Potential risk hidden danger identification method and device based on multi-dimensional information and event causal sorting, equipment, medium and product

Abstract

The invention discloses a potential risk hidden danger identification method and device based on multi-dimensional information and event causal sorting, equipment, a medium and a product, and relates to the field of artificial intelligence. The method comprises the following steps: acquiring multi-modal data from a subscribed public opinion monitoring object, performing feature extraction, and extracting multi-modal information features; performing data structured analysis based on the multi-modal information features, and extracting key attribute structured data of each event; performing clustering analysis on all the events, and dividing all the events into a series of large event sets; performing information deduplication and integration analysis on each small event in the large event set to obtain key attribute structured data of each large event set, performing causal chain sorting on an occurrence sequence and a causal relationship of a series of large event sets to obtain an event causal chain, and performing potential risk hidden danger identification on the event to obtain a hidden danger identification result. The comprehensiveness and timeliness of public opinion monitoring and the accuracy of public opinion analysis can be improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method, device, equipment, medium and product for identifying potential risks and hidden dangers based on multi-dimensional information and event causal analysis. Background Art

[0002] Internet public opinion refers to the popular opinions on social events on the Internet, which is a manifestation of social public opinion. In recent years, the impact of Internet public opinion on life order and social stability has been increasing day by day. Some major Internet public opinion events have made people begin to realize the great role of the Internet in social supervision. At the same time, with the development of the big data era, the requirements for public opinion monitoring are getting higher and higher, and the monitoring of the status of Internet public opinion has become particularly important.

[0003] Identification of potential risks and hidden dangers in public opinion monitoring refers to the integration of Internet information collection technology and intelligent information processing technology, which realizes users' information needs such as online public opinion monitoring and hot event tracking through automatic capture, automatic classification and clustering, topic detection, and special focus on massive Internet information, and forms analysis results such as briefings, reports, and charts to provide an analytical basis for identifying potential risks and hidden dangers of online public opinion events and making correct public opinion guidance.

[0004] At present, the method of identifying potential risks and hidden dangers in public opinion monitoring is mostly to screen public opinion through public opinion analysts, and then track public opinion, so as to have a rough judgment on the public opinion status of the event in terms of its influence. However, this method has the problem that the monitoring data source is not comprehensive enough, and it is impossible to achieve real-time analysis and monitoring, and it is impossible to intuitively understand the public opinion status and risk level of online public opinion, which leads to the inability to obtain effective information in a timely and accurate manner to assist business decision-making. Summary of the invention

[0005] The purpose of this application is to provide a method, device, equipment, medium and product for identifying potential risks and hidden dangers based on multi-dimensional information and event causal analysis, so as to improve the comprehensiveness and timeliness of public opinion monitoring and the accuracy of public opinion analysis.

[0006] To achieve the above objectives, this application provides the following solutions.

[0007] In the first aspect, the present application provides a method for identifying potential risks and hidden dangers based on multi-dimensional information and event causal analysis, including:

[0008] Obtain multimodal data from subscribed public opinion monitoring objects, including audio data, video data, and text data;

[0009] Perform feature extraction on the multimodal data to extract multimodal information features including timbre features, face features, and text information features;

[0010] Performing data structured analysis based on multimodal information features to extract key attribute structured data of each event; the key attribute structured data of each event includes time, location, people involved, specific event information, and risk index;

[0011] Perform cluster analysis on all events based on the key attribute structured data of each event, and divide all events into a series of large event sets;

[0012] Perform information deduplication and integration analysis on each small event in the large event set to obtain key attribute structured data of each large event set;

[0013] Based on the key attribute structured data of each major event set, the causal chain of the occurrence sequence and causal relationship of a series of major event sets is sorted out to obtain the causal chain of public opinion monitoring events;

[0014] Identify potential risks of public opinion monitoring events based on the causal chain of the events.

[0015] Optionally, the feature extraction of the multimodal data to extract multimodal information features including timbre features, facial features and text information features specifically includes:

[0016] Use the audio processing library Librosa to extract the timbre features of audio data and convert the audio data into text data;

[0017] Use deep convolutional neural networks to identify facial features in video data and use the ArcFace loss function to enhance its recognition ability;

[0018] With the help of large language models, text information features are extracted from text data, including time, place, people involved, and specific information of events.

[0019] Optionally, the data structured analysis is performed based on the multimodal information features to extract key attribute structured data of each event, specifically including:

[0020] Extract the time, location and specific event information corresponding to each event based on text information features;

[0021] The characters involved in each event are extracted by integrating the timbre features, face features and text information features;

[0022] The risk index is calculated based on the time, location, event details and people involved in each event.

[0023] Optionally, the calculating of the risk index based on the time, location, event specific information and persons involved corresponding to each event specifically includes:

[0024] Calculate the number of hours t from the current time to the time corresponding to each event, and use the formula Calculate the time risk index R_t;

[0025] Calculate the straight-line distance s between the location of each event corresponding point and the public opinion monitoring center of the region, and use the formula Calculate the spatial risk index R_s;

[0026] Based on the specific information of each event and the people involved, according to the formula Calculate the public opinion risk index R_p; where p is the total number of articles involved in the event, including the number of views, comments, reposts, and likes; when the people involved include public figures or the specific information of the event involves social hot spots, it is recorded as flag = true, otherwise it is recorded as flag = false;

[0027] The risk index R is calculated according to the formula R = εt×R_t+εs×R_s+εp×R_p; where εt, εs and εp are the weights of the time risk index, space risk index and public opinion risk index respectively.

[0028] Optionally, the cluster analysis is performed on all events based on the key attribute structured data of each event, and all events are divided into a series of large event sets, specifically including:

[0029] Construct the feature vector corresponding to each event based on the key attribute structured data of each event;

[0030] Calculate the Euclidean distance between two events based on the feature vector corresponding to each event;

[0031] The similarity between two events is calculated based on the Euclidean distance between the two events;

[0032] Construct a similarity matrix based on the similarity between two events;

[0033] Based on the similarity matrix, a graph adjacency matrix is ​​constructed, and the spectral clustering algorithm is used for clustering, dividing all events into a series of large event sets; each large event set includes multiple small events.

[0034] Optionally, the information deduplication and integration analysis is performed on each small event in the large event set to obtain key attribute structured data of each large event set, specifically including:

[0035] The time of each major event set is the earliest time of occurrence of all the small events in the major event set; the location of each major event set includes the locations of all the small events in the major event set; the people involved in each major event set include the people involved in all the small events in the major event set; the event specific information of each major event set is the integration of the event specific information of all the small events in the major event set; the risk index of each major event set is the maximum value of the total risk coefficient of all the small events in the major event set.

[0036] In the second aspect, the present application provides a potential risk identification device based on multi-dimensional information and event causal analysis, including:

[0037] The multimodal data acquisition module is used to obtain multimodal data from the subscribed public opinion monitoring objects, including audio data, video data and text data;

[0038] A data feature extraction module is used to extract features from multimodal data, and extract multimodal information features including timbre features, facial features, and text information features;

[0039] A data structured analysis module is used to perform data structured analysis based on multimodal information features to extract key attribute structured data of each event; the key attribute structured data of each event includes time, location, people involved, specific event information and risk index;

[0040] The event clustering analysis module is used to perform clustering analysis on all events based on the key attribute structured data of each event, and divide all events into a series of large event sets;

[0041] The information deduplication and integration module is used to perform information deduplication and integration analysis on each small event in the large event set to obtain key attribute structured data of each large event set;

[0042] The event causal chain combing module is used to comb the occurrence sequence and causal relationship of a series of major event sets based on the key attribute structured data of each major event set, and obtain the causal chain of public opinion monitoring events;

[0043] The potential risk identification module is used to identify potential risks and hidden dangers of events based on the causal chain of public opinion monitoring events.

[0044] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for identifying potential risks and hidden dangers based on multi-dimensional information and event causal combing.

[0045] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for identifying potential risks and hidden dangers based on multi-dimensional information and event causal analysis.

[0046] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the method for identifying potential risks and hidden dangers based on multi-dimensional information and event causal analysis.

[0047] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0048] The present application provides a method, device, equipment, medium and product for identifying potential risks and hidden dangers based on multi-dimensional information and event causal combing, which can comprehensively integrate multimodal data such as text, audio and video involved in public opinion events, and significantly improve the accuracy and robustness of public opinion information processing; through structured analysis of multimodal information features such as timbre features, facial features and text information features, the key attribute structured data of each event (including time, place, people involved, specific event information and risk index) is extracted, and the clustering analysis method can be dynamically adjusted to achieve a more accurate clustering effect, thereby improving the accuracy of event analysis, and helping to quickly identify and respond to key nodes of public opinion events; through combing and understanding the causal development context of events, the development trend of events and the judgment of potential risks and hidden dangers are more effectively analyzed, the depth and breadth of complex event analysis are improved, and complex public opinion events can be better predicted and handled, providing a strong basis for quickly responding to various emergencies and social events. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0050] Figure 1 This is a flowchart of a method for identifying potential risks and hidden dangers based on multi-dimensional information and event causal analysis for this application. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0052] In response to the defects in the prior art, this application proposes a method, device, equipment, medium and product for identifying potential risks and hidden dangers based on multi-dimensional information and event causal analysis, so as to improve the comprehensiveness and timeliness of public opinion monitoring and the accuracy of public opinion analysis.

[0053] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0054] In an exemplary embodiment, Figure 1 As shown, a method for identifying potential risk hazards based on multi-dimensional information and event causal analysis is provided, including the following steps 1 to 7.

[0055] Step 1: Obtain multimodal data from the subscribed public opinion monitoring object, including audio data, video data, and text data.

[0056] The subscribed public opinion monitoring object can be a specific enterprise, group, individual, etc. It should be noted that the acquisition of multimodal data in this application is carried out under the premise of complying with the corresponding data protection laws and policies of the country where it is located and obtaining authorization from the corresponding device owner.

[0057] By regularly crawling multimodal data such as audio, video, and text from different channels / platforms, we can obtain the latest multidimensional information and ensure the comprehensiveness and timeliness of the data. Furthermore, we can identify the format of the crawled data (multidimensional information) and identify the data type by the extension of the crawled data. For example, we can classify MP4, AVI, and MOV files into the video data category, MP3 and WAV files into the audio data category, and HTML, TXT, PDF, and other types of files or web page texts into the text data category.

[0058] Step 2: Perform feature extraction on the multimodal data to extract multimodal information features including timbre features, facial features, and text information features.

[0059] Extract features from audio, video, text and other data and integrate feature information. For example, for audio data, use the audio processing library Librosa to extract timbre features such as MFCCs, and then perform voiceprint comparison and voiceprint recognition. For video data, use deep convolutional neural networks (DCNNs) to learn deep features for face recognition, extract facial features, and use the ArcFace loss function to enhance its recognition ability. For text data, use the large language model (LLM) to analyze and classify important text information features such as time, place, people involved, and specific information of events in the text. Then integrate the extracted feature information into local files (such as csv files, etc.).

[0060] Step 3: Perform data structured analysis based on multimodal information features to extract key attribute structured data of each event; the key attribute structured data of each event includes time, location, people involved, specific event information and risk index.

[0061] In the context of social governance and risk management, the analysis and processing of each event needs to be based on structured data to ensure the accuracy and operability of the information. To this end, event data needs to include the following five key attributes: time, location, people involved, specific event information, and risk index, and each event should be arranged in order according to the risk index and tabulated for further analysis.

[0062] The key attribute structured data definitions and uses of each event are as follows. (a) Time: recorded in a standard time format (such as ISO 8601ss), used to assess time risks and analyze the temporal relationship of related events. (b) Location: the geographical location where the event occurred, used for spatial risk assessment, measuring the geographical impact of the event and its proximity to the target area of ​​public opinion monitoring. (c) People involved: key people or entities involved in the event, used to determine the public opinion influence and social attention of the event. (d) Specific information about the event: a detailed description of the event, as the basis for qualitative analysis, to help identify the nature and potential consequences of the event. (e) Risk index: a comprehensive risk score obtained after quantitative analysis of specific information, to help quickly identify and handle high-risk events.

[0063] Specifically, the publishing time of the text can be directly obtained from the text information features as the event time, and the sentences describing the geographical location can be found in the text as the event location. The parties involved in the event can be comprehensively judged as the persons involved based on the facial features extracted from the video data, the timbre features in the audio data, and the names of the people appearing in the text data. The audio data is also integrated into text data, combined with the directly obtained text data, and the extracted text gist is summarized by the large language model as the specific information of the event. Furthermore, based on the key attribute information such as time, place, person, and specific information of the event involved in the integration, the risk index of the event can be calculated. The steps for calculating the risk index are as follows.

[0064] (3.1) Calculate the time risk index R_t

[0065] The time corresponding to each event (also called event time) is recorded as t hours from the current time, and the time risk index of the event is R_t. The time development of public opinion is divided into the incubation period, outbreak period, continuous period and decline period, and the corresponding values ​​of R_t in each stage are given:

[0066] In the latent period (0≤t<10), the event has just occurred and the risk index is low, but there is a risk of rapid spread. The risk index of this stage is recorded as R_t=0.5.

[0067] During the outbreak period (10≤t<60), the public opinion event has been fermenting for a period of time, and the direction is not yet clear. At this time, the risk index is extremely high, which may trigger negative public opinion and chain reactions. The time risk index of this stage is recorded as R_t=1.

[0068] During the duration period (60≤t<120), there are risks in public opinion events, and the risk growth trend slows down or tends to stabilize. At this time, the risk index is high, but relatively stable. The risk index of this stage is recorded as R_t=0.75.

[0069] In the fallback period (t≥120), the heat of the public opinion event begins to decline. At this time, the risk index gradually decreases, but we still need to pay attention to the subsequent impact. The risk index of this stage is recorded as R_t=0.25.

[0070] In summary, the calculation formula of the time risk index R_t can be expressed as follows:

[0071]

[0072] (3.2) Calculation of spatial risk index R_s

[0073] Here, we take the local public opinion response as an example, and record the straight-line distance between the location of each event corresponding point (referred to as the event location) and the public opinion monitoring center of the region as s kilometers. The occurrence range of social events is divided into high-risk areas and low-risk areas, and the corresponding values ​​of R_s in each area are given. If an event has no exact location, its spatial risk index is not considered, and the spatial risk index weight of such events is εs = 0.

[0074] Among them, in the high-risk area (0≤s<5), the event location is close to the regional public opinion monitoring center, the risk is greater and the change with distance is small. A quadratic function that is symmetric about s=0 and opens downward is used to describe the attenuation of the spatial risk index in this area, that is,

[0075] In the low-risk area (s≥5), the event location is far away from the regional public opinion monitoring center, the risk is small, and the initial decay is faster with the increase of distance, and the decay is slower in the later stage. The power function is used to describe the decay of the spatial risk index in this area, that is,

[0076] In summary, the calculation formula of the spatial risk index R_s can be expressed as:

[0077]

[0078] (3.3) Calculate the public opinion risk index R_p

[0079] The sum of the number of events involved (including readings, comments, reposts, likes, etc.) is recorded as p. Based on the facial features and event-specific information in the multimodal information features, the face database is compared. If a public figure is identified or the specific information of the event is analyzed using LLM to involve social hot spots, then flag = true, otherwise flag = false. Flag is a label given for convenience. The numerical relationship between the people involved in the event and the public opinion risk index is described by an exponential function, that is:

[0080]

[0081] (3.4) Calculation of total risk index

[0082] Multiplying R_t, R_s, and R_p of an event by their corresponding weights and then adding them together is the total risk index of the event, recorded as the risk index R of the event:

[0083] R=εt×R_t+εs×R_s+εp×R_p (4)

[0084] Among them, εt, εs and εp are the weights of the time risk index, space risk index and public opinion risk index respectively.

[0085] After the key attribute structured data of each event (referred to as event data) is collected, the event data is sorted and stored in a structured manner for subsequent analysis. (a) Data formatting: All event data is entered into a structured database according to the defined format and stored using a relational database (such as MySQL) or a NoSQL database (such as MongoDB). (b) Data verification and cleaning: The accuracy and consistency of the data are ensured through a combination of automated tools such as OpenRefine and Weka and manual inspection.

[0086] In order to further optimize resource allocation and emergency response strategies, events need to be sorted according to their risk index, as follows: (a) Algorithmic sorting: Sort all events from high to low according to their risk index. Use efficient sorting algorithms such as quick sort to ensure fast processing on large-scale data sets. (b) Table generation: The sorted event data will be presented in a table format, containing five fields: time, location, people involved, event details, and risk index. The table can be exported to a common analysis format (CSV file) for further analysis and modeling.

[0087] Step 4: Perform cluster analysis on all events based on the key attribute structured data of each event, and divide all events into a series of large event sets.

[0088] The cluster analysis of all event data specifically includes the following steps:

[0089] (4.1) Construct the feature vector corresponding to each event based on the key attribute structured data of each event.

[0090] Construct the feature vector xi=[xi1,xi2,xi3,xi4,xi5] corresponding to each event data. The 5 elements contained in the feature vector come from the key attribute structured data in step 3, which are time, location, people involved, event specific information and risk index. The obtained feature vector is used to calculate the Euclidean distance and similarity between events, which is the basis for the subsequent construction of similarity matrix and event clustering.

[0091] (4.2) The Euclidean distance between two events is calculated based on the feature vector corresponding to each event.

[0092] For all events, the Euclidean distance is used to compare them pairwise. The similarity of each pair of events i and j can be calculated, and a similarity matrix can be established based on the similarity. The calculation formula for the Euclidean distance between two events is as follows:

[0093]

[0094] where xi k and xj k(k=1,2,3,4,5) represents the kth element of event i and j in feature vectors xi=[xi1,xi2,xi3,xi4,xi5] and xj=[xj1,xj1,xj3,xj4,xj5] respectively. ij represents the Euclidean distance between events i and j.

[0095] (4.3) The similarity between two events is calculated based on the Euclidean distance between two events. The formula is as follows:

[0096]

[0097] Where S ij Represents the similarity between events i and j.

[0098] (4.4) Construct a similarity matrix based on the similarity between two events.

[0099] Construct an N×N similarity matrix, where N is the number of events and each element S in the matrix ij Represents the similarity between event i and event j. The similarity matrix is ​​filled according to (4.3), and the similarity matrix is ​​a symmetric matrix, that is, S ij =S ji , so only half of the matrix (such as the upper triangle or the lower triangle) needs to be calculated. The final similarity matrix is ​​shown in the following matrix type:

[0100]

[0101] Where S ij is the similarity between the corresponding event i and event j. All elements on the diagonal are 1, indicating that the similarity between the event and itself is 1.

[0102] (4.5) Based on the similarity matrix, an adjacency matrix of a graph (which can be an undirected graph, a directed graph, or a weighted graph, etc.) is constructed, and the spectral clustering algorithm is used for clustering, thereby dividing all events into a series of large event sets M1, M2, ..., M n ; n is the number of large event sets. The multiple events included in each large event set are called small events m i , i=1,2,...p, p is the number of small events included in the big event set. All events in each big event set are related to this big event and are part of this big event. Then, through the sequence and causal relationship of each small event, the causal chain analysis can be performed to analyze the context and risk index of the big event.

[0103] Step 5: Perform information deduplication and integration analysis on each small event in the large event set to obtain key attribute structured data of each large event set.

[0104] Record each major event set M j ={m1,m2,...,m p}, j = 1, 2, ... n, for m i Perform deduplication and information merging to obtain the integrated event node M′ j . M′ j is the integrated set of large events, and each integrated small event m′ i All are M' j A thread of all m' i The information of j Events. Specifically, each large event set M′ j The time is the big event set M j All small events in m i The earliest time of occurrence, that is, the minimum time; each major event set M′ j Locations include M j The locations of all small events in each large event set M′ j The characters involved include M j All small events in m i The people involved; each major event set M′ j The event details are M j All small events in m i The integration of event specific information; each large event set M′ j The risk index is M j All minor events i The maximum value of the total risk factor.

[0105] Step 6: Based on the key attribute structured data of each major event set, the occurrence sequence and causal relationship of a series of major event sets are sorted out to obtain the causal chain of public opinion monitoring events.

[0106] The event node M′ is sorted in chronological order j Perform selection sorting to obtain a linear event causal chain, that is, the public opinion monitoring event causal chain. The risk index of the event causal chain is recorded as the maximum value of the risk index of the event nodes, and then the causal chain information is stored in the database for further analysis.

[0107] Step 7: Identify potential risks and hidden dangers of the incident based on the causal chain of public opinion monitoring events.

[0108] The causal chain of public opinion monitoring events arranges event nodes in chronological order, which can help analyze the causes, development context and key nodes of public opinion events, evaluate risk index, and is used in event tracing, risk assessment, decision support and early warning mechanisms. It can effectively identify high-risk events and predict their development direction. When dealing with complex public opinions, the causal chain provides a basis for response strategies. By taking measures against high-risk nodes and updating feedback results in real time, response measures can be optimized. Cross-analysis of multiple causal chains can also reveal the relationship between events and prevent similar events from recurring.

[0109] This application can help relevant institutions grasp the dynamics of public opinion in a timely manner and provide strong support for decision-making by integrating and analyzing public opinion information from different social platforms. This application method has important application value in many fields such as government public relations, corporate brand management, and emergency response, and can significantly improve the level of social management and public security protection capabilities.

[0110] For example, in a specific embodiment, a series of discussions about "a certain food safety issue" were found on a social media in the city using the crawler program using the method of the present application. After crawling these data, a large number of negative comments were found through analysis. These comments are associated with the original post, which describes in detail the location and time of the food safety issue. The event chain of the food safety issue is generated by the method of the present application, and further integrated analysis reveals potential public health risks. Related to the above-mentioned food safety issue, a large number of negative comments about the inadequate government regulatory measures subsequently appeared. These comments were also analyzed by the method of the present application, associated with relevant discussion posts, and the time, location and other information of these posts were identified, and connected into a new event chain of regulatory measures. A comprehensive analysis of these two event chains found that food safety issues not only pose a threat to public health, but may also cause a negative impact on government trust. Based on this, a detailed risk assessment report was generated based on the risk index of the event causal chain, and specific suggestions such as strengthening food safety inspections, improving regulatory mechanisms, and strengthening publicity and education were proposed, and presented to relevant departments through data visualization tools to help them take timely action. After new posts and comments appear, the event chain will be automatically updated, and the risk assessment and strategy recommendations will be adjusted according to the latest data. For example, after a citizen's complaint was properly handled, a series of positive comments appeared. This application method will incorporate this new information into the analysis, adjust the risk assessment of the event chain, and provide the latest strategic recommendations. This application method ensures the timely discovery and effective response to social risks through effective event chain connection and real-time data update, and maintains social stability and public safety.

[0111] In an exemplary embodiment, the present application also proposes a potential risk and hidden danger identification device based on multi-dimensional information and event causal analysis, including the following functional modules. A multimodal data acquisition module is used to obtain multimodal data from subscribed public opinion monitoring objects, including audio data, video data and text data; a data feature extraction module is used to extract features from multimodal data and extract multimodal information features including timbre features, facial features and text information features; a data structured analysis module is used to perform data structured analysis based on multimodal information features and extract key attribute structured data of each event; the key attribute structured data of each event includes time, location, involved persons, specific event information and risk index; an event clustering analysis module is used to perform clustering analysis on all events based on the key attribute structured data of each event and divide all events into a series of large event sets; an information deduplication and integration module is used to perform information deduplication and integration analysis on each small event in the large event set to obtain the key attribute structured data of each large event set; an event causal chain combing module is used to perform causal chain combing on the occurrence order and causal relationship of a series of large event sets based on the key attribute structured data of each large event set to obtain the public opinion monitoring event causal chain. The potential risk identification module is used to identify potential risks and hidden dangers of events based on the causal chain of public opinion monitoring events.

[0112] The potential risk and hidden danger identification device based on multi-dimensional information and event causal combing proposed in the present application can crawl multi-modal data such as audio, video, text, etc. from different channels, regularly obtain the latest information, and ensure the comprehensiveness and timeliness of the data; perform structured analysis on the crawled multi-modal data and separate different types of data; perform feature extraction and integrate feature information on audio, video, text and other data respectively; extract key information such as time, place, and people involved in the integrated information through a large language model, and use a deep learning algorithm to infer the main context of the events involved in the information and calculate the risk index corresponding to the event, and store all events and their key attribute structured data in all event sets M; perform similarity and correlation analysis on the event set M through a similarity calculation algorithm, construct a similarity matrix, cluster all events in the event set M, and classify similar events into one category, thereby dividing M into M1, M2...M n A series of major events; M1, M2...M n For the collection of major events, the multimodal causal chains of the same event are sorted out, and then the public opinion risk levels are ranked. According to the order of risk levels, response strategies and action plans are generated for the corresponding events. High-risk events are given priority, and the results of the processing are evaluated and fed back. Unprocessed events are stored in the database for future reference.

[0113] The characteristics of the potential risk identification method and device based on multi-dimensional information and event causal combing proposed in this application are: (1) This application can efficiently integrate multi-modal data such as text, audio and video involved in social events, significantly improving the accuracy and robustness of public opinion information processing; even in the face of incomplete or noisy data, it can ensure the reliability of the analysis results, which is crucial for public opinion monitoring and risk assessment in social governance; (2) By analyzing the characteristics of multi-modal public opinion information, it can dynamically adjust the similarity calculation method to achieve a more accurate clustering effect, thereby improving the accuracy of event analysis, and helping to quickly identify and respond to key nodes of social events; (3) By combing and understanding the causal development context of events, more effective analysis of events can be achieved. The application can analyze the development trends of public opinion events and potential risks, improve the depth and breadth of complex event analysis, better predict and handle complex public opinion events, and provide strong support for social governance; (4) The technology of this application is highly portable and can quickly adapt to different application scenarios and needs, showing cross-domain application potential and being able to respond quickly to various emergencies and social events; (5) This application uses in-depth analysis of multimodal data, integrating voice, face and text information to construct a comprehensive event context map, achieve efficient discovery of potential risks and intelligent generation of response strategies, and has broad application value, especially in the fields of government public relations, corporate brand management and public safety management, and can significantly improve the level of social governance and public safety protection capabilities.

[0114] In an exemplary embodiment, the present application also provides a computer device, which may be a server or a terminal. The computer device includes a processor, a memory, an input / output interface, and a communication interface. The processor, the memory, and the input / output interface are connected via a system bus, and the communication interface is connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, the potential risk and hidden danger identification method based on multi-dimensional information and event causal combing is implemented.

[0115] In an exemplary embodiment, the present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for identifying potential risks and hidden dangers based on multi-dimensional information and event causal analysis.

[0116] In an exemplary embodiment, the present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the method for identifying potential risks and hidden dangers based on multi-dimensional information and event causal combing.

[0117] It can be understood by a person of ordinary skill in the art that all or part of the processes in the above-mentioned embodiment method can be completed by hardware related to computer program instructions, and the computer program can be stored in a non-volatile computer-readable storage medium, and the computer program can include the process of the embodiment of the above-mentioned method when executed. Among them, any reference to memory or other media in each embodiment provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0118] It should be noted that the information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0119] This application uses crawler technology to obtain multimodal data such as audio, video and text in specific social hot spots from different channels, and uses a large language model to process different types of data. For audio data, timbre features are extracted to identify voiceprints, and the core essence of the audio content is identified through semantic analysis. For video data, the face area in the video is identified and the corresponding face depth features are extracted to identify the characters in the video. For text data, the main content is extracted and summarized. On the basis of extracting these features, the large language model is used to further identify key elements such as time, place, person and event specific information, and the main context of the event is inferred using a deep learning algorithm, and the key attribute structured data of the event is stored in the database. After that, all events in the database are clustered based on the spectral clustering algorithm, and the evolution chain of the event is further obtained using time series and causal associations, so as to achieve more convenient public opinion information grasp and rapid public opinion development analysis, and predict and respond to potential risks and hidden dangers of the event. The method of this application plays an important role in many fields such as government public relations, corporate brand management, and emergency response, and significantly improves the level of social management and public security protection capabilities.

[0120] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0121] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for identifying potential risks and hidden dangers based on multi-dimensional information and event causal analysis, characterized in that: include: Obtain multimodal data from subscribed public opinion monitoring objects, including audio data, video data, and text data; Perform feature extraction on the multimodal data to extract multimodal information features including timbre features, face features, and text information features; Performing data structured analysis based on multimodal information features to extract key attribute structured data of each event; the key attribute structured data of each event includes time, location, people involved, specific event information, and risk index; Perform cluster analysis on all events based on the key attribute structured data of each event, and divide all events into a series of large event sets; Perform information deduplication and integration analysis on each small event in the large event set to obtain key attribute structured data of each large event set; Based on the key attribute structured data of each major event set, the causal chain of the occurrence sequence and causal relationship of a series of major event sets is sorted out to obtain the causal chain of public opinion monitoring events; Identify potential risks of public opinion monitoring events based on the causal chain of the events.

2. The potential risk identification method based on multi-dimensional information and event causal analysis according to claim 1 is characterized in that: The feature extraction of the multimodal data to extract multimodal information features including timbre features, facial features and text information features specifically includes: Use the audio processing library Librosa to extract the timbre features of audio data and convert the audio data into text data; Use deep convolutional neural networks to identify facial features in video data and use the ArcFace loss function to enhance its recognition ability; With the help of large language models, text information features are extracted from text data, including time, place, people involved, and specific information of events.

3. The potential risk identification method based on multi-dimensional information and event causal analysis according to claim 1 is characterized in that: The data structured analysis based on multimodal information features extracts key attribute structured data of each event, including: Extract the time, location and specific event information corresponding to each event based on text information features; The characters involved in each event are extracted by integrating the timbre features, face features and text information features; The risk index is calculated based on the time, location, event details and people involved in each event.

4. The potential risk identification method based on multi-dimensional information and event causal analysis according to claim 3 is characterized in that: The risk index is calculated based on the time, location, event specific information and people involved in each event, including: Calculate the number of hours t from the current time to the time corresponding to each event, and use the formula Calculate the time risk index R_t; Calculate the straight-line distance s between the location of each event corresponding point and the public opinion monitoring center of the region, and use the formula Calculate the spatial risk index R_s; Based on the specific information of each event and the people involved, according to the formula Calculate the public opinion risk index R_p; where p is the total number of articles involved in the event, including the number of views, comments, reposts, and likes; when the people involved include public figures or the specific information of the event involves social hot spots, it is recorded as flag = true, otherwise it is recorded as flag = false; The risk index R is calculated according to the formula R = εt×R_t+εs×R_s+εp×R_p; where εt, εs and εp are the weights of the time risk index, space risk index and public opinion risk index respectively.

5. The method for identifying potential risks and hidden dangers based on multi-dimensional information and event causal analysis according to claim 1 is characterized in that: The cluster analysis is performed on all events based on the key attribute structured data of each event, and all events are divided into a series of large event sets, including: Construct the feature vector corresponding to each event based on the key attribute structured data of each event; Calculate the Euclidean distance between two events based on the feature vector corresponding to each event; The similarity between two events is calculated based on the Euclidean distance between the two events; Construct a similarity matrix based on the similarity between two events; Based on the similarity matrix, a graph adjacency matrix is ​​constructed, and the spectral clustering algorithm is used for clustering, dividing all events into a series of large event sets; each large event set includes multiple small events.

6. The method for identifying potential risks and hidden dangers based on multi-dimensional information and event causal analysis according to claim 5 is characterized in that: The information of each small event in the large event set is deduplicated and integrated for analysis to obtain key attribute structured data of each large event set, including: The time of each major event set is the earliest time of occurrence of all the small events in the major event set; the location of each major event set includes the locations of all the small events in the major event set; the people involved in each major event set include the people involved in all the small events in the major event set; the event specific information of each major event set is the integration of the event specific information of all the small events in the major event set; the risk index of each major event set is the maximum value of the total risk coefficient of all the small events in the major event set.

7. A device for identifying potential risks and hidden dangers based on multi-dimensional information and event causal analysis, characterized in that: include: The multimodal data acquisition module is used to obtain multimodal data from the subscribed public opinion monitoring objects, including audio data, video data and text data; A data feature extraction module is used to extract features from multimodal data, and extract multimodal information features including timbre features, facial features, and text information features; A data structured analysis module is used to perform data structured analysis based on multimodal information features to extract key attribute structured data of each event; the key attribute structured data of each event includes time, location, people involved, specific event information and risk index; The event clustering analysis module is used to perform clustering analysis on all events based on the key attribute structured data of each event, and divide all events into a series of large event sets; The information deduplication and integration module is used to perform information deduplication and integration analysis on each small event in the large event set to obtain key attribute structured data of each large event set; The event causal chain combing module is used to comb the occurrence sequence and causal relationship of a series of major event sets based on the key attribute structured data of each major event set, and obtain the causal chain of public opinion monitoring events; The potential risk identification module is used to identify potential risks and hidden dangers of events based on the causal chain of public opinion monitoring events.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the potential risk and hidden danger identification method based on multi-dimensional information and event causal combing as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for identifying potential risks and hidden dangers based on multi-dimensional information and event causal analysis as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for identifying potential risks and hidden dangers based on multi-dimensional information and event causal analysis as described in any one of claims 1 to 6 is implemented.

Images