Intelligent public opinion monitoring system and method based on AI

Abstract

The invention provides an AI-based intelligent public opinion monitoring system and method, and the system comprises an information receiving and processing module which is used for receiving target data information in real time, carrying out the modal recognition and analysis of the target data information, and obtaining the public opinion information in the target data information; the intelligent prediction and early warning module is used for performing intelligent prediction and early warning on the public opinion information, determining potential public opinion crisis and performing multi-channel information alarm on the potential public opinion crisis; and the decision report output module is used for analyzing the potential public opinion crisis based on the target large model and outputting a decision report of the potential public opinion crisis. The scientificity and effectiveness of decision making are improved, and the intelligence and effectiveness of public opinion monitoring and the accuracy of public opinion prediction are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent public opinion monitoring, and particularly relates to an intelligent public opinion monitoring system and method based on AI. BACKGROUND

[0002] At present, with the rapid development of the Internet and information technology, information dissemination has become extremely rapid and extensive, and various channels such as social media, news platforms, and network forums produce massive amounts of information every day. In such an environment, the generation and dissemination of public opinion have the characteristics of complexity, diversity, and real-time, so public opinion monitoring is very important.

[0003] However, the traditional public opinion monitoring method often cannot cope with such a large and rapidly changing amount of data, and has problems such as incomplete monitoring, untimely analysis, and inaccuracy, which cannot effectively monitor the situation and trend of public opinion, and cannot quickly generate a corresponding solution according to the specific situation of public opinion, thereby greatly reducing the monitoring effect of public opinion.

[0004] Therefore, in order to overcome the above-mentioned defects, the present application provides an intelligent public opinion monitoring system based on AI. SUMMARY

[0005] The present application provides an intelligent public opinion monitoring system and method based on AI, which collects target data in real time through data crawling technology and receives it, ensures the timeliness of data acquisition, classifies and understands the data by using multi-modal recognition technology, extracts the content related to public opinion from it, effectively realizes the effectiveness and accuracy of multi-modal data processing, effectively determines the potential crisis, and effectively performs multi-channel alarm to determine the relevant personnel to respond quickly, determines the decision report, thereby providing a basis for reasonably dealing with the crisis, improving the scientificity and effectiveness of decision-making, and improving the intelligence, effectiveness of public opinion monitoring and the accuracy of public opinion prediction.

[0006] The present application provides an intelligent public opinion monitoring system based on AI, which comprises:

[0007] An information receiving and processing module is configured to receive target data information in real time, and to perform modal recognition and analysis on the target data information to obtain public opinion information in the target data information.

[0008] An intelligent prediction and early warning module is configured to perform intelligent prediction and early warning on the public opinion information, determine potential public opinion crises, and perform multi-channel information alarm on the potential public opinion crises.

[0009] A decision report output module is configured to analyze the potential public opinion crises based on a target large model, and output a decision report of the potential public opinion crises.

[0010] Preferably, an AI-based intelligent public opinion monitoring system, an information receiving and processing module comprises:

[0011] A data receiving unit is configured to receive target data information in real time.

[0012] A modality recognition unit is configured to perform multi-modal recognition on the target data information to determine a data modality corresponding to the target data information.

[0013] A data analysis unit is configured to recognize the target data information according to the data modality to determine public opinion information in the target data information.

[0014] Preferably, an AI-based intelligent public opinion monitoring system, a data receiving unit comprises:

[0015] A data extraction requirement reading subunit is configured to read a data extraction requirement, determine a data tag and a data extraction target of the data extraction requirement.

[0016] An authorized webpage calling subunit is configured to obtain a webpage information address of an authorized website, and call the authorized webpage according to the webpage information address.

[0017] A data crawling subunit is configured to perform information login operation on the authorized webpage, and crawl target data source information in the authorized webpage after the information login succeeds based on the data tag of the data extraction requirement.

[0018] A target data information obtaining subunit is configured to:

[0019] Position the data extraction target in the target data source information to determine a plurality of key information segments of the data extraction target in the target data source information.

[0020] Integrate the plurality of key information segments to obtain the target data information.

[0021] Preferably, an AI-based intelligent public opinion monitoring system, a modality recognition unit comprises:

[0022] An information reading subunit is configured to read target data information to determine a form of the target data information, wherein a data of the form is equal to or greater than 1.

[0023] A data modality determination subunit is configured to determine a data modality corresponding to the target data information according to the form of the target data information.

[0024] Preferably, an AI-based intelligent public opinion monitoring system, a data analysis unit comprises:

[0025] A data modality reading subunit is configured to read a data modality of target data information, and obtain a modality identifier of the data modality.

[0026] The target recognition mode determination subunit is configured to input the modality identifier of the data modality into a preset recognition library for matching, and determine a target recognition mode matched with the modality identifier.

[0027] The text expression determination subunit is configured to perform information conversion on the target data information according to the target recognition mode, and obtain a text expression corresponding to the target data information.

[0028] The public opinion information extraction subunit is configured to:

[0029] identify the text expression to obtain a public opinion theme corresponding to the text expression, and simultaneously acquire a key text field related to the public opinion theme;

[0030] acquire a public opinion information extraction dimension, and extract effective information in the key text field according to the public opinion information extraction dimension to obtain public opinion information in the target data information;

[0031] The public opinion information includes media attention, user attitude in a preset time period, and media attention and user attitude in a ring ratio of the preset time period.

[0032] Preferably, an AI-based intelligent public opinion monitoring system includes an intelligent prediction and early warning module.

[0033] The first division unit is configured to read the public opinion information, determine an information category of the public opinion information, and simultaneously divide the public opinion information according to the information category to obtain sub-public opinion information corresponding to each information category.

[0034] The second division unit is configured to divide the sub-public opinion information corresponding to each information category according to a preset time period to obtain a plurality of data segments of each sub-public opinion information.

[0035] The data transformation curve acquisition unit is configured to map the plurality of data segments of each sub-public opinion information in a preset rectangular coordinate system according to the preset time period to obtain a data point of each sub-public opinion information in the preset rectangular coordinate system, and simultaneously perform smoothing processing on the data point of each sub-public opinion information in the preset rectangular coordinate system to obtain a data transformation curve of each sub-public opinion information.

[0036] The curve analysis unit is configured to:

[0037] read the data transformation curve of each sub-public opinion information to determine a curve inflection point of the data transformation curve corresponding to each sub-public opinion information.

[0038] divide the data transformation curve into a plurality of curve segments according to the curve inflection point of the data transformation curve, simultaneously read data monotonicity corresponding to each curve segment, and simultaneously determine a reference range of the data transformation curve according to a preset standard.

[0039] According to the data monotonicity corresponding to each curve segment, the trend of change of each curve segment is determined, and the data monotonicity corresponding to the last preset time period is obtained, and the target trend of change of the next preset time period is predicted according to the data monotonicity corresponding to the last preset time period.

[0040] The qualified determination unit is configured to determine whether the target curve segment of the next time period corresponding to the last preset time period is qualified according to the reference range of the data transformation curve and the target trend of change.

[0041] The alarm unit is configured to generate a potential public opinion crisis report when it is determined in the prediction result that the target curve segment of the next time period is unqualified, and perform multi-channel information alarm on the potential public opinion crisis report.

[0042] Preferably, an AI-based intelligent public opinion monitoring system, a decision report output module, comprises:

[0043] The model training unit is configured to call multiple industry data, and obtain a basic model framework, and train the basic model framework according to the multiple industry data to obtain a target large model.

[0044] The association unit is configured to call a preset industry knowledge base in the system, and associate the target large model with the preset industry knowledge base.

[0045] Preferably, an AI-based intelligent public opinion monitoring system, a decision report output module, comprises:

[0046] The public opinion crisis analysis unit is configured to:

[0047] read a potential public opinion crisis based on the target large model, and extract attribute parameters of the potential public opinion crisis based on the reading result;

[0048] analyze the attribute parameters based on the target large model to obtain a risk level and a corresponding influence range of the potential public opinion crisis, and obtain an evaluation result of the potential public opinion crisis based on the risk level and the influence range;

[0049] The decision report generation unit is configured to parse the evaluation result based on the industry knowledge base to obtain a coping decision corresponding to the potential public opinion crisis, and generate a decision report of the potential public opinion crisis based on the coping decision.

[0050] The decision dynamic adjustment unit is configured to:

[0051] respond to the potential public opinion crisis based on the decision report, and monitor a response result in real time to obtain a change feature of the potential public opinion crisis.

[0052] The target large model is used to analyze the change characteristics in real time, and the decision report is dynamically modified based on the real-time analysis result.

[0053] The application provides an AI-based intelligent public opinion monitoring method, which comprises the following steps:

[0054] Step 1: Real-time receiving of target data information, and modal recognition and analysis of the target data information to obtain public opinion information in the target data information;

[0055] Step 2: Intelligent prediction and early warning of the public opinion information to determine a potential public opinion crisis, and multi-channel information alarm of the potential public opinion crisis;

[0056] Step 3: Analysis of the potential public opinion crisis based on a target large model to output a decision report of the potential public opinion crisis.

[0057] Preferably, the AI-based intelligent public opinion monitoring method comprises the following steps:

[0058] Real-time receiving of target data information;

[0059] Multi-modal recognition of the target data information to determine a data mode corresponding to the target data information;

[0060] Recognition of the target data information according to the data mode to determine public opinion information in the target data information.

[0061] Compared with the prior art, the application has the following beneficial effects:

[0062] The target data is collected and received in real time through the data crawling technology, ensuring the timeliness of data acquisition, the multi-modal recognition technology is used to classify and understand the data, the content related to public opinion is extracted, the effectiveness and accuracy of multi-modal data processing are effectively realized, the multi-channel alarm is effectively performed by determining the potential crisis, the relevant personnel can be determined to respond quickly, the decision report is determined, and thus the basis for reasonably coping with the crisis is provided, the scientificity and effectiveness of decision-making are improved, and the intelligence, effectiveness of public opinion monitoring and the accuracy of public opinion prediction are improved.

[0063] Other features and advantages of the application will be set forth in the following description of the application, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the application. The objects and other advantages of the application can be achieved and obtained by the structures specifically pointed out in the application document.

[0064] The technical solutions of the application will be further described in detail below with reference to the drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0065] The accompanying drawings are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and are used to explain the application without restricting it. In the drawings:

[0066] Figure 1 The structure diagram of an AI-based intelligent public opinion monitoring system in an embodiment of the application;

[0067] Figure 2 The structure diagram of an information receiving and processing module in an AI-based intelligent public opinion monitoring system in an embodiment of the application;

[0068] Figure 3 The flowchart of an AI-based intelligent public opinion monitoring method in an embodiment of the application. DETAILED DESCRIPTION

[0069] The preferred embodiments of the application are described below in conjunction with the accompanying drawings, and it should be understood that the preferred embodiments described herein are only used to explain and illustrate the application, and do not limit the application.

[0070] In one embodiment, an AI-based intelligent public opinion monitoring system is provided, as shown in Figure 1 , comprising:

[0071] An information receiving and processing module is configured to receive target data information in real time, and to perform modal recognition and analysis on the target data information to obtain public opinion information in the target data information.

[0072] An intelligent prediction and early warning module is configured to perform intelligent prediction and early warning on the public opinion information, determine potential public opinion crises, and perform multi-channel information alarm on the potential public opinion crises.

[0073] A decision report output module is configured to analyze the potential public opinion crises based on a target large model, and output a decision report of the potential public opinion crises.

[0074] In this embodiment, the target data information can include pictures, videos, text and other information related to the public opinion status obtained from web pages, wherein the modal recognition can be used to identify the presentation form of the target data information, which can be a picture, a video or a text presentation form.

[0075] In this embodiment, the public opinion information can include media attention, user attitude in a preset time period, and the media attention and user attitude in the preset time period.

[0076] In this embodiment, the multi-channel information alarm can be a short message, an email, a system prompt, etc.

[0077] In this embodiment, the decision report can include the situation of the potential crisis, the possible impact and the suggested coping strategies.

[0078] In this embodiment, the target data information can be based on AI intelligent crawler to realize efficient collection and semantic understanding of network information.

[0079] In this embodiment, the most advanced large model is used as the base model, trained and fine-tuned with industry data, and connected to the industry knowledge base to realize the intelligence and accuracy of decision report generation.

[0080] The working principle and beneficial effects of the above technical solution are: through the technology of data crawling, the target data is collected and received in real time, ensuring the timeliness of data acquisition, through the use of multi-modal recognition technology, the data is classified and understood, and the content related to public opinion is extracted, effectively realizing the effectiveness and accuracy of multi-modal data processing, through determining the potential crisis, effectively performing multi-channel alarm can effectively determine the relevant personnel to respond quickly, by determining the decision report, thereby facilitating the provision of basis for reasonably dealing with crisis, improving the scientificity and effectiveness of decision-making, and improving the intelligence, effectiveness of public opinion monitoring and accuracy of public opinion prediction.

[0081] In one embodiment, an AI-based intelligent public opinion monitoring system is provided, as shown in Figure 2 The information receiving and processing module includes:

[0082] The data receiving unit is configured to receive target data information in real time.

[0083] The modal recognition unit is configured to perform multi-modal recognition on the target data information to determine the data modal corresponding to the target data information.

[0084] The data analysis unit is configured to identify the target data information according to the data modal to determine the public opinion information in the target data information.

[0085] In this embodiment, multi-modal recognition can be recognition of different types of data in the target data information.

[0086] In this embodiment, the data modal can be the data representation form contained in the target data information, such as pictures, videos, and texts, etc.

[0087] The working principle and beneficial effects of the above technical solution are: by receiving the target data information in real time, and performing multi-modal recognition on the received target data information, the data modal contained in the target data information is effectively determined, and finally the target data information is identified according to the data modal, realizing accurate and effective determination of the public opinion information in the target data information, providing reliable data support for realizing public opinion monitoring.

[0088] In one embodiment, an AI-based intelligent public opinion monitoring system is provided, and the data receiving unit includes:

[0089] The data extraction requirement reading subunit is configured to read the data extraction requirement, determine a data tag of the data extraction requirement, and determine a data extraction target;

[0090] The authorized webpage calling subunit is configured to obtain a webpage information address of the authorized website, and call the authorized webpage according to the webpage information address;

[0091] The data crawling subunit is configured to perform information login operation on the authorized webpage, and crawl target data source information in the authorized webpage based on the data tag of the data extraction requirement after the information login is successful,

[0092] The target data information obtaining subunit is configured to:

[0093] According to the data extraction target, the target data source information is positioned, and a plurality of key information segments of the data extraction target in the target data source information are determined;

[0094] The plurality of key information segments are integrated to obtain the target data information.

[0095] In this embodiment, the data tag represents a data extraction representation determined according to the data extraction requirement, which is used to realize a key element for extracting and identifying data.

[0096] In this embodiment, the authorized website is a website that can extract information from a known preset website library, wherein the preset website library stores webpage information addresses of each authorized website.

[0097] In this embodiment, the key information segment is core information in the target data source information determined according to the data extraction target.

[0098] The working principle and beneficial effects of the above technical solution are as follows: the data tag and the extraction target are accurately determined, which provides a clear direction for the entire data extraction process, avoids blind data extraction, and improves the accuracy of data acquisition; through the authorization mechanism, only users or systems with legal authorization can obtain webpage information, thereby protecting the security and privacy of webpage data; according to the data tag, relevant data in the authorized webpage can be accurately found, thereby avoiding obtaining a large amount of irrelevant data and improving data acquisition efficiency; through positioning and integrating the key information segment, information meeting the data extraction target can be accurately extracted from complex data sources, thereby improving the usability of data.

[0099] In one embodiment, an AI-based intelligent public opinion monitoring system is provided, and a modal recognition unit includes:

[0100] The information reading subunit is configured to read the target data information and determine a form of the target data information, wherein the data of the form is equal to or greater than 1;

[0101] The data modality determination subunit is configured to determine the data modality corresponding to the target data information according to the form of the target data information.

[0102] The above technical solution has the beneficial effects of accurately reading the target data information and determining the form thereof, and then determining the data modality, which helps to classify and understand the data, makes the data processing more targeted, and improves the data utilization efficiency.

[0103] In one embodiment, an AI-based intelligent public opinion monitoring system is provided, and the data analysis unit comprises:

[0104] The data modality reading subunit is configured to read the data modality of the target data information, and simultaneously obtain the modality identifier of the data modality.

[0105] The target recognition mode determination subunit is configured to input the modality identifier of the data modality into a preset recognition library for matching, and determine the target recognition mode matched with the modality identifier.

[0106] The text expression determination subunit is configured to perform information conversion on the target data information according to the target recognition mode, and obtain the text expression corresponding to the target data information.

[0107] The public opinion information extraction subunit is configured to:

[0108] identify the text expression, obtain the public opinion theme corresponding to the text expression, and simultaneously obtain the key text field related to the public opinion theme.

[0109] obtain the public opinion information extraction dimension, and extract the effective information in the key text field according to the public opinion information extraction dimension, to obtain the public opinion information in the target data information.

[0110] The public opinion information comprises the media attention degree, the user attitude in a preset time period, and the media attention degree and the user attitude in the preset time period.

[0111] In this embodiment, the preset recognition library is set in advance, and it contains the mapping relationship between each data modality identifier and the response recognition mode.

[0112] In this embodiment, the target data information is converted, for example, if the target data information is numerical, it can be converted into a text form describing the numerical meaning; if it is an image, the key elements in the image are extracted and described in words; through the conversion, the text expression corresponding to the target data information is obtained.

[0113] In this embodiment, the text expression determined by the text expression determination subunit is analyzed in depth. Natural language processing techniques such as morphological analysis, syntactic analysis, semantic understanding, etc. are used to identify the main topic in the text expression, i.e. the public opinion theme. At the same time, the key text fields closely related to the public opinion theme are determined, which may be parts containing key opinions, event descriptions, etc.

[0114] The working principle and beneficial effects of the above technical solutions are: by determining the data modalities and determining the appropriate target recognition method, different types of data are converted into text expressions, so that the data can be presented in a unified and easy-to-understand form; this helps to integrate and analyze different types of data, improving the availability of data; based on text expression, public opinion theme recognition and effective information extraction can accurately capture the media attention, user attitude and its ring ratio change in the preset time period, thereby improving the accuracy and effectiveness of obtaining public opinion information.

[0115] In one embodiment, an AI-based intelligent public opinion monitoring system is provided, and the intelligent prediction and early warning module comprises:

[0116] The first division unit is configured to read the public opinion information, determine the information category of the public opinion information, and simultaneously divide the public opinion information according to the information category to obtain sub-public opinion information corresponding to each information category;

[0117] The second division unit is configured to divide the sub-public opinion information corresponding to each information category according to a preset time period to obtain a plurality of data segments of each sub-public opinion information;

[0118] The data transformation curve acquisition unit is configured to map the plurality of data segments of each sub-public opinion information in a preset rectangular coordinate system according to the preset time period to obtain data points of each sub-public opinion information in the preset rectangular coordinate system, and simultaneously perform smoothing processing on the data points of each sub-public opinion information in the preset rectangular coordinate system to obtain a data transformation curve of each sub-public opinion information;

[0119] The curve analysis unit is configured to:

[0120] read the data transformation curve of each sub-public opinion information, and determine a curve inflection point of the data transformation curve corresponding to each sub-public opinion information;

[0121] divide the data transformation curve into a plurality of curve segments according to the curve inflection point of the data transformation curve, simultaneously read the data monotonicity corresponding to each curve segment, and simultaneously determine a reference range of the data transformation curve according to a preset standard;

[0122] According to the data monotonicity corresponding to each curve segment, the change trend of each curve segment is determined, and meanwhile, the data monotonicity corresponding to the last preset time period is obtained, and the target change trend of the next preset time period is predicted according to the data monotonicity corresponding to the last preset time period.

[0123] The qualified determination unit is configured to determine whether the target curve segment of the next time period corresponding to the last preset time period is qualified according to the reference range of the data change curve and the target change trend.

[0124] The alarm unit is configured to generate a potential public opinion crisis report when it is determined in the prediction result that the target curve segment of the next time period is unqualified, and perform multi-channel information alarm on the potential public opinion crisis report.

[0125] In this embodiment, the sub-public opinion information is the result of dividing the public opinion information based on the information category, and the sub-public opinion information is one-to-one corresponding to the information category.

[0126] In this embodiment, the plurality of data segments are the result of dividing the sub-public opinion information according to the preset time period (i.e. the time period set in advance).

[0127] In this embodiment, the target change trend of the next preset time period predicted according to the data monotonicity corresponding to the last preset time period can be obtained by linkage learning (i.e. feature learning of the monotonicity of each curve segment and the change trend of the next stage curve segment) of the monotonicity corresponding to each curve segment and the change trend, and then obtaining the change feature of the monotonicity to the change trend of the next time period, so that the target change trend of the next preset time period can be predicted according to the data monotonicity corresponding to the last preset time period.

[0128] In this embodiment, the determination of whether the target curve segment of the next time period corresponding to the last preset time period is qualified according to the reference range of the data change curve and the target change trend means that the reference range of the data change curve is taken as the target standard, so that the target change trend is compared with the reference range of the data change curve to determine whether the target curve of the next time period corresponding to the last preset time period is qualified (i.e. when the target change trend belongs to the reference range of the data change curve, it is determined that the target curve of the next time period corresponding to the last preset time period is qualified, otherwise, it is determined that the target curve of the next time period corresponding to the last preset time period is unqualified).

[0129] The working principle and beneficial effects of the above technical solution are: through the information category, the information of public opinion information is effectively divided, and through the preset time period, the sub-public opinion information is divided, so as to effectively obtain a plurality of data segments, and then lay a foundation for obtaining the data transformation curve of each sub-public opinion information. Through the analysis of each sub-public opinion information, the target trend of the next preset time period corresponding to the last preset period can be effectively predicted, so as to determine whether the target curve segment of the next time period corresponding to the last preset time period is qualified, and then effectively realize the rapid and accurate generation of potential public opinion report. It is beneficial to ensure the accuracy and timeliness of generating potential public opinion report, so as to realize effective monitoring of public opinion and timely warning of potential crisis.

[0130] In one embodiment, an AI-based intelligent public opinion monitoring system is provided, and a decision report output module includes:

[0131] A model training unit is configured to call multi-industry data, obtain a basic model framework, and train the basic model framework based on the multi-industry data to obtain a target large model.

[0132] An association unit is configured to call a preset industry knowledge base in the system and associate the target large model with the preset industry knowledge base.

[0133] In this embodiment, the basic model framework is set in advance and serves as a basic structural support for training multi-industry data.

[0134] In this embodiment, the preset industry knowledge base is set in advance.

[0135] In this embodiment, the association between the target large model and the preset industry knowledge base can be achieved by establishing data mapping, indexing, and feature matching to realize corresponding connection in semantics, structure, and function, thereby achieving association.

[0136] The working principle and beneficial effects of the above technical solution are: by calling multi-industry data, the basic model framework is trained to effectively obtain the target large model, thereby associating the target large model with the preset industry knowledge base, thereby improving the processing capacity and accuracy of public opinion decision-making, and improving work efficiency.

[0137] In one embodiment, an AI-based intelligent public opinion monitoring system is provided, and a decision report output module includes:

[0138] A public opinion crisis analysis unit is configured to:

[0139] Based on the target large model, a potential public opinion crisis is read, and based on the reading result, the attribute parameters of the potential public opinion crisis are extracted.

[0140] The target large model is used to analyze the attribute parameters, to obtain a risk level and a corresponding influence range of the potential public opinion crisis, and to obtain an evaluation result of the potential public opinion crisis based on the risk level and the influence range.

[0141] The decision report generation unit is configured to analyze the evaluation result based on the industry knowledge base, to obtain a coping decision corresponding to the potential public opinion crisis, and to generate a decision report of the potential public opinion crisis based on the coping decision.

[0142] The decision dynamic adjustment unit is configured to:

[0143] The decision dynamic adjustment unit is configured to:

[0144] The target large model is used to analyze the attribute parameters, to obtain a risk level and a corresponding influence range of the potential public opinion crisis, and to obtain an evaluation result of the potential public opinion crisis based on the risk level and the influence range.

[0145] In this embodiment, the attribute parameters of the potential public opinion crisis include the public opinion propagation speed, the involved subjects, and the severity of the event.

[0146] In this embodiment, the attribute parameters are analyzed by the target large model. Through the algorithm and pre-training knowledge inside the model, the risk level (such as low, medium, or high) and the corresponding influence range (for example, local influence or widespread influence) of the potential public opinion crisis are evaluated, and finally the evaluation result of the potential public opinion crisis is obtained.

[0147] In this embodiment, the industry knowledge base is used as a reference to analyze the evaluation result. The industry knowledge base contains knowledge such as coping experience for different types of public opinion crises and industry standards. According to the evaluation result, the appropriate coping strategy is found in the knowledge base, so that the coping decision for the potential public opinion crisis is obtained, and a complete decision report of the potential public opinion crisis is generated based on the coping decision.

[0148] In this embodiment, the potential public opinion crisis is responded to according to the decision content in the decision report, and the development of the public opinion crisis is monitored in real time during the response process. For example, whether the heat of the public opinion is reduced, whether the propagation direction is changed, and the like are monitored, so that the change characteristics of the potential public opinion crisis are obtained. Finally, the target large model is used to analyze the real-time obtained change characteristics, and the previously generated decision report is dynamically modified according to the analysis result, so that the coping decision can adapt to the dynamic development of the public opinion crisis.

[0149] The working principle and beneficial effects of the above technical solution are: through the multi-dimensional analysis of the target large model on the potential public opinion crisis, the attribute parameters can be more accurately extracted and the risk level and influence range can be determined, avoiding the subjectivity and limitations of manual evaluation, providing a scientific basis for subsequent response decisions; using the industry knowledge base to analyze and evaluate the results to generate a decision report, which can quickly draw on mature experience and best practices within the industry, improve the efficiency and rationality of decision-making, and ensure that the response decisions meet industry standards and requirements; real-time monitoring of changes in public opinion crises and timely correction of decision reports enable response measures to be continuously optimized as public opinion develops, enhancing the flexibility and adaptability of responding to public opinion crises and improving the overall effectiveness of responding to public opinion crises.

[0150] In one embodiment, an AI-based intelligent public opinion monitoring method is provided, as shown in Figure 3 , comprising:

[0151] Step 1: Real-time reception of target data information, while performing modal recognition and analysis on the target data information to obtain public opinion information in the target data information;

[0152] Step 2: Intelligent prediction and early warning of public opinion information, determining potential public opinion crises, and simultaneously performing multi-channel information alarm on the potential public opinion crises;

[0153] Step 3: Analysis of potential public opinion crises based on a target large model, outputting a decision report for the potential public opinion crises.

[0154] The working principle and beneficial effects of the above technical solution are: through the use of data crawling technology to collect and receive target data in real time, ensuring the timeliness of data acquisition, through the use of multi-modal recognition technology to classify and understand the data, extracting the content related to public opinion, effectively realizing the effectiveness and accuracy of multi-modal data processing, effectively determining the relevant personnel to respond quickly through multi-channel alarm by determining potential crises, and providing a basis for rational response to crises by determining a decision report, improving the scientificity and effectiveness of decision-making, and improving the intelligence, effectiveness of public opinion monitoring and the accuracy of public opinion prediction.

[0155] In one embodiment, an AI-based intelligent public opinion monitoring method is provided, step 1, comprising:

[0156] Real-time reception of target data information;

[0157] Multi-modal recognition of the target data information to determine the data modal corresponding to the target data information;

[0158] According to the data modal, the target data information is recognized to determine the public opinion information in the target data information.

[0159] The working principle and beneficial effects of the technical solution are as follows: target data information is received in real time, the received target data information is subjected to multi-modal recognition, the data modalities contained in the target data information are effectively determined, the target data information is recognized according to the data modalities, the public opinion information in the target data information is accurately and effectively determined, and reliable data support is provided for realizing public opinion monitoring.

[0160] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the claims of the present application and their equivalents, the present application also intends to include these modifications and variations.

Claims

1. An AI-based intelligent public opinion monitoring system, characterized in that, include: The information receiving and processing module is used to receive target data information in real time, and at the same time, perform modal recognition and analysis on the target data information to obtain public opinion information from the target data information; The intelligent prediction and early warning module is used to intelligently predict and warn about public opinion information, identify potential public opinion crises, and simultaneously send multi-channel information alerts about potential public opinion crises. The decision report output module is used to analyze potential public opinion crises based on the target big model and output decision reports on potential public opinion crises.

2. The AI-based intelligent public opinion monitoring system according to claim 1, characterized in that, The information receiving and processing module includes: The data receiving unit is used to receive target data information in real time; The modality recognition unit is used to perform multimodal recognition on target data information and determine the data modality corresponding to the target data information; The data analysis unit is used to identify target data information based on data modalities and determine the public opinion information within the target data information.

3. The AI-based intelligent public opinion monitoring system according to claim 2, characterized in that, The data receiving unit includes: The data extraction requirement reading subunit is used to read the data extraction requirements and determine the data labels and data extraction targets for the data extraction requirements. The authorized webpage retrieval subunit is used to obtain the webpage information address of the authorized website, and at the same time, retrieve the authorized webpage based on the webpage information address; The data crawling subunit is used to perform information login operations on authorized web pages. Simultaneously, based on data extraction requirements, it crawls target data source information from the authorized web pages after successful information login using data tags. The target data information acquisition subunit is used for: Based on the location of the data extraction target in the target data source information, the multiple key information segments of the data extraction target in the target data source information are determined. By integrating multiple key information segments, the target data information can be obtained.

4. The AI-based intelligent public opinion monitoring system according to claim 2, characterized in that, Modality recognition unit, including: The information reading subunit is used to read the target data information and determine the representation of the target data information, wherein the data of the representation is equal to or greater than 1; The data modality determination subunit is used to determine the data modality corresponding to the target data information based on the form in which the target data information is presented.

5. The AI-based intelligent public opinion monitoring system according to claim 2, characterized in that, The data analysis unit includes: The data modality reading subunit is used to read the data modality of the target data information and, at the same time, obtain the modality identifier of the data modality; The target recognition method determination subunit is used to input the modality identifier of the data modality into a preset recognition library for matching, and determine the target recognition method that matches the modality identifier; The text representation determination subunit is used to convert target data information according to the target recognition method to obtain the text representation corresponding to the target data information. The public opinion information extraction subunit is used for: The text expression is identified to obtain the corresponding public opinion theme, and at the same time, the key text fields related to the public opinion theme are obtained. Obtain the dimensions for extracting public opinion information, and extract the effective information from the key text fields according to the dimensions to obtain the public opinion information in the target data information; The public opinion information includes: media attention and user attitudes within a preset time period, as well as the month-on-month comparison of media attention and user attitudes within the preset time period.

6. The AI-based intelligent public opinion monitoring system according to claim 1, characterized in that, The intelligent prediction and early warning module includes: The first segmentation unit is used to read public opinion information, determine the information category of public opinion information, and segment the public opinion information according to the information category to obtain the sub-public opinion information corresponding to each information category. The second division unit is used to divide the sub-public opinion information corresponding to each information category according to a preset time period, and obtain several data segments for each sub-public opinion information. The data transformation curve acquisition unit is used to map several data segments of each sub-public opinion information in a preset rectangular coordinate system according to a preset time period to obtain the data points of each sub-public opinion information in the preset rectangular coordinate system. At the same time, the data points of each sub-public opinion information in the preset rectangular coordinate system are smoothed to obtain the data transformation curve of each sub-public opinion information. Curve analysis unit, used for: Read the data transformation curve for each sub-public opinion information and determine the inflection point of the corresponding data transformation curve for each sub-public opinion information; The data transformation curve is divided into several curve segments based on the inflection point of the curve. At the same time, the monotonicity of the data corresponding to each curve segment is read, and the reference range of the data transformation curve is determined according to the preset standard. The trend of change of each curve segment is determined based on the monotonicity of the data corresponding to each curve segment. At the same time, the monotonicity of the data corresponding to the last preset time period is obtained, and the target trend of change in the next preset time period is predicted based on the monotonicity of the data corresponding to the last preset time period. The pass / fail determination unit is used to predict whether the target curve segment corresponding to the next time period is qualified based on the baseline range of the data transformation curve and the target change trend. The alarm unit is used to generate a potential public opinion crisis report when the target curve segment for the next time period is deemed unqualified in the prediction results, and to send the potential public opinion crisis report to multiple information alarms.

7. The AI-based intelligent public opinion monitoring system according to claim 1, characterized in that, The decision report output module includes: The model training unit is used to retrieve data from multiple industries and obtain a basic model framework. At the same time, it trains the basic model framework based on the data from multiple industries to obtain the target large model. The association unit is used to retrieve a preset industry knowledge base in the system and associate the target large model with the preset industry knowledge base.

8. The AI-based intelligent public opinion monitoring system according to claim 1, characterized in that, The decision report output module includes: The public opinion crisis analysis unit is used for: The potential public opinion crisis is read based on the target large model, and the attribute parameters of the potential public opinion crisis are extracted based on the reading results; Based on the target big model, the attribute parameters are analyzed to obtain the risk level and corresponding impact range of potential public opinion crises, and the assessment results of potential public opinion crises are obtained based on the risk level and impact range. The decision report generation unit is used to analyze the evaluation results based on the industry knowledge base, obtain the corresponding response decisions for potential public opinion crises, and generate a decision report on potential public opinion crises based on the response decisions. The decision-making dynamic adjustment unit is used for: Responses to potential public opinion crises are based on decision-making reports, and the results of these responses are monitored in real time to obtain the changing characteristics of potential public opinion crises. The system performs real-time analysis of change characteristics based on the target large model, and dynamically corrects the decision report based on the real-time analysis results.

9. An AI-based intelligent public opinion monitoring method, characterized in that, include: Step 1: Receive target data information in real time, and simultaneously perform modal recognition and analysis on the target data information to obtain public opinion information from the target data information; Step 2: Conduct intelligent prediction and early warning of public opinion information, identify potential public opinion crises, and simultaneously send multi-channel information alerts about potential public opinion crises; Step 3: Analyze potential public opinion crises based on the target big model and output a decision report on potential public opinion crises.

10. The AI-based intelligent public opinion monitoring method according to claim 9, characterized in that, Step 1 includes: Receive target data information in real time; Multimodal identification is performed on the target data information to determine the corresponding data modality; based on the data modality, the public opinion information in the target data information is identified.

Images