Cloud computing-based financial risk management method and system

Abstract

This invention relates to the field of financial risk management technology, and particularly to a cloud-based financial risk management method and system. The method includes: identifying risk subjects requiring financial risk management; selecting a data source platform; capturing behavioral data containing the risk subjects; querying a pre-constructed key dictionary, wherein the key dictionary consists of keywords and corresponding sentiment scores; calculating the sentiment scores of the behavioral data; and summing all sentiment scores to generate a total score, where each risk subject corresponds to a total score. This invention, by writing communication reminders, helps to promptly detect the operational risk information of risk subjects, controlling risks at the nascent stage and avoiding the impact of insufficient or missing information on financial decisions. This not only improves the accuracy and timeliness of financial risk identification but also enhances the effectiveness of risk assessment, providing more reliable data support for individual financial decisions.

Description

Technical Field

[0001] This invention relates to the field of financial risk management technology, and in particular to a cloud-based financial risk management method and system. Background Technology

[0002] Financial risk management aims to analyze various risks in financial activities that may affect financial assets, investments, businesses, or individuals, and to take corresponding measures to reduce potential losses.

[0003] In the current financial environment, individuals typically rely on publicly available financial statements and news reports to assess a company's operations and potential financial risks when purchasing stocks, making financial investments, or engaging in partnerships. However, this publicly available information often lacks comprehensiveness, making it difficult for individuals to accurately judge the actual level of a company's financial risk. For example, gathering opinions from employees of the target company can provide insights into the company's true operating information and offer a reference for risk assessment.

[0004] Therefore, "how to obtain multi-source data for financial risk management" is the technical problem that this invention aims to solve. Summary of the Invention

[0005] The purpose of this invention is to provide a cloud computing-based financial risk management method and system to solve the problem of "how to obtain multi-source data for financial risk management" mentioned in the background.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A cloud-based financial risk management method, comprising:

[0008] Identify risk subjects that require financial risk management, select data source platforms, capture behavioral data containing risk subjects, query a pre-built key dictionary, which consists of keywords and corresponding sentiment scores, calculate the sentiment scores of behavioral data, and aggregate all sentiment scores to generate a total score, where each risk subject corresponds to a total score, count the number of behavioral data, and calculate the average score.

[0009] Collect attribute data of risk subjects and cluster the data source platforms into industry-related platforms and broad monitoring platforms. When the average score deviates from the preset fluctuation range, extract the core features from the behavioral data, write them into a preset template, generate a topic, and publish it to the industry-related platforms. Through the behavioral data, traverse the relationship accounts of the risk subjects, generate interaction invitations, and write them into the topic.

[0010] Obtain feedback data under the topic, construct a word set composed of sensitive words, determine whether there are sensitive words in the feedback data, if so, summarize all feedback data containing sensitive words, generate risk management results, write communication reminder information, and send it to preset terminals, if not, activate the pre-edited targeted monitoring mechanism and update the risk management results.

[0011] Furthermore, the steps for identifying risk entities requiring financial risk management and selecting data source platforms include:

[0012] From the data source platform, define the publishers and disseminators of behavioral data, and draw the dissemination chain;

[0013] Set several filtering criteria to remove useless parts of the behavioral data.

[0014] Furthermore, the step of calculating the sentiment score of the behavioral data, summing all the sentiment scores, and generating a total score includes:

[0015] The total score is updated according to a preset step size, and a score evolution curve is plotted with time as the horizontal axis and the total score at the corresponding time as the vertical axis.

[0016] The score evolution curve is divided into an ascending segment and a descending segment. The descending segment is identified, and the behavioral data is updated.

[0017] Furthermore, the step of extracting the core features from the behavioral data and writing them into a preset template includes:

[0018] The most frequently occurring keywords in the statistical behavioral data are defined as target words. The target words and preset templates are input into the language model, and the output is the topic.

[0019] Set the validity period of the topic, extract the expiration time, obtain the generation time of the risk management result, and align the expiration time and generation time.

[0020] Furthermore, the step of obtaining feedback data under the topic and constructing a word set composed of sensitive words includes:

[0021] Insert abnormal words into the word set and establish a mapping between abnormal words and risk levels;

[0022] When abnormal words are detected in the feedback data, the communication reminder information is updated and a tag generated by the risk level is inserted.

[0023] Furthermore, the step of updating the risk management result includes:

[0024] Configure influencing factors for a preset frequency and dynamically adjust the preset frequency, wherein the influencing factors include at least: time period;

[0025] Set up a cloud computing node and connect it to the data source platform to upload the key dictionary to the cloud computing node.

[0026] Furthermore, the system includes:

[0027] The calculation module is used to identify risk subjects that need financial risk management, select data source platforms, capture behavioral data containing risk subjects, query a pre-built key dictionary, which consists of keywords and corresponding sentiment scores, calculate the sentiment scores of behavioral data, and superimpose all sentiment scores to generate a total score, where each risk subject corresponds to a total score, count the number of behavioral data, and calculate the average score.

[0028] The writing module is used to collect attribute data of risk subjects and cluster the data source platforms into industry-related platforms and broad monitoring platforms. When the average score deviates from the preset fluctuation range, the core features in the behavioral data are extracted, written into the preset template, a topic is generated, and published to the industry-related platform. Through the behavioral data, the relationship accounts of the risk subjects are traversed, an interaction invitation is generated, and written into the topic.

[0029] The judgment module is used to obtain feedback data under the topic, construct a word set composed of sensitive words, determine whether there are sensitive words in the feedback data, if so, summarize all feedback data containing sensitive words, generate risk management results, write communication reminder information, and send it to preset terminals, if not, activate the pre-edited targeted monitoring mechanism and update the risk management results.

[0030] Furthermore, the computing module includes:

[0031] The definition unit is used to define the publishers and disseminators of behavioral data from the data source platform and to draw the dissemination chain;

[0032] The delete unit is used to set several filtering criteria to delete useless parts of behavioral data;

[0033] The drawing unit is used to update the total score according to a preset step size, and draw the score evolution curve with time as the horizontal axis and the total score at the corresponding time as the vertical axis.

[0034] The identification unit is used to divide the score evolution curve into an ascending segment and a descending segment, identify the descending segment, and update the behavioral data.

[0035] Furthermore, the writing module includes:

[0036] The statistical unit is used to analyze the most frequently occurring keywords in the statistical behavioral data and defines them as target words. The target words and preset templates are input into the language model, and the output is the topic.

[0037] The setting unit is used to set the validity period of a topic, extract the expiration time, obtain the generation time of the risk management result, and align the expiration time and generation time.

[0038] Furthermore, the determination module includes:

[0039] An insertion unit is used to insert abnormal words into the word set and establish a mapping between abnormal words and risk levels;

[0040] An update unit is used to update the communication reminder information and insert a tag generated by the risk level when abnormal words are detected in the feedback data.

[0041] A configuration unit is used to configure influencing factors of a preset frequency and dynamically adjust the preset frequency, wherein the influencing factors include at least: time periods;

[0042] The upload unit is used to build a cloud computing node and connect to the data source platform to upload the key dictionary to the cloud computing node.

[0043] Compared with the prior art, the beneficial effects of the present invention are:

[0044] By calculating the total score, the risk level of a risk entity can be quantified, facilitating comprehensive judgment. It also enables trend analysis and prediction of risk entities. Calculating the average score allows for comparison and analysis with historical data, enabling timely detection of anomalies. Creating topics guides users to express their opinions, shifting from "passive reception" to "active acquisition," thus collecting more targeted feedback and obtaining implicit information about the true operational status of enterprises. This significantly broadens the sources of financial risk data. Writing communication reminders helps to promptly identify operational risk information of risk entities, controlling risks at their nascent stage and preventing insufficient or missing information from affecting financial decisions. This not only improves the accuracy and timeliness of financial risk identification but also enhances the effectiveness of risk assessment, providing more reliable data support for individual financial decisions. Attached Figure Description

[0045] Figure 1 A flowchart illustrating a cloud-based financial risk management method provided in an embodiment of the present invention;

[0046] Figure 2 A first sub-process flowchart of a cloud-based financial risk management method provided in an embodiment of the present invention;

[0047] Figure 3 A second sub-process flowchart of the cloud-based financial risk management method provided in this embodiment of the invention;

[0048] Figure 4 A third sub-process flowchart of the cloud computing-based financial risk management method provided in this embodiment of the invention;

[0049] Figure 5 A block diagram illustrating the composition of a cloud-based financial risk management system provided in an embodiment of the present invention;

[0050] Figure 6 A block diagram illustrating the composition of the computing module in a cloud-based financial risk management system provided in this embodiment of the invention;

[0051] Figure 7 A block diagram illustrating the composition of the write module in a cloud-based financial risk management system provided in this embodiment of the invention;

[0052] Figure 8 This is a block diagram of the judgment module in a cloud-based financial risk management system provided in an embodiment of the present invention. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0054] In Example 1, Figure 1 The implementation flow of the cloud-based financial risk management method provided by an embodiment of the present invention is illustrated below:

[0055] S100: Identify risk subjects that require financial risk management, select data source platforms, capture behavioral data containing risk subjects, query a pre-built key dictionary, wherein the key dictionary consists of keywords and corresponding sentiment scores, calculate the sentiment scores of the behavioral data, superimpose all sentiment scores to generate a total score, wherein each risk subject corresponds to a total score, count the number of behavioral data, and calculate the average score.

[0056] Identify the risk entities requiring financial risk management. For example, if a user wants to cooperate with Company A on raw material supply, they need to fully understand Company A's operational status and personnel changes before making the purchase. In this process, Company A is the risk entity. Select data source platforms that may involve the risk entity. These platforms can be social media platforms, news platforms, forums, and public information disclosure platforms. Use data scraping technology to extract behavioral data containing information related to the risk entity from all data source platforms. This behavioral data includes text comments, speech records, news reports, and interactive information.

[0057] A key dictionary is created, consisting of keywords and corresponding sentiment scores. The sentiment score characterizes the risk tendency corresponding to the keyword. For example, negative keywords such as "loss," "layoffs," "default," and "tight cash flow" are assigned negative values, while positive keywords such as "profit growth," "expansion," "successful financing," and "increased orders" are assigned positive values. By segmenting the text content in the behavioral data and comparing it with the key dictionary, the sentiment score of each keyword in the behavioral data is determined. The sentiment scores of the behavioral data are then calculated by summing these scores, with each behavioral data point corresponding to a sentiment score. The total score is generated by summing all the sentiment scores corresponding to the risk subject. The number of captured behavioral data points is counted, and the quotient of the total score and the number of points is defined as the average score. A higher average score indicates that the overall operating condition of the risk subject is relatively stable and the potential financial risk is lower.

[0058] S200: Collect attribute data of risk subjects and cluster the data source platforms into industry-related platforms and broad monitoring platforms. When the average score deviates from the preset fluctuation range, extract the core features in the behavioral data, write them into the preset template, generate topics, and publish them to the industry-related platforms. Through the behavioral data, traverse the relationship accounts of the risk subjects, generate interaction invitations, and write them into the topics.

[0059] The system collects attribute data of risk entities, including their type, business registration information, supply chain data, and equity structure. Based on type, data source platforms are clustered into industry-related platforms and broad monitoring platforms. Industry-related platforms primarily refer to vertical platforms highly relevant to the risk entity's industry, used to obtain highly specialized and targeted industry data, such as industry forums and supply chain platforms. For example, for an internet company, its corresponding industry-related platforms include its own app, developer communities, and related interactive communities of sister companies. Broad monitoring platforms refer to social media platforms, news and information platforms, and public information release platforms, used to obtain comprehensive and timely operational data. The advantage of this approach is that by collecting and integrating data from these two types of platforms in a layered manner, it can improve the relevance of the data while ensuring the comprehensiveness of the information, thereby building a multi-dimensional data foundation and providing reliable support for subsequent risk assessment. Based on the historical fluctuations of the risk entity's average score, a fluctuation range is set for each risk entity. When the average score exceeds the fluctuation range, it indicates that the risk entity may be in an abnormal operational state. Core features are extracted from behavioral data. These core features refer to key information characterizing the current operational status of the risk entity. Core features include, but are not limited to, changes in operational status, public opinion keywords, abnormal behavior patterns, and related event tags. For example, a core feature might be: multiple instances of delayed payments and an increase in supplier complaints. These core features are then written into a pre-defined template to generate guiding topics, which are then published on relevant industry platforms. For example, a topic could be: "Has this risk entity recently experienced any payment delays? What has been the supplier cooperation experience like?"

[0060] The process involves identifying the related accounts of the risk entity. These accounts can be user accounts that have previously posted comments about the risk entity, or user accounts in the same industry as the risk entity. These user accounts share similarities or connections with the risk entity in terms of business scope, operating model, or industry category. In short, related accounts are user accounts that may have information about the risk entity's operations. Interaction invitations are then generated using these related accounts. These invitations should include: a description of the topic, guidance for participation, and a feedback entry point. The interaction invitation is then incorporated into the topic.

[0061] S300: Obtain feedback data under the topic, construct a word set composed of sensitive words, determine whether there are sensitive words in the feedback data, if so, summarize all feedback data containing sensitive words, generate risk management results, write communication reminder information, and send it to the preset terminal, if not, activate the pre-edited targeted monitoring mechanism and update the risk management results.

[0062] The system receives feedback data from related accounts regarding a topic, creating a term set consisting of several sensitive words. These sensitive words include keywords, but their coverage is broader than that of keywords. The term set is matched against the feedback data, and each piece of feedback data is iterated and checked to determine if it contains sensitive words. If so, all feedback data corresponding to sensitive words are integrated to generate a risk management result. The risk management result is proactively acquired operational information about the risk entity. Communication reminders are inserted into the risk management result, including risk warnings and suggested measures. For example, a communication reminder might state: "The risk entity may have multiple instances of overdue payments to suppliers; it is recommended to promptly communicate with relevant personnel of the risk entity to verify its true operational status." The risk management result is then distributed to preset terminals, which are the devices of users who intend to cooperate with the risk entity. If no sensitive words are found in the feedback data, a targeted monitoring mechanism is activated. This targeted monitoring mechanism refers to a specific method for continuously monitoring and analyzing all information published by related accounts. When new information is detected from a related account, this information is used to update the risk management result.

[0063] In Example 2, Figure 2 The first sub-process flowchart of the cloud-based financial risk management method provided by an embodiment of the present invention is shown. The steps of identifying the risk subjects that need financial risk management and selecting the data source platform are described in detail below:

[0064] S101: Define the publishers and disseminators of behavioral data from the data source platform and draw the dissemination chain.

[0065] In the data source platform, users who publish behavioral data are defined as publishers, and users who participate in behaviors such as commenting, forwarding, or liking the behavioral data are defined as disseminators, generating a propagation chain, which is the path of information dissemination in the behavioral data. For example, A posts a status update on a social media platform about their work experience and feelings at a risk entity. B agrees with A's post and forwards it. Subsequently, C likes B's forwarded content. In this process, A is the publisher, and B and C are the disseminators. The propagation path is: A-B-C.

[0066] S102: Set several filtering indicators to delete useless parts of behavioral data.

[0067] After identifying the behavioral data of the risk subjects, several screening indicators are set, including time range and behavior type. Based on the screening indicators, useless parts of the behavioral data are deleted. For example, if the time range of the screening indicator is January 1st to February 1st, then useless parts of the behavioral data outside this time period will be deleted.

[0068] In Example 3, Figure 2 The first sub-process flowchart of the cloud-based financial risk management method provided by an embodiment of the present invention is shown. The following details the steps of calculating the sentiment scores of behavioral data, superimposing all sentiment scores, and generating a total score:

[0069] S103: Update the total score according to the preset step size, and plot the score evolution curve with time as the horizontal axis and the total score at the corresponding time as the vertical axis.

[0070] The preset step size can be 1 day or multiple days. In other words, the total score is updated once a day to establish a correspondence between the total score and time. The score evolution curve is plotted with time as the horizontal axis and the total score at the corresponding time as the vertical axis. The score evolution curve is used to show the trend of the total score over time.

[0071] S104: Divide the score evolution curve into an ascending segment and a descending segment, identify the descending segment, and update the behavioral data.

[0072] The system identifies the rising and falling segments in the score evolution curve. When the total score is in the falling segment, the behavioral data is updated. Specifically, when the total score is in the falling segment, it indicates that the business situation of the risk subject may be abnormal, so the behavioral data is re-captured.

[0073] In Example 4, Figure 3 The second sub-process flowchart of the cloud-based financial risk management method provided in this embodiment of the invention is shown. The following details the step of extracting the core features from the behavioral data and writing them into a preset template:

[0074] S201: The keyword that appears most frequently in the statistical behavioral data is defined as the target word. The target word and the preset template are input into the language model, and the topic is output.

[0075] The most frequently occurring keywords in statistical behavioral data are identified and defined as target words. These target words, along with pre-defined template information, are then input into a language model to output topics. The language model refers to a deep learning-based natural language processing model. The advantages of this approach are: it ensures clear topic semantics, conforms to pre-designed formats or rules, and provides standardized and consistent output results, facilitating subsequent analysis and monitoring.

[0076] S202: Set the validity period of the topic, extract the expiration time, obtain the generation time of the risk management result, and align the expiration time and generation time.

[0077] Select two time endpoints, set the validity period of the topic, and terminate the topic after determining the risk management results. For example, generate and publish a topic by capturing behavioral data, calculating average scores, and extracting core features. After collecting a preset number of feedback data (the preset number can be determined based on the correlation between the feedback data and the risk subject; if the correlation is low, more feedback data should be selected), process the data, and generate the risk management results. At this point, the topic can be terminated because continuing to discuss the topic after completing risk monitoring is no longer valuable.

[0078] In Example 5, Figure 4 The diagram illustrates the third sub-process flowchart of the cloud-based financial risk management method provided in this embodiment of the invention. The steps of obtaining feedback data under a specific topic and constructing a word set composed of sensitive words are described in detail below:

[0079] S301: Insert abnormal words into the word set and establish a mapping between abnormal words and risk levels.

[0080] Add anomalous words to the existing keyword set. Compared to keywords, anomalous words have a stronger sentiment connotation. Anomalous words can include risk warnings such as production stoppage and business cessation, or information about equity changes. Each anomalous word corresponds to a risk level, which includes high, medium, and low.

[0081] S302: When abnormal words are detected in the feedback data, update the communication reminder information and insert a tag generated by the risk level.

[0082] When abnormal words are detected in the feedback data, the communication reminder information is updated immediately, a label containing the risk level is generated, and the label is inserted into the corresponding feedback data.

[0083] In Example 6, Figure 4 The diagram illustrates the third sub-process flowchart of the cloud-based financial risk management method provided in this embodiment of the invention. The steps for updating the risk management result are described in detail below:

[0084] S303: Configure the influencing factors of the preset frequency and dynamically adjust the preset frequency, wherein the influencing factors include at least: time period.

[0085] After generating risk management results, the results are updated at a preset frequency. The preset frequency is influenced by factors including time periods. In other words, different time periods correspond to different preset frequencies; for example, the preset frequency should be increased during peak trading hours.

[0086] S304: Set up a cloud computing node and connect it to the data source platform, then upload the key dictionary to the cloud computing node.

[0087] A cloud computing node is set up as a data processing center. Various data source platforms are connected to the cloud computing node through interfaces, and the key dictionary is uploaded to the cloud computing node for storage, so as to improve the key dictionary's ability to retrieve, match and analyze behavioral data.

[0088] Figure 5 The diagram illustrates the structural composition of a cloud-based financial risk management system 1 provided in an embodiment of the present invention. The cloud-based financial risk management system 1 includes:

[0089] The calculation module 11 is used to identify risk subjects that need to be managed for financial risks, select data source platforms, capture behavioral data containing risk subjects, query a pre-built key dictionary, wherein the key dictionary consists of keywords and corresponding sentiment scores, calculate the sentiment scores of behavioral data, superimpose all sentiment scores to generate a total score, wherein each risk subject corresponds to a total score, count the number of behavioral data, and calculate the average score.

[0090] The writing module 12 is used to collect attribute data of risk subjects and cluster the data source platforms into industry-related platforms and broad monitoring platforms. When the average score deviates from the preset fluctuation range, the core features in the behavioral data are extracted, written into the preset template, a topic is generated, and published to the industry-related platform. Through the behavioral data, the relationship accounts of the risk subjects are traversed, an interaction invitation is generated, and written into the topic.

[0091] The judgment module 13 is used to obtain feedback data under the topic, construct a word set composed of sensitive words, determine whether there are sensitive words in the feedback data, if so, summarize all feedback data containing sensitive words, generate risk management results, write communication reminder information, and send it to the preset terminal, if not, activate the pre-edited targeted monitoring mechanism and update the risk management results.

[0092] Figure 6 The diagram illustrates the composition of the computing module 11 in a cloud-based financial risk management system provided in an embodiment of the present invention. The computing module 11 includes:

[0093] Definition unit 111 is used to define the publishers and disseminators of behavioral data from the data source platform and draw the dissemination chain;

[0094] Deletion unit 112 is used to set several filtering indicators and delete useless parts of behavioral data;

[0095] The drawing unit 113 is used to update the total score according to a preset step size, and draw a score evolution curve with time as the horizontal axis and the total score at the corresponding time as the vertical axis.

[0096] The identification unit 114 is used to divide the score evolution curve into an ascending segment and a descending segment, identify the descending segment, and update the behavioral data.

[0097] Figure 7 This diagram illustrates the structural composition of the writing module 12 in a cloud-based financial risk management system provided in an embodiment of the present invention. The writing module 12 includes:

[0098] Statistical unit 121 is used to count the most frequently occurring keywords in the behavioral data and define them as target words. The target words and preset templates are input into the language model, and the topic is output.

[0099] Setting unit 122 is used to set the validity period of a topic, extract the expiration time, obtain the generation time of the risk management result, and align the expiration time and generation time.

[0100] Figure 8 The diagram shows the structural composition of the judgment module 13 in the cloud-based financial risk management system provided in an embodiment of the present invention. The judgment module 13 includes:

[0101] Insertion unit 131 is used to insert abnormal words into the word set and establish a mapping between abnormal words and risk levels;

[0102] The update unit 132 is used to update the communication reminder information and insert a tag generated by the risk level when abnormal words are detected in the feedback data.

[0103] Configuration unit 133 is used to configure influencing factors of a preset frequency and dynamically adjust the preset frequency, wherein the influencing factors include at least: time period;

[0104] Upload unit 134 is used to build a cloud computing node and connect to the data source platform to upload the key dictionary to the cloud computing node.

[0105] The calculation module 11 is mainly used to complete step S100, the writing module 12 is mainly used to complete step S200, and the judgment module 13 is mainly used to complete step S300.

[0106] The definition unit 111 is mainly used to complete step S101, the deletion unit 112 is mainly used to complete step S102, the drawing unit 113 is mainly used to complete step S103, and the recognition unit 114 is mainly used to complete step S104.

[0107] The statistics unit 121 is mainly used to complete step S201, and the setting unit 122 is mainly used to complete step S202.

[0108] The insertion unit 131 is mainly used to complete step S301, the update unit 132 is mainly used to complete step S302, the configuration unit 133 is mainly used to complete step S303, and the upload unit 134 is mainly used to complete step S304.

[0109] The technical features of the above embodiments can be combined in any way. For the sake of brevity, 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.

[0110] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

[0111] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A cloud computing-based financial risk management method, characterized in that, The method includes: Identify risk subjects that require financial risk management, select data source platforms, capture behavioral data containing risk subjects, query a pre-built key dictionary, which consists of keywords and corresponding sentiment scores, calculate the sentiment scores of behavioral data, and aggregate all sentiment scores to generate a total score, where each risk subject corresponds to a total score, count the number of behavioral data, and calculate the average score. Collect attribute data of risk subjects and cluster the data source platforms into industry-related platforms and broad monitoring platforms. When the average score deviates from the preset fluctuation range, extract the core features from the behavioral data, write them into a preset template, generate a topic, and publish it to the industry-related platforms. Through the behavioral data, traverse the relationship accounts of the risk subjects, generate interaction invitations, and write them into the topic. Obtain feedback data under the topic, construct a word set composed of sensitive words, determine whether there are sensitive words in the feedback data, if so, summarize all feedback data containing sensitive words, generate risk management results, write communication reminder information, and send it to preset terminals, if not, activate the pre-edited targeted monitoring mechanism and update the risk management results.

2. The cloud-based financial risk management method according to claim 1, characterized in that, The steps for identifying risk entities requiring financial risk management and selecting data source platforms include: From the data source platform, define the publishers and disseminators of behavioral data, and draw the dissemination chain; Set several filtering criteria to remove useless parts of the behavioral data.

3. The cloud-based financial risk management method according to claim 2, characterized in that, The steps for calculating the sentiment score of the behavioral data, summing all the sentiment scores, and generating the total score include: The total score is updated according to a preset step size, and a score evolution curve is plotted with time as the horizontal axis and the total score at the corresponding time as the vertical axis. The score evolution curve is divided into an ascending segment and a descending segment. The descending segment is identified, and the behavioral data is updated.

4. The cloud-based financial risk management method according to claim 3, characterized in that, The step of extracting the core features from the behavioral data and writing them into a preset template includes: The most frequently occurring keywords in the statistical behavioral data are defined as target words. The target words and preset templates are input into the language model, and the output is the topic. Set the validity period of the topic, extract the expiration time, obtain the generation time of the risk management result, and align the expiration time and generation time.

5. The cloud-based financial risk management method according to claim 1, characterized in that, The steps of obtaining feedback data under the topic and constructing a word set composed of sensitive words include: Insert abnormal words into the word set and establish a mapping between abnormal words and risk levels; When abnormal words are detected in the feedback data, the communication reminder information is updated and a tag generated by the risk level is inserted.

6. The cloud-based financial risk management method according to claim 1, characterized in that, The steps for updating the risk management results include: Configure influencing factors for a preset frequency and dynamically adjust the preset frequency, wherein the influencing factors include at least: time period; Set up a cloud computing node and connect it to the data source platform to upload the key dictionary to the cloud computing node.

7. A cloud-based financial risk management system, characterized in that, The system includes: The calculation module is used to identify risk subjects that need financial risk management, select data source platforms, capture behavioral data containing risk subjects, query a pre-built key dictionary, which consists of keywords and corresponding sentiment scores, calculate the sentiment scores of behavioral data, and superimpose all sentiment scores to generate a total score, where each risk subject corresponds to a total score, count the number of behavioral data, and calculate the average score. The writing module is used to collect attribute data of risk subjects and cluster the data source platforms into industry-related platforms and broad monitoring platforms. When the average score deviates from the preset fluctuation range, the core features in the behavioral data are extracted, written into the preset template, a topic is generated, and published to the industry-related platform. Through the behavioral data, the relationship accounts of the risk subjects are traversed, an interaction invitation is generated, and written into the topic. The judgment module is used to obtain feedback data under the topic, construct a word set composed of sensitive words, determine whether there are sensitive words in the feedback data, if so, summarize all feedback data containing sensitive words, generate risk management results, write communication reminder information, and send it to preset terminals, if not, activate the pre-edited targeted monitoring mechanism and update the risk management results.

8. The cloud-based financial risk management system according to claim 7, characterized in that, The computing module includes: The definition unit is used to define the publishers and disseminators of behavioral data from the data source platform and to draw the dissemination chain; The delete unit is used to set several filtering criteria to delete useless parts of behavioral data; The drawing unit is used to update the total score according to a preset step size, and draw the score evolution curve with time as the horizontal axis and the total score at the corresponding time as the vertical axis. The identification unit is used to divide the score evolution curve into an ascending segment and a descending segment, identify the descending segment, and update the behavioral data.

9. The cloud-based financial risk management system according to claim 8, characterized in that, The writing module includes: The statistical unit is used to analyze the most frequently occurring keywords in the statistical behavioral data and defines them as target words. The target words and preset templates are input into the language model, and the output is the topic. The setting unit is used to set the validity period of a topic, extract the expiration time, obtain the generation time of the risk management result, and align the expiration time and generation time.

10. The cloud-based financial risk management system according to claim 7, characterized in that, The judgment module includes: An insertion unit is used to insert abnormal words into the word set and establish a mapping between abnormal words and risk levels; An update unit is used to update the communication reminder information and insert a tag generated by the risk level when abnormal words are detected in the feedback data. A configuration unit is used to configure influencing factors of a preset frequency and dynamically adjust the preset frequency, wherein the influencing factors include at least: time periods; The upload unit is used to build a cloud computing node and connect to the data source platform to upload the key dictionary to the cloud computing node.

Images