AI-driven bank asset allocation intelligent optimization decision method and system

Abstract

The present application relates to the technical field of financial asset management, in particular to an AI-driven bank asset allocation intelligent optimization decision method and system. Obtain the public opinion risk index of the target enterprise; call the dispute information of the target enterprise, analyze the self dispute risk index of the target enterprise based on the pre-trained dispute information analyzer; call multiple dispute risk indexes of multiple cooperative enterprises of the target enterprise, and calculate the business relationship risk index of the target enterprise; obtain the investment rationality index and the industry prospect index of the target enterprise, and calculate the enterprise prospect index of the target enterprise based on the investment rationality index and the industry prospect index; the weighted sum of the multiple risk indexes of the target enterprise is obtained, and the comprehensive investment risk index of the target enterprise is obtained, which is used for asset allocation recommendation. The present application significantly improves the comprehensiveness, accuracy and forward-looking of the bank's enterprise investment risk assessment.

Description

Technical Field

[0001] This invention relates to the field of financial asset management technology, and in particular to an AI-driven intelligent optimization decision-making method and system for bank asset allocation. Background Technology

[0002] Currently, bank asset allocation decisions largely rely on traditional risk assessment methods, using corporate financial statement data as the core to analyze debt repayment and profitability, supplemented by industry macro data and market experience. However, this model has significant shortcomings: the risk assessment dimensions are too singular, and it relies too heavily on financial data, ignoring key dimensions such as public opinion and commercial disputes, making it difficult to identify relevant potential risks. Furthermore, existing asset allocation decisions often rely on manual analysis, which is inefficient in processing massive amounts of information, highly subjective, lacks intelligent data support, and is not forward-looking enough to provide accurate and forward-looking guidance for asset allocation, easily leading to high risks in asset allocation decisions. Summary of the Invention

[0003] This invention addresses the technical problems of existing technologies that rely on a single dimension for risk assessment and excessive financial data by providing an AI-driven intelligent optimization decision-making method and system for bank asset allocation.

[0004] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: In a first aspect, this invention provides an AI-driven intelligent optimization decision-making method for bank asset allocation, comprising: analyzing the public opinion status of a target enterprise across the entire network within a preset time range to obtain the public opinion risk index of the target enterprise; retrieving dispute information of the target enterprise and analyzing it based on a pre-trained dispute information analyzer to obtain the target enterprise's own dispute risk index; retrieving dispute information and performance information of multiple partner enterprises of the target enterprise, calculating the historical default rate of multiple partner enterprises based on the performance information, obtaining multiple dispute risk indices of multiple partner enterprises based on the dispute information and dispute information analyzer of multiple partner enterprises, and calculating the business relationship risk index of the target enterprise by combining the multiple cooperation scales between the target enterprise and multiple partner enterprises, as well as the historical default rate and dispute risk index of multiple partner enterprises; obtaining the investment rationality index of the target enterprise based on the investment behavior of the target enterprise, obtaining the industry prospect index of the target enterprise based on the core business of the target enterprise, and calculating the enterprise prospect index of the target enterprise based on the investment rationality index and the industry prospect index; weighting and summing the public opinion risk index, the target enterprise's own dispute risk index, the business relationship risk index, and the enterprise prospect index to obtain the comprehensive investment risk index of the target enterprise, and continuing to calculate multiple comprehensive investment risk indices of multiple target enterprises for asset allocation recommendation.

[0005] Optionally, analyze the public opinion status of the target company across the entire network within a preset time range to obtain the public opinion risk index of the target company. This includes: using web crawling technology to capture public opinion information about the target company across the entire network within a preset time range and adding it to the public opinion information set; using sentiment analysis algorithms to perform sentiment analysis on each piece of public opinion information and label it with sentiment tags, including negative, positive, and neutral; and calculating the proportion of public opinion information with negative sentiment tags to the total number of public opinion information in the public opinion information set as the public opinion risk index.

[0006] Optionally, the dispute information of the target enterprise is retrieved, and the dispute risk index of the target enterprise is obtained by analyzing it based on a pre-trained dispute information analyzer. This includes: extracting multiple historical defaulting enterprises from the bank's non-performing asset database; using the first default time of each historical defaulting enterprise as a node, collecting dispute information of multiple historical defaulting enterprises within a preset period before the node as a historical dispute information set, and labeling multiple default severity levels of multiple historical defaulting enterprises corresponding to multiple historical dispute information as a historical default information set, wherein the default severity level includes secondary, doubtful, and loss; building the dispute information analyzer based on machine learning, and training the dispute information analyzer using the historical dispute information set and the historical default information set until convergence; inputting the dispute information of the target enterprise into the dispute information analyzer, thereby predicting and outputting the dispute risk index of the target enterprise.

[0007] Optionally, based on performance information, the historical default rates of multiple cooperating enterprises are calculated. Based on dispute information of multiple cooperating enterprises and a dispute information analyzer, multiple dispute risk indices of multiple cooperating enterprises are obtained, including: identifying multiple enterprises with commercial contracts with the target enterprise as the multiple cooperating enterprises; obtaining the performance of multiple cooperating enterprises in fulfilling the contractual obligations within a preset period as performance information; using the ratio of the amount of non-performance to the amount of performance due within the preset period as the historical default rate of the cooperating enterprise, and continuing to calculate multiple historical default rates of multiple cooperating enterprises; retrieving multiple dispute information of multiple cooperating enterprises within the preset period, inputting the multiple dispute information into the dispute information analyzer, and obtaining multiple dispute risk indices of multiple cooperating enterprises.

[0008] Optionally, by combining the target company's multiple cooperation scales with multiple partners, as well as the historical default rates and dispute risk indices of the multiple partners, a business relationship risk index of the target company can be calculated. This includes: calculating the accounts receivable of the target company with each of the multiple partners as the cooperation scale; weighting and fusing the historical default rates and dispute risk coefficients of the multiple partners as multiple business risk coefficients of the multiple partners; and weighting and fusing the multiple business risk coefficients of the multiple partners using the proportion of the cooperation scale of the multiple partners to the target company's total accounts receivable as the weight to obtain the business relationship risk index of the target company.

[0009] Optionally, based on the target company's investment behavior, an investment rationality index is obtained; based on the target company's core business, an industry prospect index is obtained; and based on the investment rationality index and the industry prospect index, a corporate prospect index is calculated, including: obtaining the target company's investment areas and, based on preset standards, obtaining the financing hotspots in the target company's industry; calculating the proportion of the target company's investment in the financing hotspots to its total investment, as the target company's investment rationality index; identifying several listed companies with the same core business as the target company, obtaining the average stock price growth rate of the several listed companies within a preset time range, as the industry prospect index; and multiplying the target company's investment rationality index by the industry prospect index to obtain the target company's corporate prospect index.

[0010] Optionally, the public opinion risk index, self-dispute risk index, business relationship risk index, and enterprise prospect index of the target enterprise are weighted and summed to obtain the comprehensive investment risk index of the target enterprise. Multiple comprehensive investment risk indices of multiple target enterprises are then calculated for asset allocation recommendations. This includes: obtaining multiple target enterprises to be invested in; calculating the public opinion risk index, self-dispute risk index, business relationship risk index, and enterprise prospect index of each target enterprise separately, and summing them according to a preset weighting rule to obtain the comprehensive investment risk index of each target enterprise; and listing the multiple target enterprises in ascending order of comprehensive investment risk index into an asset allocation recommendation list.

[0011] Secondly, this invention provides an AI-driven intelligent optimization decision-making system for bank asset allocation, comprising: The public opinion risk analysis module is used to analyze the public opinion status of a target company across the entire network within a preset time range and obtain the public opinion risk index of the target company. The self-dispute risk analysis module is used to retrieve the dispute information of the target enterprise and obtain the self-dispute risk index of the target enterprise based on the pre-trained dispute information analyzer. The business relationship risk analysis module is used to retrieve dispute and performance information of multiple partner companies of the target company. Based on the performance information, it calculates the historical default rate of multiple partner companies. Based on the dispute information of multiple partner companies and the dispute information analyzer, it obtains multiple dispute risk indices of multiple partner companies. Combining the multiple cooperation scales between the target company and multiple partner companies, as well as the historical default rate and dispute risk index of multiple partner companies, it calculates the business relationship risk index of the target company. The enterprise prospect analysis module is used to obtain the investment rationality index of the target enterprise based on its investment behavior, obtain the industry prospect index of the target enterprise based on its core business, and calculate the enterprise prospect index of the target enterprise based on the investment rationality index and the industry prospect index. The comprehensive investment risk assessment module is used to weight and sum the public opinion risk index, self-dispute risk index, business relationship risk index, and enterprise prospect index of the target enterprise to obtain the comprehensive investment risk index of the target enterprise. It then calculates multiple comprehensive investment risk indices for multiple target enterprises for asset allocation recommendations.

[0012] By implementing this invention, it is possible to analyze the public opinion status of a target company across the entire network within a preset time range, obtain the public opinion risk index of the target company, and transform vague public opinion information into measurable risk indicators by quantifying the proportion of negative public opinion. This provides banks with intuitive and objective data support for assessing corporate investment risks and avoids risk misjudgment caused by subjective judgment. By implementing this invention, it is possible to retrieve the dispute information of the target enterprise and analyze it based on a pre-trained dispute information analyzer to obtain the target enterprise's own dispute risk index. This provides an effective way for banks to gain a deeper understanding of the risk status of the target enterprise's own operation process, and helps banks to discover potential signs of default in advance and reduce credit risk. By implementing this invention, it is possible to retrieve dispute and performance information of multiple partner companies of a target enterprise; calculate the historical default rate of multiple partner companies based on the performance information; obtain multiple dispute risk indices of multiple partner companies based on the dispute information and dispute information analyzer of multiple partner companies; and calculate the business relationship risk index of the target enterprise by combining the multiple cooperation scales between the target enterprise and multiple partner companies, as well as the historical default rates and dispute risk indices of multiple partner companies. By combining key factors such as cooperation scale, business relationship risk can be accurately quantified, providing a strong basis for banks to judge the degree of influence of partners on the target enterprise, and further reducing investment decision risk. By implementing this invention, it is possible to obtain the investment rationality index of a target company based on its investment behavior, obtain the industry prospect index of the target company based on its core business, and calculate the enterprise prospect index of the target company based on the investment rationality index and the industry prospect index. This comprehensively evaluates the enterprise's prospects from the perspectives of investment behavior and industry prospects, not only focusing on the rationality of the enterprise's own investment decisions but also combining industry development trends, making the enterprise prospect assessment more comprehensive and objective. This provides banks with clear and quantifiable indicators to judge the future development potential of target companies, helps banks to more accurately grasp the investment value of enterprises, and provides forward-looking support for asset allocation decisions. By implementing this invention, a weighted sum of the target company's public opinion risk index, self-dispute risk index, business relationship risk index, and corporate prospect index can be obtained to obtain the target company's comprehensive investment risk index. Multiple comprehensive investment risk indices for multiple target companies can then be calculated for asset allocation recommendations. Based on the comprehensive investment risk index, an asset allocation recommendation list can be generated, providing banks with clear and feasible investment decision-making references, helping banks optimize their asset allocation structure, and improve the safety and profitability of asset allocation.

[0013] In summary, by implementing this invention, banks can significantly improve the comprehensiveness, accuracy, and foresight of their corporate investment risk assessments, providing intelligent and data-driven support for bank asset allocation decisions, effectively reducing investment risks, optimizing asset allocation structures, and improving the efficiency and profitability of bank asset operations. Attached Figure Description

[0014] Figure 1 A flowchart illustrating an AI-driven intelligent optimization decision-making method for bank asset allocation provided by this invention; Figure 2 This is a schematic diagram of the structure of an AI-driven intelligent optimization decision-making system for bank asset allocation provided by the present invention.

[0015] In the attached diagram, the components represented by each number are as follows: Module 11: Public Opinion Risk Analysis; Module 12: Self-Dispute Risk Analysis; Module 13: Business Relationship Risk Analysis; Module 14: Enterprise Prospect Analysis; Module 15: Comprehensive Investment Risk Assessment. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0018] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0019] Example 1, as Figure 1 As shown, this embodiment of the invention provides an AI-driven intelligent optimization decision-making method for bank asset allocation, including: S100: Analyze the public opinion status of the target company across the entire network within a preset time range to obtain the public opinion risk index of the target company; S200: Retrieve dispute information of the target company and analyze it based on a pre-trained dispute information analyzer to obtain the target company's own dispute risk index; S300: Retrieve dispute and performance information of multiple partner companies of the target company. Based on the performance information, calculate the historical default rate of multiple partner companies. Based on the dispute information of multiple partner companies and the dispute information analyzer, obtain multiple dispute risk indices of multiple partner companies. Combine the multiple cooperation scales between the target company and multiple partner companies, as well as the historical default rate and dispute risk index of multiple partner companies, calculate the business relationship risk index of the target company. S400: Based on the investment behavior of the target company, obtain the investment rationality index of the target company; based on the core business of the target company, obtain the industry prospect index of the target company; and based on the investment rationality index and the industry prospect index, calculate the enterprise prospect index of the target company. S500: The target company's public opinion risk index, self-dispute risk index, business relationship risk index, and enterprise prospect index are weighted and summed to obtain the target company's comprehensive investment risk index. Multiple comprehensive investment risk indices for multiple target companies are then calculated for asset allocation recommendations.

[0020] In step S100 of this application embodiment, the public opinion status of the target enterprise across the entire network within a preset time range is analyzed to obtain the public opinion risk index of the target enterprise, including: Based on web crawling technology, capture public opinion information about the target company from the entire network within a preset time range and add it to the public opinion information set; Based on sentiment analysis algorithms, each piece of public opinion information is analyzed for public opinion sentiment and then labeled with public opinion sentiment tags, which include negative, positive, and neutral. The proportion of public opinion information with a negative sentiment label is calculated as the total number of public opinion information in the set, and is used as the public opinion risk index.

[0021] In this embodiment of the application, the purpose of step S100 is to transform the vague and scattered public opinion information of the target enterprise into a quantifiable and comparable "public opinion risk index," providing banks with an objective and intuitive basis for assessing the investment risk of the target enterprise, and filling the gap in the assessment of non-financial risk factors such as the enterprise's market reputation and public trust in traditional asset allocation decisions.

[0022] To achieve the above objectives, firstly, it is necessary to use web crawling technology to capture public opinion information about the target company from the entire network within a preset time range.

[0023] Specifically, it is necessary to first clarify the two core screening criteria: "preset time range" and "target company". The preset time range, such as the past 3 months or half a year, uses web crawling technology to automatically capture all public opinion information related to the target company from the entire network, including news platforms, social media, industry forums, and financial websites. Then, this information is uniformly summarized to form a "public opinion information collection" to ensure that the information coverage is comprehensive and avoids missing key public opinion content.

[0024] Next, based on sentiment analysis algorithms, each piece of public opinion information needs to be analyzed for public sentiment and then labeled with sentiment tags. The sentiment analysis algorithm is a machine learning-based text classification algorithm, such as BERT or SVM.

[0025] Specifically, each piece of public opinion information in the "public opinion information set," such as news reports, user comments, and forum posts, needs to be input into a sentiment analysis algorithm. The algorithm will analyze the sentiment tendency of the text content and determine whether the information is biased towards negative, positive, or neutral. Negative information includes criticism, complaints, and negative rumors, while positive information can include praise and good news. Neutral information can include objective statements of facts and real-time news that does not have commercial value.

[0026] Each piece of public opinion information is labeled with a corresponding "public opinion sentiment tag," completing the transformation from "text information" to "emotional classification."

[0027] Furthermore, it is necessary to calculate the proportion of public opinion information with a negative sentiment label to the total number of public opinion information in the set, as a public opinion risk index.

[0028] Specifically, the ratio of the total number of public opinion messages in the "public opinion information set" to the number of messages marked "negative" is calculated as follows: Public Opinion Risk Index = Number of Negative Public Opinion Messages / Total Number of Public Opinion Messages × 100%. The "Negative Public Opinion Ratio" calculated above is directly defined as the "Public Opinion Risk Index." A higher index value indicates a higher public opinion risk for the target company; a lower value indicates a lower risk. This quantifies public opinion risk and facilitates subsequent comparison and comprehensive calculation with other risk indices.

[0029] In step S200 of this application embodiment, the dispute information of the target enterprise is retrieved, and the target enterprise's own dispute risk index is obtained based on the analysis of a pre-trained dispute information analyzer, including: Extract multiple historical defaulting companies from the bank's non-performing asset database; Using the first default time of each historical defaulting enterprise as a node, dispute information of multiple historical defaulting enterprises within a preset period before the node is collected as a historical dispute information set, and the severity of multiple defaults of multiple historical defaulting enterprises corresponding to multiple historical dispute information is marked as a historical default information set, wherein the severity of default includes secondary, suspicious and loss; The dispute information analyzer is built based on machine learning, and the dispute information analyzer is trained using the historical dispute information set and the historical breach of contract information set until convergence. By inputting the dispute information of the target company into the dispute information analyzer, the dispute risk index of the target company can be predicted and output.

[0030] In this embodiment of the application, the purpose of step S200 is to train an intelligent analysis model using historical defaulting enterprise data, accurately quantify the investment risks hidden by the target enterprise due to its own commercial disputes, generate an "individual dispute risk index", and fill the gap in the traditional asset allocation decision-making process regarding the insufficient assessment of enterprise dispute risks.

[0031] To achieve the above objectives, it is first necessary to extract multiple historical defaulting companies from the bank's non-performing asset database to ensure that the default data is highly consistent with the default risk scenarios actually faced by the bank.

[0032] Next, using "historically defaulting enterprises" as the main sample, and based on the key node of "first default time"—collecting enterprise dispute information within the "pre-set period" prior to this node—it is necessary to form a "historical dispute information set." This dispute information includes contract disputes, litigation records, arbitration cases, etc. Simultaneously, according to the bank's non-performing asset classification standards, the "severity of default" corresponding to these historically defaulting enterprises is marked, i.e., substandard, doubtful, or loss-making, thus forming the "historical default information set."

[0033] For example, suppose a bank plans to assess the dispute risk of Company A, which mainly manufactures electronic products. Company A is the target company.

[0034] Then, it is necessary to select 50 "electronic product manufacturing companies" that have defaulted in the past from the bank's non-performing asset database, such as Company B, which had defaulted on loans due to contract disputes, and Company C, whose capital chain was broken due to patent litigation.

[0035] Next, using the "first default time" of each historically defaulting company as a node, dispute information for a preset period prior to the node is collected, such as 12 months. For example, if Company B's first default time is June 2022, then three contract dispute records from June 2021 to June 2022 are collected to form one sample in the "historical dispute information set." Simultaneously, based on Company B's asset losses after the default, its default severity is marked as "suspicious." Similarly, dispute information and default severity labels for the same period from Company C and 49 other companies are collected, ultimately forming 50 sets of paired training data containing "dispute information - default severity." This data is used to train the dispute information analyzer.

[0036] Furthermore, it is necessary to build the dispute information analyzer based on machine learning, and train the dispute information analyzer using the historical dispute information set and the historical default information set until convergence.

[0037] Depending on the task type of the dispute information analyzer, the random forest classification algorithm can be used to build the dispute information analyzer.

[0038] In the parameter settings of the dispute information analyzer, the number of decision trees is set to 100; the maximum depth of each decision tree is set to 8 layers; the minimum number of samples for split nodes is set to 5; and the minimum number of samples for leaf nodes is set to 2.

[0039] The training samples for the dispute information analyzer are the aforementioned paired training data of dispute information and the severity of breach of contract. At least 100 sets of paired training data need to be collected, with 20% randomly allocated as the validation set and 80% as the training set for training the dispute information analyzer. The training rounds are set to 50 rounds, with each round using all 100 sets of samples to perform a complete training of the random forest model. The convergence criterion is set to ensure that after five consecutive rounds of training, the model's prediction accuracy on the validation set does not fluctuate by more than 1%.

[0040] The dispute information analyzer is trained using the above method until it converges. By inputting the dispute information of the target company into the analyzer, the dispute risk index of the target company can be predicted and output.

[0041] For example, if dispute information, including contract disputes and patent litigation of Company D in the past 12 months, is input into the dispute information analyzer, the dispute information analyzer will extract key features and compare historical data, and finally output Company D's own dispute risk index as 82%, with a full score of 100%. The higher the value, the higher the risk.

[0042] In step S300 of this application embodiment, based on performance information, the historical default rates of multiple cooperating enterprises are calculated and obtained. Based on the dispute information of multiple cooperating enterprises and a dispute information analyzer, multiple dispute risk indices of multiple cooperating enterprises are obtained, including: Identify multiple companies that have commercial contracts with the target company, as the multiple cooperating companies; The system obtains information on the performance of contracts by multiple partner companies within a preset period. This information is used as performance data. The ratio of the amount of contractual non-performance to the amount of contractual performance due by a partner company within the preset period is used as the historical default rate of the partner company. This process is then used to calculate and obtain multiple historical default rates for the multiple partner companies. Multiple dispute information from multiple partner companies within a preset period is retrieved, and the multiple dispute information is input into the dispute information analyzer to obtain multiple dispute risk indices for the multiple partner companies.

[0043] In step S300 of this application embodiment, the purpose of the above steps is to deconstruct risks from the perspective of "business relationship chain" and provide key basic data for the subsequent calculation of the target enterprise's "business relationship risk index" by quantifying the historical default probability of the target enterprise's partners and its own dispute risks.

[0044] To achieve the above objectives, it is first necessary to identify multiple companies with commercial contracts with the target company as the aforementioned multiple partners. For example, if the target company is an automobile manufacturer, its partners may include upstream engine suppliers, downstream car dealers, etc. The cooperation relationship must be confirmed by verifying the commercial contracts such as purchase contracts and sales contracts signed by both parties to ensure the authenticity and relevance of the cooperation relationship and avoid including companies with no substantial cooperation in the evaluation scope, which could lead to misjudgment of risk.

[0045] Next, it is necessary to obtain the performance information of multiple cooperating companies in fulfilling the contractual obligations within a preset period. The ratio of the amount of non-performance to the amount of performance due within the preset period is used as the historical default rate of the cooperating company. This process is repeated to obtain multiple historical default rates for the multiple cooperating companies.

[0046] The specific duration of the preset period mentioned here can be set according to industry characteristics and cooperation cycle, such as the past two years. "Performance information" refers to the actual situation of whether the cooperating company has fulfilled its obligations as agreed in the contract between the two parties, including the performance amount and performance time stipulated in the contract, as well as the amount of non-performance, the reasons for non-performance, etc.

[0047] The historical default rate of a partner company is calculated as follows: Historical default rate = (Amount of non-performance by the company within a preset period) ÷ (Amount of performance due by the company within a preset period) × 100%. For example, if an upstream supplier has a total contract value of 10 million yuan to fulfill to the target company within two years, and the actual non-performance amount is 500,000 yuan, then the supplier's historical default rate can be calculated as 5% using the above formula.

[0048] Furthermore, it is necessary to retrieve multiple dispute information from multiple cooperating enterprises during a preset period, input the multiple dispute information into the dispute information analyzer, and obtain multiple dispute risk indices for multiple cooperating enterprises.

[0049] This involves collecting "multiple dispute information" from the aforementioned partner companies within a "preset period." This preset period is consistent with the preset period used to calculate the historical default rate, ensuring data timeliness. The dispute information includes contract disputes, litigation cases, arbitration records, etc., involving the partner companies, which is consistent with the type of "dispute information" in step S200.

[0050] Then, the dispute information analyzer trained to convergence in step S200 is directly reused, and the dispute information of each partner company is input into the analyzer. Based on the learned correlation logic between historical disputes and default risks, the dispute information analyzer will automatically output the "dispute risk index" for the corresponding partner company. For example, if a distributor has multiple payment disputes, the dispute information analyzer will output its dispute risk index as 70%. The higher the risk index value, the more serious the dispute risk.

[0051] In step S300 of this application embodiment, the business relationship risk index of the target enterprise is calculated by combining the multiple cooperation scales between the target enterprise and multiple cooperating enterprises, as well as the historical default rates and dispute risk indices of the multiple cooperating enterprises, including: Calculate the accounts receivable of the target company and multiple partner companies separately, and use this as the scale of cooperation with multiple partner companies; The historical default rates and dispute risk coefficients of multiple partner companies are weighted and integrated to form multiple business risk coefficients for the multiple partner companies; Using the proportion of the cooperation scale of multiple cooperating enterprises to the total accounts receivable of the target enterprise as the weight, the multiple business risk coefficients of the multiple cooperating enterprises are weighted and integrated to obtain the business relationship risk index of the target enterprise.

[0052] In this embodiment of the application, the purpose of the above steps in step S300 is to associate the risk of the cooperating enterprise with the scale of cooperation of the target enterprise, and to quantify the overall risk faced by the target enterprise due to the business cooperation relationship through multi-level weighted fusion, thereby generating a "business relationship risk index" to solve the problem that traditional assessments only consider the risk of the cooperating party in isolation and ignore the difference in the impact of the degree of cooperation on the target enterprise.

[0053] To achieve the above objectives, it is first necessary to calculate the accounts receivable of the target company from each of its multiple partner companies, which will be considered as the scale of cooperation with these partners. For example, in the annual cooperation between the target company and distributor A, distributor A has outstanding payments of 8 million yuan, and in the cooperation with distributor B, distributor B has outstanding payments of 5 million yuan. Therefore, the scale of cooperation between the target company and A and B is 8 million yuan and 5 million yuan, respectively.

[0054] Next, the historical default rates and dispute risk coefficients of multiple partner companies need to be weighted and integrated to form multiple business risk coefficients for each company. The weighting rules can be set according to the bank's risk appetite, such as a 60% weight for historical default rate and a 40% weight for dispute risk index. The two indicators for a single partner company are then weighted and summed to obtain that company's "business risk coefficient." For example, if partner company A has a historical default rate of 5% (quantified as 0.05) and a dispute risk index of 0.6, its business risk coefficient = 0.05 × 60% + 0.6 × 40% = 0.03 + 0.24 = 0.27; and partner company B has a historical default rate of 3% (0.03) and a dispute risk index of 0.4, its business risk coefficient = 0.03 × 60% + 0.4 × 40% = 0.018 + 0.16 = 0.178.

[0055] Furthermore, it is necessary to use the proportion of the cooperation scale of multiple cooperating enterprises to the total accounts receivable of the target enterprise as a weight to weight and integrate the multiple business risk coefficients of multiple cooperating enterprises to obtain the business relationship risk index of the target enterprise.

[0056] Specifically, the first step is to calculate the target company's "total accounts receivable," which is the sum of accounts receivable from all partner companies. For example, in the above scenario, total accounts receivable = 8 million yuan in accounts receivable from company A + 5 million yuan in accounts receivable from company B = 13 million yuan. Next, calculate the "partnership scale percentage" for each partner company, which is the proportion of that company's accounts receivable to the total accounts receivable, used as the weight for weighted merging. For instance, the scale percentage of partner company A = 8 million ÷ 13 million ≈ 61.5%, and the scale percentage of partner company B = 5 million ÷ 13 million ≈ 38.5%.

[0057] Next, the "business risk coefficients" of all cooperating companies need to be weighted and summed using the "proportion of cooperation scale" as the weight. The result is the business relationship risk index of the target company. Continuing with the previous example, the business relationship risk index of the target company = 0.27 × 61.5% + 0.178 × 38.5% ≈ 0.166 + 0.069 ≈ 0.235 (23.5%).

[0058] In step S400 of this application embodiment, based on the investment behavior of the target enterprise, an investment rationality index of the target enterprise is obtained; based on the core business of the target enterprise, an industry prospect index of the target enterprise is obtained; and based on the investment rationality index and the industry prospect index, an enterprise prospect index of the target enterprise is calculated, including: Identify the target company's investment areas and, based on preset criteria, identify the hottest financing areas within the target company's industry. The proportion of the target company's investment in the financing hotspots to its total investment is calculated as the target company's investment rationality index. Identify several listed companies whose core businesses are the same as the target company, and obtain the average stock price growth rate of the listed companies within a preset time range as an industry prospect index. The target company's investment rationality index is obtained by multiplying its industry prospect index by the target company's investment rationality index.

[0059] In this embodiment of the application, the purpose of step S400 is to quantitatively assess the future development potential of the target company from the dual dimensions of investment decision rationality and industry development trend, and generate a "company prospect index" to fill the gap in traditional asset allocation decision-making, which lacks a systematic quantitative method for assessing company prospects and relies too much on subjective experience.

[0060] To achieve the above objectives, it is first necessary to identify the target company's investment areas and, based on preset criteria, identify the financing hotspots in the target company's industry.

[0061] First, retrieve the target company's investment records to identify its "investment areas." For example, if the target company's main business is new energy vehicles, its investment areas might include battery R&D, autonomous driving technology, and charging pile construction. Then, based on pre-defined criteria such as industry development trends, growth reports released by industry associations, and fund flows monitored by financial institutions, determine the "hot areas for financing" in the target company's industry. For instance, through the aforementioned information channels, it can be determined that the hot areas for financing in the new energy vehicle industry in 2025 might be solid-state batteries and smart cockpits.

[0062] Next, it is necessary to calculate the proportion of the target company's investment in the hottest financing areas to its total investment, which serves as the target company's investment rationality index. The specific calculation method is: Investment Rationality Index = (Target Company's Investment in Hot Financing Areas) ÷ (Target Company's Total Investment) × 100%. For example, if the target company's total annual investment is 1 billion yuan, of which 600 million yuan is invested in hot areas such as solid-state batteries, then its investment rationality index = 6 ÷ 10 × 100% = 60%. The higher the investment rationality index, the higher the match between the company's investment decisions and industry opportunities, and the more rational its investment behavior.

[0063] Next, it is necessary to identify several listed companies with the same core business as the target company, and obtain the average stock price growth rate of these listed companies within a preset time range, which will serve as the industry prospect index. For example, if the target company's main business is photovoltaic module production, then 10 listed companies in the A-share market that also primarily engage in photovoltaic modules can be selected. The stock price data of these listed companies within the "preset time range," such as the past year, is obtained, and the stock price growth rate of each company is calculated. The calculation method for the stock price growth rate is ((end-of-period stock price - beginning-of-period stock price) ÷ beginning-of-period stock price × 100%). Then, the arithmetic mean of the stock price growth rates of all companies is taken as the "industry prospect index." For example, if the average stock price growth rate of the 10 listed photovoltaic module companies over the past year is 25%, then the industry prospect index for the target company's industry is 25%. This industry prospect index directly reflects the overall market recognition and development potential of the industry; the higher the value, the broader the industry's prospects.

[0064] Finally, the investment rationality index of the target company needs to be multiplied by the industry prospect index to obtain the company prospect index.

[0065] The two indices obtained in the first two steps are multiplied together, and the result is the target company's "Company Prospect Index". Using the previous example, if the target company's investment rationality index is 60% (0.6) and its industry prospect index is 25% (0.25), then its company prospect index = 0.6 × 0.25 = 0.15 (15%).

[0066] In step S500 of this application embodiment, the public opinion risk index, self-dispute risk index, business relationship risk index, and enterprise prospect index of the target enterprise are weighted and summed to obtain the comprehensive investment risk index of the target enterprise. Multiple comprehensive investment risk indices for multiple target enterprises are then calculated for asset allocation recommendations, including: Acquire multiple target companies for investment; Calculate the public opinion risk index, self-dispute risk index, business relationship risk index, and enterprise prospect index for each target enterprise, and sum them according to the preset weighting rules to obtain the comprehensive investment risk index for each target enterprise; Multiple target companies are listed in ascending order of comprehensive investment risk index to form a recommended asset allocation list.

[0067] In this embodiment of the application, step S500 serves to integrate multi-dimensional risk and prospect indicators, generate a quantitative "comprehensive investment risk index" through weighted summation, and rank multiple investable companies based on the comprehensive investment risk index to form a clear asset allocation recommendation list, thereby solving the problems of fragmented risk assessment and vague decision-making basis in traditional asset allocation decisions.

[0068] First, it is necessary to identify multiple target companies for investment. For example, if a bank plans to allocate 500 million yuan of assets in the new energy industry, it may initially screen out 10 companies that are mainly engaged in new energy power generation and meet the asset size requirements as the target companies to be evaluated. This will ensure that subsequent analysis focuses on the bank's actual intended investment targets and avoids wasting resources.

[0069] Then, the public opinion risk index, self-dispute risk index, business relationship risk index and enterprise prospect index of each target enterprise need to be calculated respectively through the calculation methods recorded in steps S100-S400, and the comprehensive investment risk index of each target enterprise is obtained by weighting and summing according to the preset weighting rules.

[0070] Next, all the potential investee companies are ranked in ascending order of their "comprehensive investment risk index," that is, sorted from low to high risk. For example, among the 10 potential investee companies, if company B has a comprehensive investment risk index of 0.28 (lowest), company A has a comprehensive investment risk index of 0.3475, and company C has a comprehensive investment risk index of 0.42 (highest), then the ranking would be B > A > ... > C.

[0071] Finally, an "Asset Allocation Recommendation List" is generated based on the ranking results. The list should clearly indicate the comprehensive investment risk index and ranking of each enterprise, providing banks with an intuitive reference for investment priorities. Banks can prioritize the allocation of assets to enterprises with higher rankings, such as allocating more investment to enterprise B and reducing or suspending investment in enterprise C with lower rankings, thereby optimizing the asset allocation structure and reducing the overall investment portfolio risk.

[0072] Example 2, as Figure 2 As shown, based on the same inventive concept as the AI-driven intelligent optimization decision-making method for bank asset allocation provided in Embodiment 1, this embodiment of the invention also provides an AI-driven intelligent optimization decision-making system for bank asset allocation, comprising: The public opinion risk analysis module 11 is used to analyze the public opinion status of the target company on the entire network within a preset time range and obtain the public opinion risk index of the target company. The self-dispute risk analysis module 12 is used to retrieve the dispute information of the target enterprise and obtain the self-dispute risk index of the target enterprise based on the pre-trained dispute information analyzer. The business relationship risk analysis module 13 is used to retrieve dispute information and performance information of multiple partner companies of the target company. Based on the performance information, it calculates the historical default rate of multiple partner companies. Based on the dispute information of multiple partner companies and the dispute information analyzer, it obtains multiple dispute risk indices of multiple partner companies. Combining the multiple cooperation scales between the target company and multiple partner companies, as well as the historical default rate and dispute risk index of multiple partner companies, it calculates the business relationship risk index of the target company. The enterprise prospect analysis module 14 is used to obtain the investment rationality index of the target enterprise based on the investment behavior of the target enterprise, obtain the industry prospect index of the target enterprise based on the core business of the target enterprise, and calculate the enterprise prospect index of the target enterprise based on the investment rationality index and the industry prospect index. The comprehensive investment risk assessment module 15 is used to weight and sum the public opinion risk index, self-dispute risk index, business relationship risk index and enterprise prospect index of the target enterprise to obtain the comprehensive investment risk index of the target enterprise. It then calculates multiple comprehensive investment risk indices of multiple target enterprises for asset allocation recommendations.

[0073] Furthermore, the public opinion risk analysis module 11 includes the following execution steps: Based on web crawling technology, capture public opinion information about the target company from the entire network within a preset time range and add it to the public opinion information set; Based on sentiment analysis algorithms, each piece of public opinion information is analyzed for public opinion sentiment and then labeled with public opinion sentiment tags, which include negative, positive, and neutral. The proportion of public opinion information with a negative sentiment label is calculated as the total number of public opinion information in the set, and is used as the public opinion risk index.

[0074] Furthermore, the self-dispute risk analysis module 12 includes the following execution steps: Extract multiple historical defaulting companies from the bank's non-performing asset database; Using the first default time of each historical defaulting enterprise as a node, dispute information of multiple historical defaulting enterprises within a preset period before the node is collected as a historical dispute information set, and the severity of multiple defaults of multiple historical defaulting enterprises corresponding to multiple historical dispute information is marked as a historical default information set, wherein the severity of default includes secondary, suspicious and loss; The dispute information analyzer is built based on machine learning, and the dispute information analyzer is trained using the historical dispute information set and the historical breach of contract information set until convergence. By inputting the dispute information of the target company into the dispute information analyzer, the dispute risk index of the target company can be predicted and output.

[0075] Furthermore, the business relationship risk analysis module 13 includes the following execution steps: Identify multiple companies that have commercial contracts with the target company, as the multiple cooperating companies; The system obtains information on the performance of contracts by multiple partner companies within a preset period. This information is used as performance data. The ratio of the amount of contractual non-performance to the amount of contractual performance due by a partner company within the preset period is used as the historical default rate of the partner company. This process is then used to calculate and obtain multiple historical default rates for the multiple partner companies. Multiple dispute information from multiple partner companies within a preset period is retrieved, and the multiple dispute information is input into the dispute information analyzer to obtain multiple dispute risk indices for the multiple partner companies.

[0076] Specifically, by combining the scale of cooperation between the target company and multiple partner companies, as well as the historical default rates and dispute risk indices of these partner companies, a business relationship risk index for the target company is calculated, including: Calculate the accounts receivable of the target company and multiple partner companies separately, and use this as the scale of cooperation with multiple partner companies; The historical default rates and dispute risk coefficients of multiple partner companies are weighted and integrated to form multiple business risk coefficients for the multiple partner companies; Using the proportion of the cooperation scale of multiple cooperating enterprises to the total accounts receivable of the target enterprise as the weight, the multiple business risk coefficients of the multiple cooperating enterprises are weighted and integrated to obtain the business relationship risk index of the target enterprise.

[0077] Furthermore, the enterprise prospect analysis module 14 includes the following execution steps: Identify the target company's investment areas and, based on preset criteria, identify the hottest financing areas within the target company's industry. The proportion of the target company's investment in the financing hotspots to its total investment is calculated as the target company's investment rationality index. Identify several listed companies whose core businesses are the same as the target company, and obtain the average stock price growth rate of the listed companies within a preset time range as an industry prospect index. The target company's investment rationality index is obtained by multiplying its industry prospect index by the target company's investment rationality index.

[0078] Furthermore, the comprehensive investment risk assessment module 15 includes the following execution steps: Acquire multiple target companies for investment; Calculate the public opinion risk index, self-dispute risk index, business relationship risk index, and enterprise prospect index for each target enterprise, and sum them according to the preset weighting rules to obtain the comprehensive investment risk index for each target enterprise; Multiple target companies are listed in ascending order of comprehensive investment risk index to form a recommended asset allocation list.

[0079] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0080] Those skilled in the art will understand that embodiments of the present invention can provide methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0081] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0082] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0083] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0084] Although preferred embodiments of the invention have been described, those skilled in the art, once they have learned the basic inventive concept, can make other changes and modifications to these embodiments.

[0085] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. An AI-driven intelligent optimization decision-making method for bank asset allocation, characterized in that, include: Analyze the public opinion situation of the target company across the entire network within a preset time range to obtain the public opinion risk index of the target company; Retrieve dispute information of the target company and analyze it based on a pre-trained dispute information analyzer to obtain the target company's own dispute risk index; The system retrieves dispute and performance information from multiple partner companies of the target company. Based on the performance information, it calculates the historical default rate of multiple partner companies. Based on the dispute information and dispute information analyzer of multiple partner companies, it obtains multiple dispute risk indices of multiple partner companies. Combining the multiple cooperation scales between the target company and multiple partner companies, as well as the historical default rate and dispute risk index of multiple partner companies, it calculates the business relationship risk index of the target company. Based on the investment behavior of the target company, obtain the investment rationality index of the target company; based on the core business of the target company, obtain the industry prospect index of the target company; and based on the investment rationality index and the industry prospect index, calculate the enterprise prospect index of the target company. The target company's public opinion risk index, self-dispute risk index, business relationship risk index, and corporate prospect index are weighted and summed to obtain the target company's comprehensive investment risk index. Multiple comprehensive investment risk indices for multiple target companies are then calculated for asset allocation recommendations.

2. The AI-driven intelligent optimization decision-making method for bank asset allocation according to claim 1, characterized in that, Analyze the public opinion situation of the target company across the entire network within a preset time range to obtain the public opinion risk index of the target company, including: Based on web crawling technology, capture public opinion information about the target company from the entire network within a preset time range and add it to the public opinion information set; Based on sentiment analysis algorithms, each piece of public opinion information is analyzed for public opinion sentiment and then labeled with public opinion sentiment tags, which include negative, positive, and neutral. The proportion of public opinion information with a negative sentiment label is calculated as the total number of public opinion information in the set, and is used as the public opinion risk index.

3. The AI-driven intelligent optimization decision-making method for bank asset allocation according to claim 1, characterized in that, The system retrieves dispute information from the target company and analyzes it using a pre-trained dispute information analyzer to obtain the target company's own dispute risk index, including: Extract multiple historical defaulting companies from the bank's non-performing asset database; Using the first default time of each historical defaulting enterprise as a node, dispute information of multiple historical defaulting enterprises within a preset period before the node is collected as a historical dispute information set, and the severity of multiple defaults of multiple historical defaulting enterprises corresponding to multiple historical dispute information is marked as a historical default information set, wherein the severity of default includes secondary, suspicious and loss; The dispute information analyzer is built based on machine learning, and the dispute information analyzer is trained using the historical dispute information set and the historical breach of contract information set until convergence. By inputting the dispute information of the target company into the dispute information analyzer, the dispute risk index of the target company can be predicted and output.

4. The AI-driven intelligent optimization decision-making method for bank asset allocation according to claim 1, characterized in that, Based on performance information, the historical default rates of multiple partner companies are calculated. Based on dispute information and a dispute information analyzer, multiple dispute risk indices for the partner companies are obtained, including: Identify multiple companies that have commercial contracts with the target company, as the multiple cooperating companies; The system obtains information on the performance of contracts by multiple partner companies within a preset period. This information is used as performance data. The ratio of the amount of contractual non-performance to the amount of contractual performance due by a partner company within the preset period is used as the historical default rate of the partner company. This process is then used to calculate and obtain multiple historical default rates for the multiple partner companies. Multiple dispute information from multiple partner companies within a preset period is retrieved, and the multiple dispute information is input into the dispute information analyzer to obtain multiple dispute risk indices for the multiple partner companies.

5. The AI-driven intelligent optimization decision-making method for bank asset allocation according to claim 1, characterized in that, By combining the scale of cooperation between the target company and multiple partner companies, as well as the historical default rates and dispute risk indices of these partner companies, a business relationship risk index for the target company is calculated, including: Calculate the accounts receivable of the target company and multiple partner companies separately, and use this as the scale of cooperation with multiple partner companies; The historical default rates and dispute risk coefficients of multiple partner companies are weighted and integrated to form multiple business risk coefficients for the multiple partner companies; Using the proportion of the cooperation scale of multiple cooperating enterprises to the total accounts receivable of the target enterprise as the weight, the multiple business risk coefficients of the multiple cooperating enterprises are weighted and integrated to obtain the business relationship risk index of the target enterprise.

6. The AI-driven intelligent optimization decision-making method for bank asset allocation according to claim 1, characterized in that, Based on the target company's investment behavior, an investment rationality index is obtained; based on the target company's core business, an industry prospect index is obtained; and based on the investment rationality index and industry prospect index, a company prospect index is calculated, including... Identify the target company's investment areas and, based on preset criteria, identify the hottest financing areas within the target company's industry. The proportion of the target company's investment in the financing hotspots to its total investment is calculated as the target company's investment rationality index. Identify several listed companies whose core businesses are the same as the target company, and obtain the average stock price growth rate of the listed companies within a preset time range as an industry prospect index. The target company's investment rationality index is obtained by multiplying its industry prospect index by the target company's investment rationality index.

7. The AI-driven intelligent optimization decision-making method for bank asset allocation according to claim 1, characterized in that, The target company's public opinion risk index, self-dispute risk index, business relationship risk index, and corporate prospect index are weighted and summed to obtain the target company's comprehensive investment risk index. Multiple comprehensive investment risk indices for multiple target companies are then calculated for asset allocation recommendations, including: Acquire multiple target companies for investment; Calculate the public opinion risk index, self-dispute risk index, business relationship risk index, and enterprise prospect index for each target enterprise, and sum them according to the preset weighting rules to obtain the comprehensive investment risk index for each target enterprise; Multiple target companies are listed in ascending order of comprehensive investment risk index to form a recommended asset allocation list.

8. An AI-driven intelligent optimization decision-making system for bank asset allocation, characterized in that: The system is used to implement the AI-driven intelligent optimization decision-making method for bank asset allocation as described in any one of claims 1-7, the system comprising: The public opinion risk analysis module is used to analyze the public opinion status of a target company across the entire network within a preset time range and obtain the public opinion risk index of the target company. The self-dispute risk analysis module is used to retrieve the dispute information of the target enterprise and obtain the self-dispute risk index of the target enterprise based on the pre-trained dispute information analyzer. The business relationship risk analysis module is used to retrieve dispute and performance information of multiple partner companies of the target company. Based on the performance information, it calculates the historical default rate of multiple partner companies. Based on the dispute information of multiple partner companies and the dispute information analyzer, it obtains multiple dispute risk indices of multiple partner companies. Combining the multiple cooperation scales between the target company and multiple partner companies, as well as the historical default rate and dispute risk index of multiple partner companies, it calculates the business relationship risk index of the target company. The enterprise prospect analysis module is used to obtain the investment rationality index of the target enterprise based on its investment behavior, obtain the industry prospect index of the target enterprise based on its core business, and calculate the enterprise prospect index of the target enterprise based on the investment rationality index and the industry prospect index. The comprehensive investment risk assessment module is used to weight and sum the public opinion risk index, self-dispute risk index, business relationship risk index, and enterprise prospect index of the target enterprise to obtain the comprehensive investment risk index of the target enterprise. It then calculates multiple comprehensive investment risk indices for multiple target enterprises for asset allocation recommendations.

Images