Social information networking monitoring and early warning method and system

Abstract

The invention relates to a social information networking monitoring and early warning method and system, and relates to the technical field of information early warning, and the method comprises the steps: collecting and analyzing the multi-source information of an information source, and determining the information source weight of the information source; according to feature vectors and feature weights extracted from the multi-source information by the distributed semantic model, obtaining semantic similarity of cross-time semantics; obtaining an event association strength network according to the information source weight and the semantic similarity; according to the distribution characteristics of the depth of each network node in the event association strength network, obtaining a risk propagation index of the influence of the assessment information propagation path on risk amplification; according to the risk propagation index, the information source weight, the semantic similarity and the association strength among the events in the event association strength network, generating an early warning comprehensive index of social information networking monitoring; and according to a preset threshold interval, determining an early warning grade result corresponding to the early warning comprehensive index. According to the invention, the accuracy and timeliness of social information networking monitoring and early warning can be improved.

Description

Technical Field

[0001] The present invention relates to the field of information early warning technology, and in particular to a social information network monitoring and early warning method, system, electronic device and non-transient computer-readable storage medium. Background Art

[0002] Nowadays, with the rapid development of information technology and the Internet, the speed of generation, dissemination and feedback of social information has increased significantly, and data-based social monitoring and early warning technologies have gradually been widely used.

[0003] However, traditional methods often rely on fixed rules and static models, making it difficult to respond in real time to emerging information sources and changes in semantic expressions, resulting in delayed warnings or missed reports. In addition, the lack of an effective network correlation analysis mechanism makes it difficult to construct the overall event propagation chain and impact path from scattered information nodes, affecting the system's perception and judgment of risk evolution trends. Summary of the Invention

[0004] In response to the technical problems existing in the prior art, the present invention provides a social information network monitoring and early warning method, system, electronic device and non-transitory computer-readable storage medium that can improve the accuracy and timeliness of social information network monitoring and early warning.

[0005] The technical solution of the present invention to solve the above technical problems is as follows:

[0006] The present invention provides a social information network monitoring and early warning method, the method comprising:

[0007] Collecting and analyzing multi-source information of information sources to determine information source weights of the information sources;

[0008] Obtaining semantic similarity of cross-temporal semantics based on feature vectors and feature weights extracted from the multi-source information by a distributed semantic model;

[0009] Obtaining an event association strength network according to the information source weights and the semantic similarity;

[0010] According to the distribution characteristics of the depth of each network node in the event correlation strength network, a risk propagation index for evaluating the impact of information propagation paths on risk amplification is obtained;

[0011] Generate a comprehensive early warning index for social information network monitoring based on the risk propagation index, the information source weight, the semantic similarity, and the correlation strength between events in the event correlation strength network;

[0012] According to the preset threshold range, the warning level result corresponding to the warning comprehensive index is determined.

[0013] Optionally, collecting and analyzing multi-source information of information sources to determine information source weights of the information sources includes:

[0014] Collect the historical credibility, information dissemination frequency and influence index of the information source;

[0015] Normalization is performed based on the historical credibility of the information source, the frequency of information dissemination, and the basic indicators corresponding to the influence index to obtain the normalized results of the above three indicators;

[0016] The normalized processing results of the three indicators are combined with the weighting coefficients to perform linear combination to obtain the information source weight of the information source.

[0017] Optionally, the information source weight is expressed as:

[0018]

[0019] Among them, W s is the information source weight, T h is the historical credibility of the information source, T max is the maximum value of historical credibility, F c is the frequency of information dissemination, F max is the maximum transmission frequency among all information sources, R i is the influence index of the information source, R max is the maximum value of the information influence index, α, β, and γ are the first, second, and third weighting coefficients, respectively.

[0020] Optionally, obtaining the semantic similarity of cross-temporal semantics by extracting feature vectors and feature weights from the multi-source information according to the distributed semantic model includes:

[0021] Perform weighted summation on the feature vectors to form the basic similarity;

[0022] Obtaining the time span of the multi-source information and a time decay coefficient representing the degree of influence of the time span on the semantic similarity;

[0023] The basic similarity is dynamically downgraded according to the time span and the time decay coefficient to obtain the semantic similarity of the cross-time semantics.

[0024] Optionally, the semantic similarity is expressed as:

[0025] S d =∑(V i W i )·exp(-λ|t i -t j |);

[0026] Among them, S d is the semantic similarity, V i is the semantic feature vector of the i-th feature, W i is the feature weight of the i-th feature, t i and t j is the timestamp of the semantic information, and λ is the time decay coefficient.

[0027] Optionally, obtaining an event association strength network according to the information source weights and the semantic similarity includes:

[0028] Multiplying the credibility weight and the semantic similarity to obtain a fusion result;

[0029] Detect the degree of deviation between the co-occurrence information of each event and the benchmark information, and obtain the logarithmic adjustment term of the co-occurrence intensity;

[0030] The event association strength network is generated according to the fusion result and the logarithmic adjustment item of the co-occurrence strength.

[0031] Optionally, obtaining a risk propagation index for evaluating the impact of information propagation paths on risk amplification based on the distribution characteristics of the depth of each network node in the event correlation strength network includes:

[0032] Constructing a depth diffusion factor according to the diffusion coefficient and the depth of the network nodes;

[0033] quantifying the cumulative effect of the correlation strength in the propagation path according to the event correlation strength and the depth diffusion factor;

[0034] According to the time influencing factor and the propagation time of the event, a time growth function including both time and space factors is constructed;

[0035] The risk propagation index is determined according to the cumulative effect and the time growth function.

[0036] Optionally, generating a comprehensive early warning index for social information network monitoring based on the risk propagation index, the information source weight, the semantic similarity, and the correlation strength between events in the event correlation strength network includes:

[0037] Obtaining the fusion result obtained by multiplying the credibility weight and the semantic similarity;

[0038] Obtaining an adjustment coefficient for adjusting the early warning comprehensive index;

[0039] The risk propagation index, the fusion result and the event association strength are weighted and summed using the adjustment coefficient to obtain the early warning comprehensive index.

[0040] Optionally, determining the warning level result corresponding to the warning comprehensive index according to a preset threshold range includes:

[0041] When the comprehensive warning index is greater than or equal to the first threshold, the corresponding warning level result is a high-level warning;

[0042] When the comprehensive warning index is greater than or equal to the second threshold and less than the first threshold, the corresponding warning level result is a medium warning;

[0043] When the comprehensive warning index is less than the third threshold, the corresponding warning level result is low risk and no warning is issued temporarily;

[0044] The first threshold is greater than the second threshold, and the second threshold is greater than the third threshold.

[0045] The present invention also provides a social information network monitoring and early warning system, the system comprising:

[0046] A weight determination module, configured to collect and analyze multi-source information of information sources and determine information source weights of the information sources;

[0047] A semantic processing module, configured to obtain semantic similarity of cross-temporal semantics based on feature vectors and feature weights extracted from the multi-source information by a distributed semantic model;

[0048] An association strength module, configured to obtain an event association strength network based on the information source weights and the semantic similarity;

[0049] A risk index module is used to obtain a risk propagation index for evaluating the impact of information propagation paths on risk amplification based on the distribution characteristics of the depth of each network node in the event correlation intensity network;

[0050] An early warning index module, configured to generate an early warning comprehensive index for social information network monitoring based on the risk propagation index, the information source weight, the semantic similarity, and the correlation strength between events in the event correlation strength network;

[0051] The monitoring and warning module is used to determine the warning level result corresponding to the warning comprehensive index according to a preset threshold range.

[0052] In addition, to achieve the above-mentioned purpose, the present invention also proposes an electronic device, comprising: a memory for storing computer software programs; a processor for reading and executing the computer software programs, thereby realizing a social information network monitoring and early warning method as described above.

[0053] In addition, to achieve the above-mentioned purpose, the present invention also proposes a non-transitory computer-readable storage medium, in which a computer software program is stored. When the computer software program is executed by a processor, it implements a social information network monitoring and early warning method as described above.

[0054] The beneficial effects of the present invention are:

[0055] (1) The present invention introduces a dynamic weight model for information sources to achieve a comprehensive evaluation of historical performance, dissemination frequency, and influence. It can adapt to the rapid changes of emerging information sources, avoid a one-size-fits-all approach of trusting or rejecting certain types of information, improve the overall quality of information input, and lay a highly reliable foundation for subsequent analysis.

[0056] (2) The present invention integrates semantic features with time decay factors to construct dynamic semantic similarity, effectively distinguishes "repeated information" from "new variant information", suppresses historical redundant interference, improves semantic recognition accuracy, and enhances the system's response sensitivity to time-sensitive events.

[0057] (3) The present invention constructs an event correlation strength matrix based on information co-occurrence, semantic similarity and source weight, which can realize the visual analysis of the event evolution path and causal context, and shift from "isolated information" to "event map", facilitating source tracing and diffusion prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 A flowchart of a social information network monitoring and early warning method provided by the present invention;

[0059] Figure 2 A schematic diagram of the structure of a social information network monitoring and early warning system provided by the present invention;

[0060] Figure 3 A schematic diagram of the hardware structure of a possible electronic device provided by the present invention;

[0061] Figure 4 A schematic diagram of the hardware structure of a possible computer-readable storage medium provided by the present invention. DETAILED DESCRIPTION

[0062] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0063] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the specified features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0064] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or illustration". Any embodiment of the present invention described as "for example" is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes are not elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed herein.

[0065] See also Figure 1 , provides a flowchart of a social information network monitoring and early warning method of the present invention, comprising the following steps:

[0066] Step 201: Collect and analyze multi-source information of information sources, and determine the information source weights of the information sources.

[0067] In some embodiments, step 201 may include:

[0068] Collect the historical credibility, information dissemination frequency and influence index of the information source;

[0069] Normalization is performed based on the historical credibility of the information source, the frequency of information dissemination, and the basic indicators corresponding to the influence index to obtain the normalized results of the above three indicators;

[0070] The normalized processing results of the three indicators are combined with the weighting coefficients to perform linear combination to obtain the information source weight of the information source.

[0071] In some embodiments, the information source weight is expressed as:

[0072]

[0073] Among them, W s is the information source weight, T h is the historical credibility of the information source, T max is the maximum value of historical credibility, Fc is the frequency of information dissemination, F max is the maximum transmission frequency among all information sources, R i is the influence index of the information source, R max is the maximum value of the information influence index, α, β, and γ are the first, second, and third weighting coefficients, respectively.

[0074] In practice, this formula is used to calculate the source weight of an information source, reflecting its credibility and importance in the monitoring and early warning system. Dynamic credibility assessment of information from different sources ensures that subsequent processing (such as event correlation and risk propagation analysis) is based on high-quality information.

[0075] The first part is the historical credibility index T h It is the credibility of the information source in history (such as historical accuracy, low error rate, etc.). max This is the most trustworthy of all information sources and is used for normalization. α adjusts the importance of this factor in the overall weight (e.g., a higher value can be used to prioritize historically reliable sources). The more stable and accurate a source has been in the past, the higher its score in this factor.

[0076] The second part is the frequency of transmission index F c It is the frequency of dissemination of the information source in a certain period of time. max The highest dissemination frequency among all information sources is used for normalization. β is the adjustment coefficient for this term. A high dissemination frequency may indicate high activity and attention for the information source, but it may also introduce noise. Therefore, the weight (β) of this term should be set carefully based on actual needs.

[0077] The third part is the impact indicator R i It is the influence index of the information source (such as forwarding volume, audience size, citation rate, etc.). max is the maximum influence value (among all sources). γ is the adjustment coefficient for this term. Sources with greater influence should be given more weight, but excessive bias towards sources with high influence but low accuracy should also be avoided. The sum of α, β, and γ is 1, ensuring that the weighted sum of the three factors remains within the standardized range (between 0 and 1), facilitating consistent use in subsequent models.

[0078] In summary, the present invention takes into account the historical performance (credibility), current activity level (dissemination frequency), and scope of influence (information influence) of the information source, and achieves flexible adjustment through normalization and weight factors. It has good scalability and adaptability and is a key step in building a dynamic credibility model.

[0079] Step 202: Obtain semantic similarity of cross-temporal semantics based on feature vectors and feature weights extracted from the multi-source information by a distributed semantic model.

[0080] In some embodiments, step 202 may include:

[0081] Perform weighted summation on the feature vectors to form the basic similarity;

[0082] Obtaining the time span of the multi-source information and a time decay coefficient representing the degree of influence of the time span on the semantic similarity;

[0083] The basic similarity is dynamically downgraded according to the time span and the time decay coefficient to obtain the semantic similarity of the cross-time semantics.

[0084] In some embodiments, semantic similarity can be expressed as:

[0085] S d =∑(V i W i )·exp(-λ|t i -t j |);

[0086] Among them, S d is the semantic similarity, V i is the semantic feature vector of the i-th feature, W i is the feature weight of the i-th feature, t i and t j is the timestamp of the semantic information, and λ is the time decay coefficient.

[0087] In practice, this formula measures the semantic similarity between two or more pieces of information and dynamically adjusts it based on time. This serves as a crucial foundation for subsequent analysis steps, such as event correlation and risk propagation. By comprehensively considering the semantic similarity of information content and temporal proximity, the dynamic semantic matching degree between a set of information is determined.

[0088] V i is the semantic feature vector of the i-th feature, representing the numerical vector representation of the i-th semantic feature in the information text (such as the output from models such as Word2Vec, BERT, and TF-IDF). It can be a numerical vector of a keyword, phrase, syntactic structure, semantic role, etc., representing the "content itself."

[0089] W i is the feature weight of the i-th feature, which expresses its importance in the overall semantic similarity. It can be set based on factors such as feature frequency, information entropy, and sentiment intensity to control the weight differences across different semantic dimensions.

[0090] t i and t j is the timestamp of the semantic information, and is the semantic feature V in the current information. i The time at which the semantic feature corresponding to the reference information appears. This is used to measure the time interval between two pieces of information, thereby introducing a time decay factor. This represents "temporal distance" and controls the timeliness of information association.

[0091] λ is the time decay coefficient, which controls the influence of time difference (time span) on semantic similarity. λ>0. Larger values indicate a greater time distance between pieces of information and a faster decay of semantic similarity. This can typically be set based on domain characteristics or adaptively achieved through learning mechanisms. This is a means of adjusting the model's sensitivity to "real-time" requirements.

[0092] V i W i Used to measure the static semantic contribution value of the i-th semantic dimension, ∑(V i W i ) is the basic similarity formed by weighted summation. exp(-λ|t i -t j |) is a time decay term. A longer time span results in a stronger decay, ultimately reducing the role of this semantic feature in similarity calculation. The overall summation represents the sum of the weighted temporal effects of multiple semantic dimensions to obtain the total similarity.

[0093] It can be understood that if two pieces of information are highly overlapped in key semantic dimensions and are close in time, then S d Approaches the maximum value. If the semantics are similar but the time span is too large, then S d It will be significantly depressed by the time decay term. If the semantic difference is large, the similarity will be low even if the time is close.

[0094] The present invention is applied to social event monitoring. If two messages both mention "epidemic outbreak in a certain place", they are semantically similar; if one is sent yesterday and the other is sent today (close in time), then S d High; if one is from half a year ago and the other is from today, even if the semantics are exactly the same, S d It will also decrease due to time decay to avoid being misjudged as a new event.

[0095] Step 203: Obtain an event association strength network based on the information source weights and the semantic similarity.

[0096] In some embodiments, step 203 may include:

[0097] Multiplying the credibility weight and the semantic similarity to obtain a fusion result;

[0098] Detect the degree of deviation between the co-occurrence information of each event and the benchmark information, and obtain the logarithmic adjustment term of the co-occurrence intensity;

[0099] The event association strength network is generated according to the fusion result and the logarithmic adjustment item of the co-occurrence strength.

[0100] Among them, the correlation strength between the i-th time and the j-th event in the event correlation strength network can be expressed as:

[0101]

[0102] Among them, R ij is the correlation strength between the i-th time and the j-th event in the event correlation strength network, W s is the information source weight, S d is the semantic similarity, I c is the co-occurrence information between two events, and I0 is the baseline information.

[0103] In practice, this formula is used to quantify the strength of associations between different events (or event fragments) in a multidimensional information flow, providing a numerical basis for constructing event networks and tracing information propagation paths. By integrating information source credibility, semantic similarity, and event co-occurrence, it derives an evaluation metric for determining the interconnectedness and strength of a set of events.

[0104] W s It is the weight of the information source, obtained by the aforementioned weight calculation formula, which reflects the credibility, dissemination frequency, influence, etc. of the information source. The higher the weight, the more reliable the information is, and the higher the credibility of the correlation between the information source and the events it participates in. This ensures that information from highly credible sources has more weight in participating in network construction and suppresses interference from noise sources.

[0105] S d It is semantic similarity, which is obtained by the combined effect of the semantic features of the information content and time decay. The higher the similarity of "content relevance" between information, the closer the semantics between events are, which highlights the true semantic relevance and improves the rationality of event connection.

[0106] I c It is the amount of co-occurrence information between two events, indicating the amount of information in which the two events are mentioned simultaneously or together within a certain time / space window. It can include co-occurrence features such as the same keyword, common location, name, topic, etc.

[0107] I0 is the reference information amount, which is used to c A normalization constant is used to prevent the denominator from being zero, and it also serves as a relative reference for the co-occurrence intensity.

[0108] is the logarithmic adjustment term of the co-occurrence intensity, if Ic >>I0, indicating that events frequently co-occur and have a strong relationship; if I c ≈I0 indicates that the co-occurrence frequency is normal. Using a logarithmic function helps to suppress the impact of extreme values and prevent high-frequency events from inflating the entire network. This improves the accuracy of structured event recognition and enhances the model's ability to distinguish between "weak correlation" and "strong coupling."

[0109] It can be understood that if two events have highly similar content, come from highly credible sources, and co-occur in multiple information, the correlation strength between the events is high. If a pair of events has different semantics, or the information source is unreliable, or they hardly co-occur, c ≈I0, then the correlation strength between events is low.

[0110] R ij As the edge weights are used to construct the event association graph, the thicker the edge, the stronger the connection between the two events. ij The event nodes can be clustered into "same event clusters" or "evolution chains", supporting the identification of complex structures such as rumor propagation paths and the diffusion of similar events.

[0111] In summary, this method integrates source quality, content semantics, and behavioral co-occurrence through multi-dimensional fusion. It also exhibits strong numerical stability, using a logarithmic function to smooth out co-occurrence deviations. It is also user-friendly for network visualization and can be directly used to construct weighted graph networks.

[0112] Step 204: Obtain a risk propagation index for evaluating the impact of information propagation paths on risk amplification based on the distribution characteristics of the depth of each network node in the event correlation strength network.

[0113] In some embodiments, step 204 may include:

[0114] Constructing a depth diffusion factor according to the diffusion coefficient and the depth of the network nodes;

[0115] quantifying the cumulative effect of the correlation strength in the propagation path according to the event correlation strength and the depth diffusion factor;

[0116] According to the time influencing factor and the propagation time of the event, a time growth function including both time and space factors is constructed;

[0117] The risk propagation index is determined according to the cumulative effect and the time growth function.

[0118] Among them, the risk transmission index can be expressed as:

[0119]

[0120] Among them, P is the risk transmission index, R ij is the event correlation strength, k is the diffusion coefficient, Dn is the network node depth, μ is the time impact factor, and t is the propagation time.

[0121] In practice, this formula serves as the core computational logic for assessing how potential risk events spread through information networks and serves as a precursor to early warning decisions. It comprehensively considers the strength of connections between events, the depth of propagation within the network structure, and the temporal evolution of events to quantify the propagation capacity and impact of a potential risk event within the network.

[0122] R ij is the event association strength, indicating the strong correlation between event i and event j. The larger the value, the more likely they are to influence each other or belong to the same event chain. It comes from the previous formula and has integrated the source weight, semantic similarity, and co-occurrence degree to determine whether to spread and the initial spread weight.

[0123] k is the diffusion coefficient, which controls the effect of node depth on the spread index. A larger value indicates a greater impact from "layer-by-layer spread" (e.g., the public opinion amplifier effect). This value can be optimized experimentally or set based on industry experience, amplifying the risk level of "deep spread."

[0124] D m This is the depth of network nodes, indicating the "depth" or "distance" of an event within the information dissemination network. It is typically measured in terms of the number of hops or node layers from the source event. A deeper network node represents a longer dissemination chain and a wider impact. This is also known as the "diffusion path length" indicator for the dissemination path.

[0125] is the depth diffusion factor, and the exponential function is used to simulate the accelerated growth of the breadth of transmission; each time you go deeper, the risk gain of the path increases exponentially. To quantify the cumulative effect of the correlation strength in the propagation path.

[0126] μ is the time impact factor, which controls the contribution of the event's "growth over time" to the propagation index; it is generally set to a positive value. The larger the value, the more likely the event is to amplify the risk impact in a short period of time.

[0127] t is the propagation time, which represents the time span since the event occurred (such as hours or days); it can be updated dynamically to reflect the event life cycle.

[0128] 1-e -μ·t It is a time growth function, which indicates the natural growth of risk influence over time. It grows slowly in the early stage and approaches 1 rapidly as time goes by (similar to the front part of the S-shaped curve). It ensures that the impact of early transmission is small, and as time goes by, the impact gradually increases.

[0129] The risk contribution of each pair of related events = event correlation × network transmission effect × time accumulation impact; the overall sum = the sum of the risk contributions of all potential transmission paths in the network; and finally a numerical value is obtained to comprehensively assess the degree of risk diffusion.

[0130] This invention is highly dynamic, incorporating propagation time and supporting real-time updates. It is sensitive to structure, exponentially amplifying deep propagation paths and more closely recapitulating actual propagation mechanisms. By integrating multiple factors, including content, structure, and time, risk assessments are more comprehensive.

[0131] It is understandable that if a piece of risk information comes from a highly credible information source, is semantically highly similar to multiple events, and is spread by a large number of nodes in the deep network, then its propagation index will rise rapidly; if it only appears in a few shallow nodes, or the time is short, the propagation impact will be relatively limited.

[0132] Step 205: Generate a comprehensive early warning index for social information network monitoring based on the risk propagation index, the information source weight, the semantic similarity, and the correlation strength between events in the event correlation strength network.

[0133] In some embodiments, step 205 may include:

[0134] Obtaining the fusion result obtained by multiplying the credibility weight and the semantic similarity;

[0135] Obtaining an adjustment coefficient for adjusting the early warning comprehensive index;

[0136] The risk propagation index, the fusion result and the event association strength are weighted and summed using the adjustment coefficient to obtain the early warning comprehensive index.

[0137] Among them, the early warning comprehensive index can be expressed as:

[0138] Z=θ1·P+θ2·∑(W s ·S d )+θ3·R ij ;

[0139] Among them, Z is the early warning comprehensive index, P is the risk transmission index, W s is the information source weight, S d is the semantic similarity, R ij is the event correlation strength, θ1, θ2, and θ3 are the first, second, and third adjustment coefficients, respectively.

[0140] In practice, this formula serves as the core decision-making basis for comprehensively determining the severity and warning level of risk events in current information networks. By integrating multidimensional risk factors (propagation trends, information credibility and semantic associations, and event structure coupling), it forms a quantitative and controllable comprehensive warning index, providing computational support for the system's output of high, medium, and low-level warnings.

[0141] P is the risk diffusion index, derived from the aforementioned formula, reflecting the breadth and intensity of an event's spread within a network. A larger value indicates a stronger structural spread and greater impact. It represents the spread trend and reflects whether the risk is spreading rapidly.

[0142] ∑(W s ·S d ) is the total value of weighted semantic credibility, which is the sum of multiple pieces of information in the current network; W s It is the credibility of the information source, reflecting whether the source is reliable; S d It represents semantic similarity, indicating the degree of similarity to the content of key events. The product of the two reflects the degree of "credibility and relevance" of a piece of information. It represents the relevance of the content, reflecting "whether it is truly related to the core of the risk."

[0143] R ij Consider the tightness of coupling between key events (or core event chains) individually. This is derived from structural co-occurrence and semantic consensus, and can be understood as "whether risk is spreading outward in a structured manner from certain nodes." This represents structural coupling and reflects "whether the event network has the ability to diffuse."

[0144] θ1, θ2, and θ3 are used to adjust the weights of three types of risk factors. They can be adjusted according to actual business scenarios: for example, θ1 can be strengthened in public opinion control, and θ2 can be increased in emergency perception. Flexible parameter adjustment is supported to adapt the model to different scenarios and strategic objectives.

[0145] It can be understood that the comprehensive index Z is to assess the risk level of the current situation from three dimensions: whether it is widely spread (P); whether it comes from credible content and is semantically consistent (W); s ·S d ); whether the structure is strongly coupled by multiple events (R ij ), this integration method avoids the misjudgment of "rapid spread but untrue" or "credible information but no spread".

[0146] This invention achieves multi-dimensional fusion, integrating communication trends, content similarity, and network structure. It is flexible and adjustable, allowing rapid adaptation to different scenarios by adjusting the theta coefficient. Adaptive evolution allows for real-time data flow calculations and dynamic early warning responses.

[0147] Step 206: Determine the warning level result corresponding to the warning comprehensive index according to the preset threshold range.

[0148] In some embodiments, step 206 may include:

[0149] When the warning comprehensive index is greater than or equal to the first threshold, the corresponding warning level result is a high-level warning;

[0150] When the warning comprehensive index is greater than or equal to the second threshold and less than the first threshold, the corresponding warning level result is a medium warning;

[0151] When the early warning comprehensive index is less than the third threshold, the corresponding early warning level result is low risk and no early warning is given;

[0152] The first threshold is greater than the second threshold, and the second threshold is greater than the third threshold.

[0153] When the comprehensive early warning index is greater than or equal to the first threshold, it indicates that the current risk index is high and the incident has obvious spread and severity. The system will trigger a high-level early warning so that relevant departments can take strong response measures in a timely manner.

[0154] When the comprehensive warning index is greater than or equal to the second threshold and less than the first threshold, the risk index is at a medium level. While there may be some risk of the incident spreading, it has not yet reached the highest alert level. The system will trigger a medium-level warning, monitor the situation, and take appropriate measures.

[0155] When the warning comprehensive index is less than the third threshold, the current risk index is low, and the event's correlation and spread are insufficient to pose a significant threat. The system then determines the risk as low and does not trigger a warning, but continues monitoring.

[0156] By setting graded thresholds in the above manner, different risk levels can be effectively distinguished, missed reports and false alarms can be avoided, and the scientific nature and sensitivity of the early warning mechanism can be ensured.

[0157] See also Figure 2 , Figure 2 This is a structural diagram of a social information networking monitoring and early warning system provided by the present invention.

[0158] like Figure 2 As shown, a social information network monitoring and early warning system proposed in an embodiment of the present invention includes:

[0159] A weight determination module 301 is used to collect and analyze multi-source information of information sources and determine the information source weights of the information sources;

[0160] Semantic processing module 302, configured to obtain semantic similarity of cross-temporal semantics based on feature vectors and feature weights extracted from the multi-source information by a distributed semantic model;

[0161] An association strength module 303 is configured to obtain an event association strength network based on the information source weights and the semantic similarity;

[0162] The risk index module 304 is configured to obtain a risk propagation index for evaluating the impact of information propagation paths on risk amplification based on the distribution characteristics of the depth of each network node in the event correlation strength network;

[0163] An early warning index module 305 is configured to generate an early warning comprehensive index for social information network monitoring based on the risk propagation index, the information source weight, the semantic similarity, and the correlation strength between events in the event correlation strength network;

[0164] The monitoring and warning module 306 is used to determine the warning level result corresponding to the warning comprehensive index according to a preset threshold range.

[0165] See also Figure 3 , Figure 3 Schematic diagram of an embodiment of an electronic device provided by an embodiment of the present invention. Figure 3 As shown, an embodiment of the present invention provides an electronic device 400, including a memory 410, a processor 420, and a computer program 411 stored in the memory 410 and executable on the processor 420. When the processor 420 executes the computer program 411, the following steps are implemented:

[0166] Collecting and analyzing multi-source information of information sources to determine information source weights of the information sources;

[0167] Obtaining semantic similarity of cross-temporal semantics based on feature vectors and feature weights extracted from the multi-source information by a distributed semantic model;

[0168] Obtaining an event association strength network according to the information source weights and the semantic similarity;

[0169] According to the distribution characteristics of the depth of each network node in the event correlation strength network, a risk propagation index for evaluating the impact of information propagation paths on risk amplification is obtained;

[0170] Generate a comprehensive early warning index for social information network monitoring based on the risk propagation index, the information source weight, the semantic similarity, and the correlation strength between events in the event correlation strength network;

[0171] According to the preset threshold range, the warning level result corresponding to the warning comprehensive index is determined.

[0172] See also Figure 4 , Figure 4 Schematic diagram of an embodiment of a computer-readable storage medium provided by an embodiment of the present invention. Figure 4As shown, this embodiment provides a computer-readable storage medium 500 on which a computer program 411 is stored. When the computer program 411 is executed by a processor, the following steps are implemented:

[0173] Collecting and analyzing multi-source information of information sources to determine information source weights of the information sources;

[0174] Obtaining semantic similarity of cross-temporal semantics based on feature vectors and feature weights extracted from the multi-source information by a distributed semantic model;

[0175] Obtaining an event association strength network according to the information source weights and the semantic similarity;

[0176] According to the distribution characteristics of the depth of each network node in the event correlation strength network, a risk propagation index for evaluating the impact of information propagation paths on risk amplification is obtained;

[0177] Generate a comprehensive early warning index for social information network monitoring based on the risk propagation index, the information source weight, the semantic similarity, and the correlation strength between events in the event correlation strength network;

[0178] According to the preset threshold range, the warning level result corresponding to the warning comprehensive index is determined.

[0179] It should be noted that, in the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0180] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0181] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A system that specifies the functions of a box or boxes.

[0182] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction system that is implemented in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0183] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0184] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0185] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A social information network monitoring and early warning method, characterized in that: The method comprises: Collecting and analyzing multi-source information of information sources to determine information source weights of the information sources; Obtaining semantic similarity of cross-temporal semantics based on feature vectors and feature weights extracted from the multi-source information by a distributed semantic model; Obtaining an event association strength network according to the information source weights and the semantic similarity; According to the distribution characteristics of the depth of each network node in the event correlation strength network, a risk propagation index for evaluating the impact of information propagation paths on risk amplification is obtained; Generate a comprehensive early warning index for social information network monitoring based on the risk propagation index, the information source weight, the semantic similarity, and the correlation strength between events in the event correlation strength network; According to the preset threshold range, the warning level result corresponding to the warning comprehensive index is determined.

2. The social information network monitoring and early warning method according to claim 1, characterized in that: The collecting and analyzing multi-source information of information sources to determine the information source weights of the information sources includes: Collect the historical credibility, information dissemination frequency and influence index of the information source; Normalization is performed based on the historical credibility of the information source, the frequency of information dissemination, and the basic indicators corresponding to the influence index to obtain the normalized results of the above three indicators; The normalized processing results of the three indicators are combined with the weighting coefficients to perform linear combination to obtain the information source weight of the information source.

3. The social information network monitoring and early warning method according to claim 2, characterized in that: The information source weight is expressed as: Among them, W s is the information source weight, T h is the historical credibility of the information source, T max is the maximum value of historical credibility, F c is the frequency of information dissemination, F max is the maximum transmission frequency among all information sources, R i is the influence index of the information source, R max is the maximum value of the information influence index, α, β, and γ are the first, second, and third weighting coefficients, respectively.

4. The social information network monitoring and early warning method according to claim 3 is characterized in that: The feature vectors and feature weights extracted from the multi-source information according to the distributed semantic model to obtain the semantic similarity of cross-temporal semantics include: Perform weighted summation on the feature vectors to form the basic similarity; Obtaining the time span of the multi-source information and a time decay coefficient representing the degree of influence of the time span on the semantic similarity; The basic similarity is dynamically downgraded according to the time span and the time decay coefficient to obtain the semantic similarity of the cross-time semantics.

5. The social information network monitoring and early warning method according to claim 4, characterized in that: The semantic similarity is expressed as: S d =∑(V i ·W i )·exp(-λ|t i -t j |); Among them, S d is the semantic similarity, V i is the semantic feature vector of the i-th feature, W i is the feature weight of the i-th feature, t i and t j is the timestamp of the semantic information, and λ is the time decay coefficient.

6. The social information network monitoring and early warning method according to claim 5, characterized in that: Obtaining an event association strength network according to the information source weights and the semantic similarity includes: Multiplying the credibility weight and the semantic similarity to obtain a fusion result; Detect the degree of deviation between the co-occurrence information of each event and the benchmark information, and obtain the logarithmic adjustment term of the co-occurrence intensity; The event association strength network is generated according to the fusion result and the logarithmic adjustment item of the co-occurrence strength.

7. The social information network monitoring and early warning method according to claim 6, characterized in that: The risk propagation index for evaluating the impact of information propagation paths on risk amplification is obtained based on the distribution characteristics of the depth of each network node in the event correlation strength network, including: Constructing a depth diffusion factor according to the diffusion coefficient and the depth of the network nodes; quantifying the cumulative effect of the correlation strength in the propagation path according to the event correlation strength and the depth diffusion factor; According to the time influencing factor and the propagation time of the event, a time growth function including both time and space factors is constructed; The risk propagation index is determined according to the cumulative effect and the time growth function.

8. The social information network monitoring and early warning method according to claim 7, characterized in that: Generating a comprehensive early warning index for social information network monitoring based on the risk propagation index, the information source weight, the semantic similarity, and the correlation strength between events in the event correlation strength network includes: Obtaining the fusion result obtained by multiplying the credibility weight and the semantic similarity; Obtaining an adjustment coefficient for adjusting the early warning comprehensive index; The risk propagation index, the fusion result and the event association strength are weighted and summed using the adjustment coefficient to obtain the early warning comprehensive index.

9. The social information network monitoring and early warning method according to claim 8, characterized in that: Determining the warning level result corresponding to the warning comprehensive index according to the preset threshold range includes: When the comprehensive warning index is greater than or equal to the first threshold, the corresponding warning level result is a high-level warning; When the comprehensive warning index is greater than or equal to the second threshold and less than the first threshold, the corresponding warning level result is a medium warning; When the comprehensive warning index is less than the third threshold, the corresponding warning level result is low risk and no warning is issued temporarily; The first threshold is greater than the second threshold, and the second threshold is greater than the third threshold.

10. A social information network monitoring and early warning system, characterized in that: The system comprises: A weight determination module, configured to collect and analyze multi-source information of information sources and determine information source weights of the information sources; A semantic processing module, configured to obtain semantic similarity of cross-temporal semantics based on feature vectors and feature weights extracted from the multi-source information by a distributed semantic model; An association strength module, configured to obtain an event association strength network based on the information source weights and the semantic similarity; A risk index module is used to obtain a risk propagation index for evaluating the impact of information propagation paths on risk amplification based on the distribution characteristics of the depth of each network node in the event correlation intensity network; An early warning index module, configured to generate an early warning comprehensive index for social information network monitoring based on the risk propagation index, the information source weight, the semantic similarity, and the correlation strength between events in the event correlation strength network; The monitoring and warning module is used to determine the warning level result corresponding to the warning comprehensive index according to a preset threshold range.

Images