network for the interaction phenomenon between users in the social network layer, therefore, a three-dimensional network of the super-network nested coupling network is generated, and an information propagation dynamics model is constructed by fusing rich propagation mechanisms on the basis of the network. A real information propagation case event is utilized to carry out parameter estimation, an information propagation index system is constructed, key factors influencing information propagation evolution are analyzed, finally, a theoretical framework is provided for accurately mastering an information propagation rule in the field of network information propagation analysis, and meanwhile, technical support and strategy support are provided for network information propagation coping and guiding.

Description

Technical Field

[0001] The present invention relates to the field of information dissemination technology, and more specifically, to an information dissemination analysis method and system based on a multi-element three-dimensional network information dissemination dynamics model of a real social network. Background Art

[0002] In recent years, various online social media have developed rapidly under the background of rapid development of mobile Internet technology. With the help of these platforms, the public can express their opinions relatively freely through text. At the same time, the barriers of various online social platforms are gradually eliminated. Netizens can express their opinions while obtaining information content on multiple social platforms. They can also conduct one-way or two-way cross-platform interactions through "forwarding" and "sharing". This has greatly reduced the threshold for ordinary people to express their opinions and express their opinions through the Internet, making the expression of public opinion smoother.

[0003] In addition, as an external driving force of the information dissemination process, the information dissemination environment has become more complex. In addition to the online and offline environments as external forces that drive the dissemination of network information, the impact of the offline information environment created by media and activities such as radio and television cannot be ignored. Corresponding to the external driving force is the internal driving force. This part focuses on the personal characteristics of users. The psychological and opinion characteristics of users will play an internal driving role in the dissemination of network information as internal forces. Due to differences in social relations, personal cognition, etc., netizens have different personal characteristics for information dissemination events, which has an increasingly significant impact on their information dissemination decisions. On the other hand, netizens' expressions have become more diverse, and the content generated by users themselves has become more personalized. Netizens' expression of opinions is no longer limited to neat literary terms. Users will create rich content such as buzzwords and emoticons, which further promotes the sustainable dissemination of network information.

[0004] From the overall evolution trend, the spread of network information has dynamic characteristics, and the spread mechanism is very important in forming the evolution pattern of public opinion. The factors that affect the process of network information spread are also diverse from the perspective of internal and external driving forces. The theory of network information dissemination dynamics takes dynamic systems as the research object and can well describe the dissemination mechanism, influencing factors, and dissemination laws. Therefore, exploring the laws of information dissemination based on this theory has become a popular research direction.

[0005] However, on the one hand, existing research has mostly taken a certain angle as the entry point, and has not comprehensively considered the comprehensive impact of multiple factors on information dissemination from multiple angles, so that the presented network structure is not clear enough and the consideration of the dissemination mechanism is relatively simple; on the other hand, most studies focus on building specific models according to research objectives, and have not yet conducted multi-scenario information dissemination research based on a general framework, resulting in multi-scenario simulations that are not comprehensive enough and cannot meet the needs of multi-scenario information dissemination laws; secondly, most current studies do not make high use of real data, and have not yet proposed an efficient and accurate parameter estimation method, which makes it inaccurate to grasp the laws of information dissemination in real scenarios and unable to use models to reproduce real information dissemination scenarios. Summary of the invention

[0006] In response to the above-mentioned problems existing in the existing network information propagation analysis, the present invention is based on the above-mentioned information propagation scenarios and technologies, and combines the information propagation dynamics, complex network and other theories to construct a three-dimensional network information propagation dynamics model based on multiple factors. The model is numerically fitted and analyzed based on real information propagation data, and the influence of model parameters on model indicators is studied through parameter change analysis technology. The information propagation laws are summarized, so as to formulate information propagation guidance and control strategies for general framework research.

[0007] According to one aspect of the present invention, a method for analyzing information dissemination based on a three-dimensional network information dissemination dynamics model is provided, which is used to simulate the individual state transition of netizens and the information dissemination evolution trend in network information dissemination events based on a preset three-dimensional network information dissemination dynamics model, wherein:

[0008] The structure of the three-dimensional network includes:

[0009] The social network layer is used to represent the directed complex network structure composed of the netizen nodes participating in the information dissemination process through a limited set of information dissemination subject nodes and directed edge relationships between nodes. Among them, the individual netizens on each social platform are nodes, and the edges between nodes represent the netizens' participation in the dissemination of information between users in the three-dimensional network information dissemination. The edges pointing to the user nodes represent the influence of other users on the user, and the edges radiating from the user nodes represent the user's ability to disseminate information outwards.

[0010] The environment layer is used to represent the information dissemination environment at the beginning, development, climax and end of the information dissemination event through a limited set of environment type nodes and directed edge relationships between the nodes; wherein the directed edges of the environment layer represent the temporal relationship of the release of environmental information, and the directed connections between multiple environment nodes represent that the environment continues to change;

[0011] The psychological layer is used to represent netizens' psychology towards the spread of information dissemination events and the transformation relationship between psychological types through a limited set of psychological type nodes and directed edge relationships between nodes. The psychological state transformation of the psychological layer is affected by the environmental layer;

[0012] The opinion layer is used to represent the transformation of netizens' opinions at different stages of information dissemination events through a limited set of opinion type nodes and directed edges between the nodes; the opinions in the opinion layer are divided into positive opinions and negative opinions according to the opinion theme and emotional tendency;

[0013] The hyperedge SE of the stereo network represents the subject a i In the external driving force of information dissemination environment m and the internal driving force of psychological state n Under the joint efforts of j , and finally m 、p n , k j Affects the information dissemination subject i Changes in the state of communication at the social network layer; the three-dimensional network information communication dynamics model takes the temporal relationship of the development process of the environmental layer events as the main line, takes the other factors of the environmental layer as a whole as the external driving force, takes the nested influence of the psychological layer and the opinion layer as the internal driving force, and is constructed based on the general information communication scenario formed by the evolution of the communication state of the user nodes at the social network layer under the joint action of the internal and external driving forces;

[0014] The method comprises:

[0015] Based on the selected information dissemination event, determining key time points in the development process of the information dissemination event;

[0016] Extracting the hot topic event information of the information dissemination event at the key time point and the forwarding information in the forwarding area of ​​the hot topic event information as the original dissemination data; wherein the forwarding information includes the forwarding text and forwarding time of the user forwarding the hot topic event information;

[0017] Preprocessing the original propagation data to obtain propagation valid data in the original propagation data;

[0018] Based on the effective dissemination data and the preset three-dimensional network information dissemination dynamics model, the dynamic process of group state transfer and information dissemination event development and evolution under the three-dimensional network of the social media platform is simulated.

[0019] On the other hand, the present invention also provides an information diffusion analysis system based on a three-dimensional network information diffusion dynamics model, which is used to perform information diffusion analysis using the information diffusion analysis method based on a three-dimensional network information diffusion dynamics model as described above, comprising:

[0020] The original communication data acquisition unit is used to determine the key time point in the development process of the information communication event based on the selected information communication event; and extract the hot topic event information of the information communication event at the key time point and the forwarding information in the forwarding area of ​​the hot topic event information as the original communication data; wherein the forwarding information includes the forwarding text and forwarding time of the forwarding user under the hot topic event information;

[0021] A data preprocessing unit, used for preprocessing the original propagation data to obtain propagation valid data in the original propagation data;

[0022] A simulation and analysis unit is used to simulate the dynamic process of group state transfer and information dissemination event development and evolution under the three-dimensional network of the social media platform based on the effective dissemination data and a preset three-dimensional network information dissemination dynamics model.

[0023] The above-mentioned information dissemination analysis method and system based on the three-dimensional network information dissemination dynamics model according to the present invention characterizes the multiple factors affecting information dissemination as the social network layer, environmental layer, psychological layer, and opinion layer in the super network, and constructs a coupling network for the interaction phenomenon between users in the social network layer, thereby generating a three-dimensional network of super network nested coupling networks, and integrates rich dissemination mechanisms on the basis of this network to construct an information dissemination dynamics model. Parameter estimation is carried out using real information dissemination case events, an information dissemination index system is constructed, and key factors affecting the evolution of information dissemination are analyzed, ultimately providing a theoretical framework for accurately grasping the laws of information dissemination in the field of network information dissemination analysis, and at the same time providing technical and strategic support for responding to and guiding network information dissemination.

[0024] In order to achieve the above and related purposes, one or more aspects of the present invention include features that will be described in detail later and are particularly pointed out in the claims. The following description and the accompanying drawings describe some exemplary aspects of the present invention in detail. However, these aspects indicate only some of the various ways in which the principles of the present invention can be used. In addition, the present invention is intended to include all of these aspects and their equivalents. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] By referring to the following description and claims in conjunction with the accompanying drawings, and with a more comprehensive understanding of the present invention, other objects and results of the present invention will become more apparent and easier to understand. In the accompanying drawings:

[0026] Figure 1 A schematic diagram of network information propagation according to an embodiment of the present invention;

[0027] Figure 2 A schematic flow chart of a method for performing information propagation analysis based on a three-dimensional network information propagation dynamics model according to an embodiment of the present invention;

[0028] Figure 3 A schematic diagram of the overall three-dimensional network structure of the MF-SLPCI model according to an embodiment of the present invention;

[0029] Figure 4 It is a schematic diagram of the node propagation state transfer process of the social network layer of the MF-SLPCI model according to an embodiment of the present invention;

[0030] Figure 5 A schematic diagram of a propagation cumulative amount and a propagation effective amount curve according to an embodiment of the present invention;

[0031] Figure 6 A schematic diagram of an information propagation cycle of a case event according to an embodiment of the present invention;

[0032] Figure 7 A schematic diagram of the distribution of opinions on forwarded texts on Weibo under selected information according to an embodiment of the present invention;

[0033] Figure 8 A schematic diagram of numerical fitting results of four selected pieces of information according to an embodiment of the present invention;

[0034] Fig. 9 A schematic diagram of an analysis of parameters related to environmental impact according to an embodiment of the present invention;

[0035] Fig.10 A schematic diagram of an analysis of parameters related to individual characteristics according to an embodiment of the present invention;

[0036] Fig.11 Schematic diagram of network structure related parameter analysis according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] In the following description, for the purpose of illustration, in order to provide a comprehensive understanding of one or more embodiments, many specific details are set forth. However, it is apparent that these embodiments may also be implemented without these specific details. In other examples, for ease of describing one or more embodiments, known structures and devices are shown in the form of block diagrams.

[0038] In response to the above-mentioned problems existing in the existing analysis of network information dissemination, the present invention is based on the dynamics theory and diversified information dissemination scenarios in the real world, takes the network information dissemination dynamics theory as the cornerstone, and based on internal and external driving factors such as social netizen interaction, netizen psychology, netizen views and information dissemination environment, integrates the impact of various real factors on information dissemination, and proposes a three-dimensional network information dissemination dynamics theory based on multiple factors.

[0039] The three-dimensional network information dissemination dynamics theory based on multiple factors provided by the present invention can study the internal and external driving forces that affect information dissemination in layers, and further combine external and internal driving forces to construct a three-dimensional network including social network layer, environmental layer, psychological layer and opinion layer. In addition, the dissemination mechanism is set based on individual factors, network structure factors, etc., and a general framework of the three-dimensional network information dissemination dynamics model based on multiple factors is constructed to describe the information dissemination process in complex systems. Using real information dissemination cases as data drive, the statistical analysis method is used to obtain the cumulative amount of population time series and fit the real data to verify the effectiveness of the model. After that, the influence of various physical factors on information dissemination is further considered from multiple perspectives, and the law of information dissemination is summarized, so as to provide decision support for relevant information guidance strategies.

[0040] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0041] The dissemination of online information under the new communication pattern can be summarized as: the sum of the opinions expressed by the information dissemination subject on the social platform under the influence of the internal driving force of psychology and the external driving force of the environment. Figure 1 It is a schematic diagram of network information dissemination, such as Figure 1 As shown in Figure 1, network information has developed rapidly under the influence of multiple factors such as communication carriers, external driving forces, internal driving forces, and forms of expression. At present, the research on the impact of various factors on information dissemination has been very extensive, but most studies have not comprehensively considered various factors to form a general framework suitable for rich and multi-scenario information dissemination cases.

[0042] Based on the internal and external driving forces and real data that drive netizens to participate in the development of information dissemination, the present invention constructs a three-dimensional network information dissemination dynamics model based on multiple factors to analyze the information dissemination law under the influence of multiple factors. By studying the three-dimensional network information dissemination dynamics model MF-SLPCI (Muti-factors Susceptible-Latent-Promoting-Countering-Immune) based on multiple factors (hereinafter referred to as the MF-SLPCI model or the three-dimensional network information dissemination dynamics model based on multiple factors), a general framework is built that can accurately simulate the individual state transition of netizens and the evolution trend of information dissemination in real network information dissemination events.

[0043] Figure 3 : is a schematic diagram of the overall three-dimensional network structure of the MF-SLPCI model according to an embodiment of the present invention. As the barriers between social platforms are gradually eliminated, netizens can express their own opinions while obtaining information content on multiple social platforms, and can also conduct one-way or two-way cross-platform interactions through "forwarding", "sharing", "cooperation" and other methods. At the same time, netizens are not only affected by the social platform environment and friends in the process of information dissemination, but also by the external information dissemination environment created by media and activities such as radio and television. From the perspective of individual netizens, the psychological and opinion characteristics of netizens as internal forces also have a driving effect on the dissemination of network information. Due to differences in circle culture, social relations and other aspects, netizens have different views on information events, and the influence of psychological and opinion characteristics of netizens on their information dissemination decisions is becoming increasingly significant.

[0044] like Figure 3 As shown in the figure, the multi-factor information dissemination scenario under the joint action of internal and external driving forces can be described as follows: the information dissemination subject, information dissemination environment information, netizen psychological type, and netizen opinions on the social network can be abstracted into nodes in the social network layer, environmental layer, psychological layer, and opinion layer in the super network (three-dimensional network), among which the connection between homogeneous nodes constitutes a sub-network, and the connection between heterogeneous nodes constitutes the connection between sub-networks. In the three-dimensional network, the social network layer represents the dissemination status attributes of netizens, the environmental layer represents the information dissemination environment attributes of netizens, the psychological layer represents the psychological state attributes of netizens, and the opinion layer represents the opinion expression attributes of netizens.

[0045] Specifically, as an example, in the MF-SLPCI model, the specific structure of each layer is as follows:

[0046] (1) Social Network Layer

[0047] The social network layer can be regarded as a directed complex network structure composed of Internet user nodes participating in the information dissemination process, which can be expressed as Among them, A n={a1, a2, ..., a m} represents a finite set of information dissemination subject nodes, Represents directed edge relationships between nodes. In the social network layer, nodes represent individual netizens on each platform. In social platforms, the edges between nodes represent netizens participating in the three-dimensional network information dissemination. The dissemination of information between users is regarded as edges. The edges pointing to user nodes represent the influence of users by other users, and the edges radiating from user nodes represent the ability of users to disseminate information outward.

[0048] (2) Environmental layer

[0049] The node types in the environment layer comprehensively represent the information dissemination environment at the beginning, development, climax and end of the event, which can be expressed as S E =(E, E E→E ), where E = {e1, e2, ..., e n} represents a finite set of environment type nodes, E E→E Represents the directed edge relationship between nodes. The directed edges in the environment layer represent the temporal relationship of the release of environment information, and the directed edges between multiple environment nodes represent the continuous changes in the environment.

[0050] (3) Psychological level

[0051] The psychological layer can be represented as S P =(P, E P→P ), where P = {p1, p2, ..., p n} is a set of finite psychological type nodes, E P→P is the directed edge relationship between nodes. The directed edges between nodes in this layer represent the transformation relationship between psychological types. i Psychological transformation into j There is a certain probability value in psychology, and this value is affected by other layers. Each node type represents the psychology of netizens' "following the crowd", "reservation" and "opposition" to the spread of information dissemination events. The psychological state change of this psychological layer is also affected by the environmental layer.

[0052] (4) Viewpoint

[0053] The viewpoint layer can represent S K =(K, E K→K ), where K = {k1, k2, ..., k n} is a set of finite viewpoint type nodes, E K→KIt is a directed edge between nodes. In the process of information dissemination, the public can express their own views and opinions and express their emotions through text. Directed edges are used to represent the transformation of netizens' views at different stages of information dissemination events. The types of views can be divided by combining the theme of the view and the emotional tendency: Class A represents a positive emotional tendency, which means that netizens support the view; Class B represents a negative emotional tendency, which means that netizens oppose the view.

[0054] Overall, the three-dimensional network hyperedge contains four types of heterogeneous nodes, and the hyperedge SE represents the subject a i In the external driving force of information dissemination environment m and the internal driving force of psychological state n Under the joint efforts of j , and finally m 、p n , k j Affects the information dissemination subject i The change of the propagation state at the social network layer can be expressed by formula (1):

[0055] SE = {a i , D{e m , p n}, k j |θ(a i , k j )=1,θ(a i , p n )=1,θ(a i , e m )=1} (1)

[0056] Based on the above analysis, the three-dimensional network of this embodiment can be expressed as F = {S A , S E , S P , S K , SE}. Based on the three-dimensional network constructed above, this embodiment first takes the temporal relationship of the development process of environmental layer events as the main line, and considers other factors of the environmental layer as the external driving force, and the nested influence of the psychological layer and the opinion layer as the internal driving force. Under the joint action of internal and external driving forces, the propagation state evolution process of user nodes in the social network layer is affected, and then the MF-SLPCI model is constructed based on this general scenario of information propagation.

[0057] Figure 2 The flow chart of the information propagation analysis method based on the three-dimensional network information propagation dynamics model according to an embodiment of the present invention is shown. Figure 2As shown, on the basis of constructing the above-mentioned three-dimensional network information propagation dynamics model, the information propagation analysis method based on the three-dimensional network information propagation dynamics model provided by the present invention simulates the individual state transition of netizens and the information propagation evolution trend in the network information propagation event based on the pre-constructed three-dimensional network information propagation dynamics model, and specifically includes the following steps:

[0058] S210: Based on the selected information dissemination event, determining a key time point in the development process of the information dissemination event;

[0059] S220: extracting the hot topic event information of the information dissemination event at the key time point and the forwarding information in the forwarding area of ​​the hot topic event information as the original dissemination data; wherein the forwarding information includes the forwarding text and forwarding time of the forwarding user under the hot topic event information; the hot topic event information may be the first release information released by the topic host under the hot topic;

[0060] S230: pre-processing the original propagation data to obtain propagation valid data in the original propagation data;

[0061] S240: Based on the effective dissemination data and the preset three-dimensional network information dissemination dynamics model, simulate the dynamic process of group state transfer and information dissemination event development and evolution under the three-dimensional network of the social media platform.

[0062] The following first describes in detail the pre-constructed three-dimensional network information propagation dynamics model in a more specific implementation manner.

[0063] In the social network layer, the microblog users who participate in the information dissemination process are taken as nodes. According to the role of individuals in information dissemination, the nodes can be mainly divided into the following six categories: susceptible S, lurkers L, supporters P, opponents C and immune I. Under the influence of factors such as three-dimensional network structure, influencing factors, dissemination mechanism, and interaction rules between individuals, the node status in the social network layer will change. In the topological structure of the social network layer, different types of nodes play different roles in the process of information dissemination, and the social relationships between nodes provide channels and ways for them to interact, such as Figure 4 shown.

[0064] according to Figure 4 The node propagation state transition process of the social network layer of the MF-SLPCI model shown in the figure can be described as follows based on the multi-factor three-dimensional network state transition process:

[0065] Assume that information is propagated in a closed and stable environment, and only consider the people that can be reached during the process of information propagation. This embodiment only focuses on the information diffusion generated by the forwarding behavior of individuals for each piece of information. It is assumed that the same user can choose whether to forward the information after reading the information, and a user only forwards each piece of information once. At any time, any individual in the crowd may be in six states with respect to the information, namely, Susceptible (abbreviated as S), latent (abbreviated as L), supporting (abbreviated as P), opposing (abbreviated as C), and immune (abbreviated as I). This embodiment introduces the in-degree k of the three-dimensional network node, that is, the number of edges pointing to the node in the three-dimensional network, and the out-degree l of the three-dimensional network node, that is, the number of edges issued by each node in the three-dimensional network, and classifies the people in each state according to the in-degree and out-degree of different nodes in the three-dimensional network, where:

[0066] S (k，l) (t) represents the total number of individuals in a susceptible state with in-degree k and out-degree l at time t. This group has not yet been exposed to the information but will have the opportunity to be exposed in the future and is susceptible to the information and may generate forwarding behavior;

[0067] L (k，l) (t) represents the total number of individuals in a latent state with in-degree k and out-degree l at time t. This group has been exposed to the information but has not yet spread it;

[0068] P (k，l) (t) represents the total number of individuals in the state of supporting forwarding with in-degree k and out-degree l at time t. This group has forwarded the information and their opinions are consistent with those of the original information. They are still in the exposure period of the information and are able to make individuals who are susceptible to the information know the content of the information and may generate forwarding behavior;

[0069] C (k，l) (t) represents the total number of individuals with in-degree k and out-degree l who are in the state of opposing forwarding at time t. This group has forwarded the information and their opinions are inconsistent with the original information. They are still in the exposure period of the information and are able to make individuals who are susceptible to the information know the content of the information and may generate forwarding behavior;

[0070] I (k，l)(t) represents the total number of individuals who are immune to the information with in-degree k and out-degree l at time t. This group consists of two parts: one is the total number of individuals who have forwarded the information and have exceeded the exposure period of the information over time and are no longer able to influence other users to access the content of the information and generate forwarding behavior; the other is the total number of individuals who are in the susceptible state, latent state, support forwarding state and oppose forwarding state, and are directly transformed into individuals immune to the information due to their subjective lack of interest in the information.

[0071] Therefore S ( k, l ) (t), L ( k, l ) (t), P ( k, l ) (t), C ( k, l ) (t), I ( k, l ) (t) represents the number of groups with in-degree k and out-degree l at time t, and the sum of the total number of groups at time t is N(t), which is a function that changes with time.

[0072] The MF-SLPCI model constructed in this embodiment is:

[0073]

[0074] p=p P +p C (7)

[0075]

[0076] The initial value of each state is defined as: S (k，l) (0) = S0, L (k，l) (0) = L0, P (k，l) (0) = C P (0), C (k，l) (0) = C C (0),I (k，l) (0)=0.

[0077] Among them, m((i, j)|(k, l)) represents the conditional probability that any edge in a directed network is sent from a node with degree (i, j) and points to a node with degree (k, l). (i.e. p P ) represents the probability that a node user with degree (k, l) is connected to a user in the forwarding state at time t, (i.e. p C) represents the probability of a node user with a degree of (k, l) being connected to a user in the anti-forwarding state at time t, and p represents the probability of being connected to a user in the forwarding state. For a multi-layer stereoscopic network, the directed network is a layer in the stereoscopic network, and the above specific state transition formula is for the user node relationship of the social network layer in the stereoscopic network. In this embodiment, assuming that the directed network is an unrelated network, the conditional probability is only related to the degree of the upstream node (i, j) with in-degree i and out-degree j, which can be written as:

[0078] m((i,j)|(k,l))=jm(i,j) / <l>(10)

[0079] Among them, (i, j) represents user nodes with different in-degree i and out-degree j, corresponding to a node with in-degree i and out-degree j respectively; m(i, j) represents the joint probability distribution of nodes (i, j) in the network, that is, the probability of randomly selecting a node with in-degree i and out-degree j in the network, which is also equal to the total number of nodes N with in-degree k and out-degree l in the network. kl The ratio to the total number of network nodes N, <l>represents the average degree of a directed network; then the degree distribution of any point in the network with degree (k, l) is:

[0080]

[0081] Because in any directed network, each edge leaves one node and enters another node, so in a directed network, the average in-degree is equal to the average out-degree, and the formula is expressed as:

[0082]

[0083] It can be simplified to:

[0084]

[0085] They represent the probability of a node user with degree (k, l) connecting to a user in the state of supporting forwarding or opposing forwarding in a unit time. The explanations of the other parameters are shown in Table 1 below.

[0086] Table 1 Explanation of parameters of the stereo network model

[0087]

[0088]

[0089] In this embodiment, it is assumed that all new users who migrate into the dynamic system are susceptible users. Therefore, the rate of change of the number of susceptible users reflected in formula (1) contains an additional term, the average migration rate Λ. In addition, this section assumes that all groups leave the dynamic system at an average migration rate u. When susceptible users S and hesitant users L are affected by supporters P and opponents C, there will be four state transition paths, namely: becoming hesitant users L, supporters P, opponents C, and immune users I; when supporters P and opponents C are affected by supporters P and opponents C, there will be three state transition paths, namely: becoming supporters P, opponents C, and immune users I. In the process of the state transition of netizens, in addition to the influence of the social network layer, the environmental layer, psychological layer, and opinion layer will also determine the communication state of netizens. The specific three-dimensional network parameter settings are shown in formulas (14)-(24), the relevant parameters involved are shown in Table 2, and the function setting mechanism is shown in Table 3:

[0090] f S→L =τ×(1-e -δ×n ) (15)

[0091]

[0092] f P→I (t) = 1-f P→C (t) (22)

[0093] f C→I (t) = 1-f C→P (t) (23)

[0094] f S→I (t) = 1-f S→L (t)-f S→P (t)-f S→C (t) (24)

[0095] f L→I (t) = 1-f L→P (t)-f L→C (t) (25)

[0096] Table 2 Definition of stereo network parameters

[0097]

[0098] Table 3 Definition of stereo network time-varying parameter settings

[0099]

[0100]

[0101] Specifically, in the dynamic process of this three-dimensional network communication, a user who is susceptible to information will be influenced by k individuals on average per unit time. Since in a degree-independent directed network, the conditional probability of a user with degree (k, l) being connected to another user who supports / opposes forwarding is m((i, j)|(k, l)), the probability of a user with degree (k, l) in a susceptible state contacting a user in a forwarding state per unit time is k(p P +p C ), when kpS (k，l) (t) users were exposed to supporters / opponents, where:

[0102] kpf S→L (t)S (k，l) (t) susceptible individuals will continue to pay attention to the information, that is, they will move from the S state to the L state;

[0103] kpf S→P (t)S (k，l) (t) susceptible individuals will hold a view consistent with the original information and support the forwarding information, that is, move from the S state to the P state;

[0104] kpf S→C (t)S (k，l) (t) susceptible people will forward the information with an opposing attitude that is inconsistent with the original information, that is, they will move from the S state to the C state;

[0105] The remaining susceptible people are not interested in the information, that is, there is kpf S→I (t)S (k，l) (t) susceptible individuals move from state S to state I.

[0106] During the incubation period, there will be θL (k，l) (t) lurkers increase their willingness to spread information and thus generate forwarding behavior, that is, f L→P (t)θ(t)L (k，l) (t) lurkers move from state L to state P, f L→C (t)θ(t)L (k，l) (t) lurkers move from state L to state C; the remaining f L→I (t)θ(t)L (k，l) (t) lurkers lose interest in the information and move from state L to state I. For the forwarder, when a supporter P / opponent C is influenced by a supporter P / opponent C, they will have kpf P→C (t)P (k，l) (t) supporters move from state P to state C kpf C→P (t)C (k，l) (t) opponents will move from state C to state P, and the other part of supporters P and opponents C will be kpf C→I (t)P (k，l) (t), kpf C→I (t)P (k，l) (t) The number of transitions from the P / C state to the I state, and the remaining α P (t)P (k，l) (t) / α C (t)C (k，l) (t) supporters / opponents no longer have the ability to influence others because they have passed the active exposure period, that is, they move from the P / C state to the I state.

[0107] The cumulative amount of forwarding can be directly obtained and calculated from the network communication platform. In the three-dimensional network information communication dynamics model, the change rate of the cumulative amount of forwarding by users who support forwarding and users who oppose forwarding with degree (k, l) over time is:

[0108]

[0109] The cumulative number of retweets by supporters and opponents is C P (t) and C C (t) are:

[0110]

[0111] In the process of information dissemination, how to measure the effect of public information dissemination and the degree of evolution of opinions is a major focus of the present invention in studying the evolution and dissemination laws of network information dissemination. In a specific embodiment of the present invention, the cumulative amount of group dissemination and the effective amount of group dissemination are used to characterize the dissemination situation from the two aspects of dissemination breadth and dissemination heat. Among them, the cumulative amount of group dissemination represents the total number of people who have disseminated information in the system as of a certain moment, and the effective amount of group dissemination represents the number of people who have disseminated information and still have influence in the system as of a certain moment. Therefore, the curve that characterizes the changes of these two types of quantities over time can be defined as a dissemination curve. Figure 5 The following propagation information can be obtained from the support / opposition diagram of the propagation cumulative amount and propagation effective amount curve shown:

[0112] (1) Peak Information Dissemination Heat

[0113] Information dissemination heat peak value P max / C max It plays an important role in describing the heat of the forwarding group information, indicating the maximum number of groups that still have the ability to influence others, corresponding to the peak of the effective amount of propagation curve P(t) / C(t). max / C max The larger it is, the higher the popularity of information dissemination.

[0114] (2) Final scale of information dissemination

[0115] The final scale of information dissemination It plays an important role in describing the breadth of group communication, indicating the total number of groups in the F state during the entire process of information dissemination, corresponding to the final value of the cumulative amount propagation curve P(t) / C(t). The larger it is, the wider the scope of information dissemination is in the process of information dissemination and development, which supports (opposes) the forwarder.

[0116] (3) Peak time of information dissemination

[0117] Information dissemination peak time t max Indicates the moment when the information reaches the highest heat, satisfying the condition P(t max )=P max , C(t max )=C max .

[0118] like Figure 5 As shown, its cumulative quantity propagation curve C is similar to an "S" curve, which is specifically characterized by a trend of first rising rapidly, then rising slowly, and finally stabilizing, while its effective quantity propagation curve F is similar to a bell curve (normal curve), which is specifically characterized by a trend of first rising and then falling.

[0119] Through the above specific embodiments of constructing a three-dimensional network information dissemination dynamics model, it can be seen that the present invention is based on dynamics theory and diversified information dissemination scenarios in the real world. It takes the network information dissemination dynamics theory as the cornerstone, and based on the internal and external driving force factors such as social netizens' interaction, netizens' psychology, netizens' opinions and information dissemination environment, integrates the influence of various real factors on information dissemination, and proposes a three-dimensional network information dissemination dynamics theory based on multiple factors. This theory studies the internal and external driving forces that affect information dissemination in layers, and further combines external driving forces, internal driving forces, and expression methods to construct a three-dimensional network including social network layers, environmental layers, psychological layers, and opinion layers. In addition, a dissemination mechanism is set based on individual factors, network structure factors, etc., and a general framework of a three-dimensional network information dissemination dynamics model based on multiple factors is constructed to describe the information dissemination process in complex systems. With real information dissemination cases as data drive, the statistical analysis method is used to obtain the cumulative amount of population time series and fit the real data to verify the effectiveness of the model. After that, the influence of various physical factors on information dissemination is further considered from multiple perspectives, and the law of information dissemination is summarized, so as to provide decision support for the government's relevant information dissemination guidance strategy.

[0120] On the basis of the above-mentioned pre-constructed three-dimensional network information dissemination dynamics model, the individual state transition of netizens and the information dissemination evolution trend in network information dissemination events are simulated. First, based on the selected information dissemination event, the key time points in the development process of the information dissemination event need to be determined, and the hot topic event information of the information dissemination event at the key time points and the forwarding information in the forwarding area of ​​the hot topic event information are extracted as the original dissemination data. This process corresponds to the above-mentioned steps S210 and S220.

[0121] Specifically, as an example, an information dissemination event is selected, and through the relevant analysis of the preset information analysis website "Zhiwei Event", combined with the trend of the event heat map and the popularity and data volume of the corresponding Weibo information, the key time points in the development process of the information dissemination case are selected, such as Figure 6 As shown in the red box in . Through the application programming interface (API) open to the public on the social microblog platform, a total of 4 hot topic host release information at key time nodes in this case were collected, namely: Case information 1 of the starting event topic at 10:02 on June 9, 2023, Case information 2 at the development stage of information dissemination at 16:07 on June 9, 2023, Case information 3 at the further development stage of information dissemination at 11:24 on June 11, 2023, and finally Case information 4 at 21:34 on June 11, 2023. For a specific message, relevant data is collected through the API (Application Programming Interface) interface open to the public on the social microblog platform, including the content of the message copy and the forwarding time of each user, etc.

[0122] This embodiment selects hot topic event information at key information dissemination development time points within 60 hours from the beginning of the case information dissemination to the end of the event heat, and further collects forwarding information in the forwarding area, including the forwarding text and forwarding time of the user who forwarded the information, to provide data support for the effectiveness verification of the verification model.

[0123] Then, the original propagation data is preprocessed to obtain propagation valid data in the original propagation data. This process corresponds to the above step S230.

[0124] Specifically, as an example, the present invention performs data preprocessing for each single information data collected at the key nodes in the process of information dissemination, which includes three steps. First, due to physiological factors, users are mainly active in browsing information during the day, and the night data of the original data is processed accordingly to avoid the limitation of information stagnation caused by physiological needs. Secondly, due to the randomness of user forwarding information, the data is filtered again to delete records that are irrelevant and meaningless to the content related to the forwarded copy information. Finally, the SnowNLP library is used to quantify the forwarded text opinion value under four messages to realize the division of supporters and opponents in the SLPCI model. The specific analysis results are as follows: Figure 7 As shown in the figure, the information forwarders corresponding to the texts in the interval [0-0.5] are classified as opponents, and the information forwarders corresponding to the texts with the quantitative opinion value results between [0.6-1] are classified as supporters. Figure 7 The results show that the opinions of the information forwarders at the four key nodes of information dissemination are concentrated in the interval of [0.6-0.7]. This is also related to the nature of the event itself. The information publishers at each time point are all official accounts and the information content is relatively positive and has a certain reference value, so more information forwarders forward with a supportive attitude. After these preprocessing steps, effective dissemination data is obtained, that is, the current forwarding time point and content of supporters and opponents in each message. These results are used to calculate the cumulative forwarding number. In this embodiment, the start time is set to 0 and the sampling frequency is set to one hour.

[0125] Then, the dynamic process of group state transfer and information dissemination event development and evolution under the three-dimensional network can be simulated based on the effective dissemination data and the preset three-dimensional network information dissemination dynamics model. This process corresponds to the above-mentioned step S240, reflecting the group dissemination state transfer and information dissemination event development and evolution dynamics under the influence of the social media platform network environment and individual factors of netizens.

[0126] In this embodiment, the forwarding accumulation amount, an information propagation indicator, is first used to conduct a numerical fitting experiment to estimate the model parameters. Then, by analyzing the overlap between the real data and the model fitting effect and the corresponding error indicators, it is determined whether the model is suitable for simulating the information propagation process, thereby verifying the effectiveness of the model.

[0127] In the numerical fitting experiment, nonlinear least squares (LS) is used for parameter estimation, and numerical methods are used to find the optimal solution for parameter estimation and data fitting. In this process, the initial value vector of the parameters to be estimated in the three-dimensional network information propagation dynamics model proposed in the present invention is set to Θ = (τ, δ, n, η P ,λ P , η C ,λ C , K max , L max , Λ,u,θ,α P , α C , S0, L0, N). The LS error function is obtained as

[0128]

[0129] in, Represents information C under the condition of parameter vector Θ P The numerical result of (t) corresponds to formula (27), Indicates the actual cumulative number of reposts by supporters; Represents information C under the condition of parameter vector Θ C The numerical result of (t) corresponds to formula (28), represents the actual cumulative amount of forwarding by supporters; w is the sampling time, w=0, 1, 2, ...T.

[0130] In order to evaluate the numerical fitting effect, this embodiment introduces the mean absolute percentage error (MAPE) to evaluate the goodness of fit. The result range of MAPE is [0, +oo]. A result of 0% indicates a perfect model, and a result of MAPE greater than 100% indicates a poor model. Therefore, the closer the result of this indicator is to 0, the better the performance of the model is, which can be expressed by formulas (31)-(32):

[0131]

[0132] Where m is the number of sample points in the forwarding data set supporting each case, n is the number of sample points in the forwarding data set opposing each case, and y i(P / C) is the observed value of the ith sample point, Yes i(P / C) The predicted value of .

[0133] Figure 8 (a)-(d) respectively give the schematic diagrams of the numerical fitting results of the MF-SLPCI model for information dissemination event cases 1, 2, 3, and 4, and the corresponding model parameter estimation results are shown in Tables 3-4. Among them, the red asterisk and the blue asterisk represent the true values ​​of those who support the forwarding and those who oppose the forwarding, respectively, and the red solid line and the blue solid line represent the simulated values ​​of the cumulative forwarding amount of those who support the forwarding and the cumulative forwarding amount of those who oppose the forwarding obtained by the model calculation, respectively. From the numerical simulation results, the curve of the case fitting using the MF-SLPCI model almost coincides with the real data points. In addition, according to the calculation, the MAPE values ​​of the four cases are shown in Table 4 and are all close to 0. Therefore, it can be seen that the MF-SLPCI model constructed by the present invention shows a good fitting performance. This result confirms the rationality of the three-dimensional network propagation mechanism under the influence of multiple factors formulated in the model, that is, the MF-SLPCI model can effectively simulate the information dissemination process and has the ability to reproduce the dynamics of information dissemination, which verifies the effectiveness and feasibility of the model.

[0134] Table 4 Evaluation index of fitting results of case 1-4 (in percentage)

[0135]

[0136] First, from Figure 8 The fitting results of the four key time periods of the development of the event show that the final scale of the supporter information dissemination in Case 2 is The information content involved the educational background and gender of the parties involved, which was likely to cause a "one-sided" phenomenon on social networking platforms. In this case, the voices supporting the school investigation were higher, which in turn attracted a lot of attention and discussion from netizens. In addition, the final scale of information dissemination of the opponents in Case 3 was This is the climax of the event, and the final scale of information dissemination by its opponents is Larger, and each message has more supporters. is greater than the final scale of information dissemination by opponents This is also related to Figure 6 The results of the opinion tendency analysis on real data are consistent, indicating that the opinions of most netizens will be influenced by the information released by the topic host and maintain opinions consistent with the information to spread information.

[0137] Table 5 Parameter estimation results for Case 1

[0138]

[0139] Table 6 Parameter estimation results for Case 2

[0140]

[0141] Table 7 Parameter estimation results for Case 3

[0142]

[0143] Table 8 Parameter estimation results of Case 4

[0144]

[0145] Secondly, from the perspective of environmental impact, the number of environmental media reports n and the environmental media influence δ of the four case events are relatively close, that is, the media environment that affects the development of information dissemination throughout the process is a similar atmosphere, and there is no reversal in the development process of the entire event. The overall tone of information dissemination has been presented from the beginning of the event, but in Case 4, the environmental media influence δ is smaller than that of the previous Cases 1, 2 and 3. At this time, the development of the event has come to an end, and the influence of the media has also weakened accordingly; from the perspective of the environmental impact factor τ, its size relationship is Case 2>Case 4>Case 1>Case 3, which is also consistent with the reading and discussion volume of topics at different stages.

[0146] From the perspective of individual attributes, the parameter fitting results of the four cases can all show that the probability of retaining the supported view η P are greater than the probability of retaining opposing views η C And the average forwarding probability of supporters λ P are greater than the average forwarding probability λ of opponents C , indicating that in the entire development process of this information dissemination event, netizens are more inclined to support the information released by the topic host and further spread the supportive views, which is consistent with the relevant discussion in the previous article; from the perspective of the lurkers' immune rate θ, the size relationship is case 3> case 1> case 2> case 4, which highlights the outbreak speed of each piece of information, that is, case 3 has the fastest outbreak speed. From the perspective of the immune rate of supporters and forwarders, the immune rate α of supporters in the four cases is P are greater than the opponent's α C And the α of Case 4 P and α C The values ​​are greater than those of other cases, that is, supporters are more likely to lose interest in information and exit the active state of forwarding information more quickly. In the final stage of information dissemination development, it is less attractive to netizens than in the first three stages, and they are more likely to exit the active state of forwarding information regardless of their views.

[0147] In addition, from the perspective of network structure, the parameter K m The larger the value, the faster the information bursts. From the fitting results of the four cases, we can see that K m The value relationship is Case 3 > Case 1 > Case 2 > Case 4, which is consistent with the corresponding relationship of the average transfer rate of the lurkers mentioned above. m The maximum value, K m It is the maximum value of the node degree in the network, representing the maximum number of people a user can contact during the information dissemination process. It is an important parameter reflecting the user characteristics in network information dissemination. Parameter K m The larger it is, the information publisher in Case 3 has more fans or greater influence than other information publishers, or there are other influential opinion leaders who forward the information after it is released, so that the information can reach more users during the dissemination process, resulting in a rapid outbreak of information.

[0148] Based on the existing fitting results of actual case events, we further analyze the relevant parameters involved in the model and the real physical factors they map, specifically analyzing the environmental impact factor τ, environmental media impact δ, the number of environmental media n, and the probability of retaining supporting opinions η. P , the average forwarding probability λ of supporters P , the probability of retaining opposing views η C , the average forwarding probability of opponents λ C , the initial value of the total susceptible population S0, the initial value of the total latent population L0, the average transfer rate q of the latent population, the total number of users N, and the maximum value of the node in-degree K in the network m and the maximum out-degree of the node L m These thirteen parameters respectively affect the cumulative amount of information forwarding C P(C) Since this model is a general framework for studying different information dissemination cases, the parameter impacts of different case fitting results are similar in the following analysis. This section only takes Case 3 as an example for single parameter change analysis.

[0149] Environmental impact analysis

[0150] like Fig. 9 As shown in (a)-(f) in the figure, the parameters of environmental impact factors related to physical factors are mapped, namely, the environmental impact factor τ, the environmental media impact factor δ, and the number of environmental media n on C P(C) and F P(C) They have similar effects, and the same parameter has a greater impact on the cumulative forwarding volume and instantaneous forwarding volume of supporters than on opponents. Specifically, as the parameters τ, δ, and n increase, the instantaneous forwarding volume curve becomes steeper and the peak value is higher. Similarly, the increase of these three parameters will accelerate the speed of information outbreak and the cumulative forwarding volume of users will reach a larger final scale. Further comparative analysis of the impact of each parameter on the same indicator, respectively analyzing Figures (a), (c), (e) and (b), (d), (f), parameters τ, δ and parameter n have a greater impact on the information dissemination indicator. The smaller the parameter value changes, the smaller the final scale of information dissemination caused by the changes. Information propagation peak value P max / C max The larger the range of change, the more important it is influencing information dissemination in terms of accelerating the speed of information outbreak, expanding information coverage, and enhancing netizens' information awareness. It is also an important role that cannot be ignored by relevant departments in the process of information dissemination governance.

[0151] Overall, such factors will mainly affect the peak value, scale and speed of information dissemination. Therefore, we can cooperate with mainstream media and new media platforms in the development of relevant information dissemination events, guide the direction of public opinion through news reports, commentary articles, etc., promote positive publicity and reduce negative impacts. In addition, we can also improve the public media literacy through relevant publicity and education activities, and through publicity and education activities, let the public view and disseminate information more rationally and reduce the spread of false information.

[0152] Individual characteristics impact analysis

[0153] like Fig.10 As shown in (a)-(f) in the figure, the parameters of the physical factors related to the individual characteristics of netizens are mapped to support the probability of retaining the opinion η P , the average forwarding probability λ of supporters P , the probability of retaining opposing views η C , the average forwarding probability of opponents λ C , the initial value of the total susceptible population S0, the initial value of the total latent population L0, the average transfer rate q of the latent population, the total number of users N to C P(C) and F P(C) has a similar effect. Specifically, as the parameter η P(C) , P(C) , S0, and L0 increase, the instantaneous forwarding curve of supporters (opponents) becomes steeper, and the information transmission peak value P(C) max Higher, similarly, the increase of these two parameters will accelerate the speed of information explosion and the cumulative forwarding volume of supporters (opponents) will reach a larger final scale. In addition, the increase of parameter S0 will also increase the time t when the user's instantaneous forwarding volume reaches the peak. max Compared with the above parameters, parameter N has a greater impact on the degree and effect of C. P(C) and F P(C) It has a relatively small negative impact. As N increases, C P(C) and F P(C) The increase in the size of the crowd will also increase the difficulty of information diffusion. From the perspective of the average transfer rate q of the latent group, its main impact is that as the parameter increases, the instantaneous forwarding volume decreases to 0 earlier, that is, the information dissemination reaches a stable scale earlier. When comparing and analyzing the changes of the same parameters within the same range, compared with opponents, it has a greater impact on the changes in the cumulative forwarding volume and instantaneous forwarding volume of supporters, and a relatively greater impact on the speed of information explosion in the same period of time, which is consistent with the above conclusions. Further comparative analysis of each parameter, respectively Fig.10 (a), (c), (e), (g), (i), (k) and (b), (d), (f), (h), (j), (l) are analyzed. For the same index, the parameter η P(C) and λ P(C) It has a greater impact on the cumulative amount of forwarding and the instantaneous amount of forwarding, followed by S0, L0 and q, and N is the least. This also provides a certain direction for information dissemination governance.

[0154] Overall, such factors will mainly affect the peak value, scale, information burst speed, time when the instantaneous forwarding volume reaches the peak value, and time when the final scale of information dissemination reaches stability. Therefore, information dissemination governance can start from such factors. According to the susceptibility, probability of retaining opinions, forwarding probability and other characteristics of different groups, targeted publicity strategies can be formulated to carry out targeted information delivery to increase the coverage and influence of positive information. In addition, according to the forwarding probability of different groups and the number of susceptible people, information dissemination can be managed in a hierarchical manner, high-risk information dissemination events can be discovered and warned in a timely manner, and corresponding response measures can be taken. Through online and offline activities, the media literacy of the public can be improved, enabling them to make rational judgments and share information, reducing the number of susceptible people, and thus managing and guiding the direction of information dissemination in a more scientific and effective manner.

[0155] Network structure impact analysis

[0156] like Fig.11 As shown in (a)-(d), the maximum node in-degree K m and the maximum out-degree of the node L m F P(C) and C P(C) has a similar effect. m and L m As the number of nodes increases, the instantaneous forwarding volume curve becomes steeper, the peak value becomes higher, and the burst speed becomes larger. Similarly, the final scale of the cumulative forwarding users will increase accordingly as these two parameters increase. That is, the larger the out-degree, the more information can be spread to other nodes. Generally, it has a greater dissemination ability in the network and can more effectively spread information to a wider group. The higher the in-degree, the stronger the ability to receive information from other nodes. Generally, it can obtain new information in the network faster, thus affecting whether it can become the source or disseminator of information. Further, by analyzing Figures (a) and (c), and Figures (b) and (d), we can deduce that parameter K m With L m In comparison, K m The change of parameters has a greater impact on the cumulative forwarding amount and the instantaneous forwarding amount, and the final scale variation range of the cumulative amount and the peak variation range of the instantaneous forwarding amount are larger. These rules are of great significance in analyzing or designing information dissemination strategies.

[0157] Overall, such factors will mainly affect the peak value, scale and speed of information dissemination. Relevant departments can start with such factors and monitor the information dissemination of key nodes in real time through in-degree and out-degree analysis. They can strengthen supervision on nodes with high in-degree and high out-degree, require them to exercise self-discipline, avoid spreading false information, malicious rumors and other harmful information, promptly discover and warn of events that may cause information dissemination risks, and take corresponding measures to intervene and deal with them.

[0158] It can be seen from the above embodiments that the information dissemination analysis method based on the three-dimensional network information dissemination dynamics model provided by the present invention constructs a general framework of a three-dimensional network information dissemination dynamics model based on multiple factors driven by internal and external driving forces, and divides various factors affecting information dissemination into social network layer, environmental time series layer, psychological layer, and opinion layer. The internal and external driving forces distributed in each layer are quantified into nested coupling functions and embedded into the differential equation representing the information dissemination dynamics of the social network layer. The time sequence of the environmental layer is the main line, and the cumulative amount of case information users forwarding at multiple key time points of information dissemination in the development process of a certain information dissemination event is numerically fitted, which verifies the effectiveness and rationality of the model, indicating that the constructed MF-SLPCI model can capture the dynamic process of information dissemination and information dissemination evolution. Further, by constructing information dissemination indicators to measure the dissemination and evolution effects, and then through single parameter change analysis experiments, the factors affecting information dissemination are analyzed from the perspectives of environmental influence, individual characteristic influence, and network structure influence, which fully shows that the internal and external driving factors are closely related to the dynamic dissemination evolution process of information. Among them, the degree and strength of environmental influence, the probability of users retaining opinions and the average forwarding probability, the total number of susceptible groups, the total number of latent groups and the transfer rate, and the node in-degree and node out-degree in the network are crucial to the final scale of information dissemination and the speed of information explosion.

[0159] In summary, the present invention integrates internal and external driving forces and quantifies them into coupled and nested time-varying parameters, constructs a multi-factor information propagation dynamics model, and analyzes the physical factors mapped by the model parameters based on the evolutionary propagation characteristics of network information propagation, revealing the objective effect of multiple factors on information propagation and information propagation evolution, so that relevant departments can formulate corresponding measures to more specifically manage social network information propagation and reduce the spread of bad information.

[0160] Corresponding to the above-mentioned information propagation analysis method based on the three-dimensional network information propagation dynamics model, the present invention also provides an information propagation analysis system based on the three-dimensional network information propagation dynamics model, which is used to perform information propagation analysis using the above-mentioned information propagation analysis method based on the three-dimensional network information propagation dynamics model, including:

[0161] The original communication data acquisition unit is used to determine the key time point in the development process of the information communication event based on the selected information communication event; and extract the hot topic event information of the information communication event at the key time point and the forwarding information in the forwarding area of ​​the hot topic event information as the original communication data; wherein the forwarding information includes the forwarding text and forwarding time of the forwarding user under the hot topic event information;

[0162] A data preprocessing unit, used for preprocessing the original propagation data to obtain propagation valid data in the original propagation data;

[0163] A simulation and analysis unit is used to simulate the dynamic process of group state transfer and information dissemination event development and evolution under the three-dimensional network of the social media platform based on the effective dissemination data and a preset three-dimensional network information dissemination dynamics model.

[0164] For more specific implementations of the information diffusion analysis system based on the three-dimensional network information diffusion dynamics model of the present invention, please refer to the aforementioned description of the information diffusion analysis method based on the three-dimensional network information diffusion dynamics model, which will not be described in detail here.

[0165] As described above with reference to the accompanying drawings, the information propagation analysis method and system based on the three-dimensional network information propagation dynamics model according to the present invention are described in an exemplary manner. However, those skilled in the art should understand that various improvements can be made to the information propagation analysis method and system based on the three-dimensional network information propagation dynamics model proposed by the present invention without departing from the content of the present invention. Therefore, the protection scope of the present invention should be determined by the content of the attached claims.< / l> < / l>

Claims

1. An information dissemination analysis method based on a three-dimensional network information dissemination dynamics model is used to simulate the individual state transition of netizens and the information dissemination evolution trend in network information dissemination events based on a preset three-dimensional network information dissemination dynamics model, wherein: The structure of the three-dimensional network includes: The social network layer is used to represent the directed complex network structure composed of the netizen nodes participating in the information dissemination process through a limited set of information dissemination subject nodes and directed edge relationships between nodes. Among them, the individual netizens on each social platform are nodes, and the edges between nodes represent the netizens' participation in the dissemination of information between users in the three-dimensional network information dissemination. The edges pointing to the user nodes represent the influence of other users on the user, and the edges radiating from the user nodes represent the user's ability to disseminate information outwards. The environment layer is used to represent the information dissemination environment at the beginning, development, climax and end of the information dissemination event through a limited set of environment type nodes and directed edge relationships between the nodes; wherein the directed edges of the environment layer represent the temporal relationship of the release of environmental information, and the directed connections between multiple environment nodes represent that the environment continues to change; The psychological layer is used to represent netizens' psychology towards the spread of information dissemination events and the transformation relationship between psychological types through a limited set of psychological type nodes and directed edge relationships between nodes. The psychological state transformation of the psychological layer is affected by the environmental layer; The opinion layer is used to represent the transformation of netizens' opinions at different stages of information dissemination events through a limited set of opinion type nodes and directed edges between the nodes; the opinions in the opinion layer are divided into positive opinions and negative opinions according to the opinion theme and emotional tendency; The hyperedge SE of the stereo network represents the subject a i In the information dissemination environment, this external driving force m and the internal driving force of psychological state n Under the joint efforts of j , and finally m 、p n , k j Affects the information dissemination subject i Changes in the state of communication at the social network layer; the three-dimensional network information communication dynamics model takes the temporal relationship of the development process of the environmental layer events as the main line, takes the other factors of the environmental layer as a whole as the external driving force, takes the nested influence of the psychological layer and the opinion layer as the internal driving force, and is constructed based on the general information communication scenario formed by the evolution of the communication state of the user nodes at the social network layer under the joint action of the internal and external driving forces; The method comprises: Based on the selected information dissemination event, determining key time points in the development process of the information dissemination event; Extracting the hot topic event information of the information dissemination event at the key time point and the forwarding information in the forwarding area of ​​the hot topic event information as the original dissemination data; wherein the forwarding information includes the forwarding text and forwarding time of the user forwarding the hot topic event information; Preprocessing the original propagation data to obtain propagation valid data in the original propagation data; Based on the effective dissemination data and the preset three-dimensional network information dissemination dynamics model, the dynamic process of group state transfer and information dissemination event development and evolution under the three-dimensional network of the social media platform is simulated.

2. The information diffusion analysis method based on the three-dimensional network information diffusion dynamics model according to claim 1 is characterized in that: Assuming that information is propagated in a closed and stable environment, only the people that can be reached during the information propagation process are considered. The node states in the three-dimensional network information propagation dynamics model include susceptible state S, latent state L, support state P, opposition state C and immune state I. The in-degree k and out-degree l of the three-dimensional network node are defined. According to the in-degree and out-degree of different nodes in the three-dimensional network, the people in each state are classified, including: S (k,l) (t) represents the total number of individuals with in-degree k and out-degree l in the susceptible state at time t; L (k,l) (t) represents the total number of individuals in the latent state with in-degree k and out-degree l at time t; P (k,l) (t) represents the total number of individuals with in-degree k and out-degree l in the state of supporting forwarding at time t; C (k,l) (t) represents the total number of individuals with in-degree k and out-degree l in the state of opposing forwarding at time t; I (k,l) (t) represents the total number of individuals in immune state with in-degree k and out-degree l at time t; The sum of the total number of each group at time t is a function N(t) that changes with time; Then, the three-dimensional network information propagation dynamics model is: p=p P +p C (7) Among them, f X→Y represents the three-dimensional network influence parameter of individuals migrating from state X to state Y, Λ represents the average migration rate, u represents the average migration rate, q represents the average transfer rate of lurkers, k represents the in-degree of the node in the three-dimensional network, l represents the out-degree of the node in the three-dimensional network, α P represents the average immunization rate of the support forwarder, α C represents the average immunization rate against forwarders; The initial value of each state is defined as: S (k,l) (0)==S0, L (k,l) (0) = L0, P (k,l) (0) = C P (0), C (k,l) (0) = C C (0) I (k,l) (0) = 0; m((i,j)|(k,l)) represents the conditional probability that any edge in a directed network is sent from a node with degree (i,j) and points to a node with degree (k,l). represents the probability that a node user with degree (k, l) is connected to a user in the forwarding state at time t, represents the probability of a node user with degree (k, l) connecting to a user in the anti-forwarding state at time t, and p represents the probability of connecting to a user in the forwarding state.

3. The information propagation analysis method based on the three-dimensional network information propagation dynamics model as claimed in claim 2 is characterized in that: Assuming that the stereo network is an unrelated stereo network, the conditional probability is only related to the degree of the upstream node (i, j) with in-degree i and out-degree j, which can be written as: m((i,j)|(k,l))=jm(i,j) / <l> (10)< / l> Where i and j represent the in-degree and out-degree of different nodes, (i, j) represents the user nodes with different in-degree i and out-degree j, and m(i, j) represents the joint probability distribution of nodes (i, j) in the network, which is equal to the total number of nodes N with in-degree k and out-degree l in the three-dimensional network. kl The ratio to the total number of network nodes N, <l> represents the average degree of the stereo network, and the degree distribution of any point with degree (k, l) in the stereo network is:< / l> 4. The information diffusion analysis method based on the three-dimensional network information diffusion dynamics model as claimed in claim 3 is characterized in that: In the stereoscopic network, the average in-degree is equal to the average out-degree.

5. The information diffusion analysis method based on the three-dimensional network information diffusion dynamics model as claimed in claim 4, wherein: In the three-dimensional network information dissemination dynamics model, in the process of the transfer of netizens' status, the social network layer environment layer, psychological layer, and opinion layer will affect the dissemination status of netizens. The three-dimensional network parameter settings are shown in the following formula: f S→L =τ×(1-e -δ×n ) (15) f P→I (t)=1-f P→C (t) (22) f C→I (t)=1-f C→P (t) (23) f S→I (t)=1-f S→L (t)-f S→P (t)-f S→C (t) (24) f L→I (t)=1-f L→P (t)-f L→C (t) (25); Among them, τ represents the environmental impact factor, δ represents the environmental media impact, n represents the number of environmental media, η P The probability of retaining a supporting view, λ P represents the average forwarding probability of supporters, η C represents the probability of retaining opposing views, λ C represents the average forwarding probability of opponents.

6. The information diffusion analysis method based on the three-dimensional network information diffusion dynamics model as claimed in claim 5, wherein: In the three-dimensional network, the probability that a user with degree (k, l) in the susceptible state contacts a user in the supporting forwarding state per unit time is k(p P +p C ), when kpS (k,l) (t) users were exposed to supporters / opponents, where: kpf S→L (t)S (k,l) (t) susceptible people will continue to pay attention to the information and move from the S state to the L state; kpf S→P (t)S (k,l) (t) susceptible individuals will hold a view consistent with the original information and support the forwarding information, moving from the S state to the P state; kpf S→C (t)S (k,l) (t) susceptible people will forward the information with an opposing attitude that is inconsistent with the original information, and move from the S state to the C state; The rest of the susceptible people are not interested in the information, and there are kpf S→I (t)S (k,l) (t) susceptible individuals move from state S to state I.

7. The information diffusion analysis method based on the multi-factor three-dimensional network information diffusion dynamics model as claimed in claim 6, wherein: In the stereo network, during the incubation period, With θL (k,l) (t) lurkers increase their willingness to spread information and thus generate forwarding behavior, f L→P (t)θ(t)L (k,l) (t) lurkers move from the L state to the P state; There is L→C (t)θ(t)L (k,l) (t) lurkers move from L state to C state; The rest L→I (t)θ(t)L (k,l) (t) lurkers lose interest in the information and move from the L state to the I state.

8. The information propagation analysis method based on the three-dimensional network information propagation dynamics model as claimed in claim 7, wherein: In the three-dimensional network, for the forwarder, when a supporter P / opponent C is influenced by a supporter P / opponent C, they will have kpf P→C (t)P (k,l) (t) supporters move from state P to state C kpf C→P (t)C (k,l) (t) opponents will move from state C to state P, and the other part of supporters P and opponents C will be kpf C→I (t)P (k,l) (t), kpf C→I (t)P (k,l) (t) is transferred from the P / C state to the I state, and the remaining αP(t)P (k,l) (t) / α C (t)C (k,l) (t) supporters / opponents move from P / C state to I state because they have passed the active exposure period and are no longer able to influence others.

9. The information propagation analysis method based on the three-dimensional network information propagation dynamics model according to claim 8, characterized in that: In the multi-factor three-dimensional network information dissemination dynamics model, the forwarding cumulative amount is directly obtained from the network dissemination platform, and the change rate of the forwarding cumulative amount of the supporting forwarding users and the opposing forwarding users with degree (k, l) over time is: The cumulative number of retweets by supporters and opponents is C P (t) and C C (t) are:

10. An information dissemination analysis system based on a three-dimensional network information dissemination dynamics model, characterized in that: The method for performing information dissemination analysis using the information dissemination analysis method based on the three-dimensional network information dissemination dynamics model as claimed in any one of claims 1 to 9 comprises: The original communication data acquisition unit is used to determine the key time point in the development process of the information communication event based on the selected information communication event; and extract the hot topic event information of the information communication event at the key time point and the forwarding information in the forwarding area of ​​the hot topic event information as the original communication data; wherein the forwarding information includes the forwarding text and forwarding time of the forwarding user under the hot topic event information; A data preprocessing unit, used for preprocessing the original propagation data to obtain propagation valid data in the original propagation data; A simulation and analysis unit is used to simulate the dynamic process of group state transfer and information dissemination event development and evolution under the three-dimensional network of the social media platform based on the effective dissemination data and a preset three-dimensional network information dissemination dynamics model.

Images