Method and system for matching enterprise financing needs and financial services based on artificial intelligence

Abstract

The method and system for matching enterprise financing needs and financial services based on artificial intelligence provided by the present invention relate to the field of artificial intelligence technology. In the present invention, first, an information acquisition operation is performed based on the financing needs of the target enterprise to obtain a corresponding ordered set of financing needs; secondly, a semantic information mining operation is performed on any financing demand enterprise data in the ordered set of financing needs to form a financing demand semantic vector corresponding to any financing demand enterprise data; then, based on the financing demand semantic vector corresponding to any financing demand enterprise data and the historical enterprise data corresponding to any financing demand enterprise data, a target financial service product matching any financing demand enterprise data is screened out from a plurality of predetermined candidate financial service products. Based on the above method, the problem of relatively low reliability of matching enterprise financing needs and financial services in the prior art can be improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an artificial intelligence-based enterprise financing demand and financial service matching method and system. Background Art

[0002] In the modern economic environment, the diversity and complexity of corporate financing needs are increasing, and traditional financing channels and matching mechanisms often fail to meet the personalized needs of enterprises. The diversity of corporate financing needs is reflected in the fact that companies of different sizes, industries, development stages and financial conditions may face different challenges in the financing process, which makes financing matching more difficult. Traditional financing channels such as bank loans, capital markets, equity investments, etc. often rely on manual judgment and manual operations to meet corporate financing needs, and do not make full use of modern scientific and technological means for accurate analysis.

[0003] Traditional financing matching methods are mostly based on surface features and basic information of enterprises, such as the financial status, scale, industry type, etc. However, this matching method based on fixed rules has great limitations. Because the financing needs of enterprises are not limited to simple financial data, but may also involve more complex semantic information, corporate strategy and other multi-dimensional information. This information is often presented in the form of natural language description or unstructured data. Traditional matching methods are difficult to fully obtain and process this complex semantic information, resulting in low accuracy and reliability of matching results. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a method and system for matching corporate financing needs and financial services based on artificial intelligence, so as to improve the relatively low reliability of matching corporate financing needs and financial services in the prior art.

[0005] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions:

[0006] An artificial intelligence-based method for matching corporate financing needs with financial services, comprising:

[0007] Performing information acquisition operations based on the financing needs of the target enterprise to obtain a corresponding ordered set of financing needs, wherein the ordered set of financing needs includes at least two pieces of financing need enterprise data arranged in a corresponding time sequence, the financing need enterprise data includes text data, and the financing need enterprise data is used to describe the corporate financing related situation of the target enterprise in a corresponding time period;

[0008] Performing semantic information mining operations on any piece of financing demand enterprise data in the ordered set of financing demand to form a financing demand semantic vector corresponding to the any piece of financing demand enterprise data, wherein the financing demand semantic vector is used to reflect semantic information about financing-related situations in the any piece of financing demand enterprise data;

[0009] Based on the financing demand semantic vector corresponding to any one of the financing demand enterprise data and the historical enterprise data corresponding to any one of the financing demand enterprise data, a target financial service product matching any one of the financing demand enterprise data is screened out from a plurality of predetermined candidate financial service products, wherein the historical enterprise data belongs to other financing demand enterprise data whose corresponding time is earlier than the time corresponding to any one of the financing demand enterprise data.

[0010] In some preferred embodiments, in the above-mentioned method for matching enterprise financing needs with financial services based on artificial intelligence, the step of screening out a target financial service product matching any one of the enterprise data on financing needs from a plurality of predetermined candidate financial service products based on the financing demand semantic vector corresponding to any one of the enterprise data on financing needs and the historical enterprise data corresponding to any one of the enterprise data on financing needs, comprises:

[0011] Based on the historical enterprise data corresponding to any one of the enterprise data with financing needs, and in combination with the first direction difference data between the historical enterprise data and any one of the enterprise data with financing needs, a first difference demand semantic vector corresponding to any one of the enterprise data with financing needs is mined, wherein the first direction difference data is used to reflect the difference data in the direction from the historical enterprise data to any one of the enterprise data with financing needs;

[0012] Based on the any one of the financing demand enterprise data and in combination with the second direction difference data between the any one of the financing demand enterprise data and the corresponding historical enterprise data, a second difference demand semantic vector corresponding to the any one of the financing demand enterprise data is mined, wherein the second direction difference data is used to reflect the difference data in the direction from the any one of the financing demand enterprise data to the historical enterprise data;

[0013] A target financial service product matching any one of the financing demand enterprise data is screened out from a plurality of predetermined candidate financial service products based on the financing demand semantic vector, the first difference demand semantic vector and the second difference demand semantic vector.

[0014] In some preferred embodiments, in the above-mentioned method for matching enterprise financing needs with financial services based on artificial intelligence, the step of screening out a target financial service product matching any one of the financing demand enterprise data from a plurality of predetermined candidate financial service products based on the financing demand semantic vector, the first difference demand semantic vector and the second difference demand semantic vector comprises:

[0015] Performing a semantic aggregation operation on the first difference demand semantic vector and the second difference demand semantic vector to form a corresponding aggregated difference demand semantic vector;

[0016] Performing semantic optimization operations on the aggregated difference demand semantic vector and the financing demand semantic vector to form a target financing demand semantic vector corresponding to any one of the financing demand enterprise data; wherein the vector size of the target financing demand semantic vector is the same as the number of the plurality of predetermined candidate financial service products;

[0017] Perform product matching operations according to the target financing demand semantic vector, and output a matching parameter distribution corresponding to any one of the financing demand enterprise data, wherein each matching parameter in the matching parameter distribution is used to reflect the possibility of matching between any one of the financing demand enterprise data and each of the multiple candidate financial service products;

[0018] According to the matching parameter distribution corresponding to the any one piece of enterprise data in need of financing, a target financial service product having the greatest possibility of matching with the any one piece of enterprise data in need of financing is screened out from the plurality of candidate financial service products.

[0019] In some preferred embodiments, in the above-mentioned method for matching enterprise financing needs with financial services based on artificial intelligence, the first difference demand semantic vector includes a first difference demand sub-vector and a first difference demand depth vector, the second difference demand semantic vector includes a second difference demand sub-vector and a second difference demand depth vector, any one of the difference demand depth vectors is formed by deep semantic mining of the corresponding difference demand sub-vector, and the aggregated difference demand semantic vector includes an aggregated difference demand sub-vector and an aggregated difference demand depth vector;

[0020] The step of performing a semantic aggregation operation on the first difference demand semantic vector and the second difference demand semantic vector to form a corresponding aggregated difference demand semantic vector includes:

[0021] Performing a semantic aggregation operation on the first difference demand sub-vector and the second difference demand sub-vector to form a corresponding aggregated difference demand sub-vector;

[0022] A semantic aggregation operation is performed on the first difference required depth vector and the second difference required depth vector to form a corresponding aggregated difference required depth vector.

[0023] In some preferred embodiments, in the above-mentioned method for matching enterprise financing demand and financial services based on artificial intelligence, the financing demand semantic vector includes a financing demand sub-vector and a financing demand depth vector;

[0024] The step of performing semantic optimization operation on the aggregated difference demand semantic vector and the financing demand semantic vector to form a target financing demand semantic vector corresponding to any one of the financing demand enterprise data includes:

[0025] Performing a semantic association aggregation operation on the aggregated difference demand sub-vector and the financing demand sub-vector to form a corresponding associated difference demand sub-vector;

[0026] Performing a deep semantic mining operation on the associated difference demand sub-vector to form a corresponding associated difference demand deep vector;

[0027] The aggregated difference demand depth vector, the associated difference demand depth vector and the financing demand depth vector are subjected to semantic association aggregation operations to form a target financing demand semantic vector corresponding to any one of the financing demand enterprise data.

[0028] In some preferred embodiments, in the above-mentioned artificial intelligence-based enterprise financing demand and financial service matching method, the step of performing semantic association aggregation operation on the aggregated difference demand depth vector, the associated difference demand depth vector and the financing demand depth vector to form a target financing demand semantic vector corresponding to any one of the financing demand enterprise data includes:

[0029] Performing a vector product operation on the aggregated difference demand depth vector, the associated difference demand depth vector, and the financing demand depth vector to form a corresponding multi-dimensional vector product;

[0030] Perform a vector convolution operation and / or a self-attention operation on the multi-dimensional vector product to form a corresponding depth vector product;

[0031] The depth vector product and the financing demand sub-vector are concatenated or averaged to form a target financing demand semantic vector corresponding to any one of the financing demand enterprise data.

[0032] In some preferred embodiments, in the above-mentioned method for matching enterprise financing needs with financial services based on artificial intelligence, the historical enterprise data of any one of the financing demand enterprise data includes the previous financing demand enterprise data and the previous two financing demand enterprise data of any one of the financing demand enterprise data in the ordered set of financing needs;

[0033] The step of mining the first difference demand semantic vector corresponding to any one of the financing demand enterprise data based on the historical enterprise data corresponding to any one of the financing demand enterprise data and combining the first direction difference data between the historical enterprise data and any one of the financing demand enterprise data comprises:

[0034] Performing a merging operation on the previous enterprise data of financing demand and the previous two enterprise data of financing demand of any one enterprise data of financing demand to form corresponding merged historical enterprise data, wherein the merging operation at least includes calculating the mean of the numerical data;

[0035] In the merged historical enterprise data, data different from any one of the financing demand enterprise data is determined to form corresponding first direction difference data;

[0036] Performing a semantic information mining operation on the merged historical enterprise data to form a first merged semantic vector corresponding to the merged historical enterprise data, and performing a semantic information mining operation on the first direction difference data to form a first difference semantic vector corresponding to the first direction difference data;

[0037] Performing a semantic association aggregation operation on the first merged semantic vector and the first difference semantic vector to form a first difference demand sub-vector corresponding to any one of the financing demand enterprise data;

[0038] The first difference demand sub-vector is subjected to a deep semantic mining operation to form a first difference demand deep vector corresponding to any one of the financing demand enterprise data, wherein the first difference demand semantic vector includes the first difference demand sub-vector and the first difference demand deep vector.

[0039] In some preferred embodiments, in the above-mentioned method for matching enterprise financing needs with financial services based on artificial intelligence, the historical enterprise data of any one of the financing demand enterprise data includes the previous financing demand enterprise data and the previous two financing demand enterprise data of any one of the financing demand enterprise data in the ordered set of financing needs;

[0040] The step of mining the second difference demand semantic vector corresponding to the any one piece of financing demand enterprise data based on the any one piece of financing demand enterprise data and combining the second direction difference data between the any one piece of financing demand enterprise data and the corresponding historical enterprise data comprises:

[0041] Performing a merging operation on the previous enterprise data of financing demand and the previous two enterprise data of financing demand of any one enterprise data of financing demand to form corresponding merged historical enterprise data, wherein the merging operation at least includes calculating the mean of the numerical data;

[0042] In any one of the financing demand enterprise data, determine data different from the merged historical enterprise data to form corresponding second direction difference data;

[0043] Performing a semantic information mining operation on the second direction difference data to form a second difference semantic vector corresponding to the second direction difference data;

[0044] Performing a semantic association aggregation operation on the financing demand semantic vector and the second difference semantic vector to form a second difference demand sub-vector corresponding to any one of the financing demand enterprise data;

[0045] The second difference demand sub-vector is subjected to deep semantic mining operation to form a second difference demand deep vector corresponding to any one of the financing demand enterprise data, wherein the second difference demand semantic vector includes the second difference demand sub-vector and the second difference demand deep vector.

[0046] In some preferred embodiments, in the above-mentioned method for matching enterprise financing needs with financial services based on artificial intelligence, the method further includes:

[0047] Determine an ordered set of training financing needs, wherein the ordered set of training financing needs includes at least two pieces of training financing demand enterprise data arranged in a corresponding time sequence;

[0048] Using the semantic information mining model in the candidate matching network, a semantic information mining operation is performed on any one of the training financing demand enterprise data in the ordered set of training financing demand, so as to form a financing demand training vector corresponding to the any one of the training financing demand enterprise data, wherein the candidate matching network belongs to a neural network, and the financing demand training vector is used to reflect the semantic information about the financing-related situation in the any one of the training financing demand enterprise data;

[0049] Utilizing the demand service matching model in the candidate matching network, based on the financing demand training vector corresponding to the any one training financing demand enterprise data and the training history enterprise data corresponding to the any one training financing demand enterprise data, a training financial service product matching the any one training financing demand enterprise data is screened out from a plurality of predetermined candidate financial service products, wherein the training history enterprise data belongs to other training financing demand enterprise data corresponding to a time earlier than a time corresponding to the any one training financing demand enterprise data;

[0050] Based on the difference between the training financial service products and the financial service product labels that match the ordered set of training financing needs, the parameters in the candidate matching network are updated to form a corresponding target matching network, wherein the target matching network is used to screen out target financial service products that match any one of the financing demand enterprise data.

[0051] An embodiment of the present invention also provides an artificial intelligence-based enterprise financing demand and financial service matching system, including a processor and a memory, the memory is used to store a computer program, and the processor is used to execute the computer program to implement the above-mentioned artificial intelligence-based enterprise financing demand and financial service matching method.

[0052] The artificial intelligence-based enterprise financing demand and financial service matching method and system provided in the embodiment of the present invention firstly performs information acquisition operations based on the financing needs of the target enterprise to obtain an ordered set of corresponding financing needs; secondly, performs semantic information mining operations on any financing demand enterprise data in the ordered set of financing needs to form a financing demand semantic vector corresponding to any financing demand enterprise data; then, based on the financing demand semantic vector corresponding to any financing demand enterprise data and the historical enterprise data corresponding to any financing demand enterprise data, screens out a target financial service product that matches any financing demand enterprise data from a plurality of predetermined candidate financial service products. Based on the above method, after the financing demand semantic vector corresponding to the financing demand enterprise data is mined, the product matching will not be directly based on the financing demand semantic vector, but the historical enterprise data corresponding to the financing demand enterprise data will be combined to match the product. In this way, when the historical enterprise data has a certain representation effect on the actual needs of the enterprise (the data generated in the course of business operations are mostly dynamically changing, therefore, it is difficult to characterize the actual situation of the enterprise based only on the data of one time period, such as the data of the current time period, and the historical data can actually positively characterize the data of the current time period, that is, there is a correlation between the two, so that the historical data can be combined, thereby improving the richness of the overall semantic information and the representation accuracy), the reliability of the matching analysis can be improved, thereby improving the relatively low reliability of the matching of enterprise financing needs and financial services in the existing technology.

[0053] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 A structural block diagram of the artificial intelligence-based enterprise financing demand and financial service matching system provided in an embodiment of the present invention.

[0055] Figure 2 A flowchart of the steps of the method for matching enterprise financing needs with financial services based on artificial intelligence provided in an embodiment of the present invention.

[0056] Figure 3 A schematic diagram of mining the first difference demand semantic vector provided by an embodiment of the present invention.

[0057] Figure 4 A schematic diagram of performing deep semantic mining operations on associated difference requirement sub-vectors provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0059] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] like Figure 1 As shown, an embodiment of the present invention provides an artificial intelligence-based enterprise financing demand and financial service matching system. The artificial intelligence-based enterprise financing demand and financial service matching system may include a memory and a processor.

[0061] In detail, the memory and the processor are electrically connected directly or indirectly to achieve data transmission or interaction. For example, they can be electrically connected to each other through one or more communication buses or signal lines. The memory may store at least one software function module (computer program) that may exist in the form of software or firmware. The processor may be used to execute the executable computer program stored in the memory, thereby realizing the method for matching enterprise financing needs and financial services based on artificial intelligence provided in an embodiment of the present invention.

[0062] Optionally, the memory may be, but not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), etc. The processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), a system on chip (SoC), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0063] and, Figure 1 The structure shown is for illustration only. The AI-based enterprise financing demand and financial service matching system may also include Figure 1 More or fewer components as shown, or with Figure 1 The different configurations shown, for example, may include a communication unit for information exchange with other devices. In addition, in an alternative example, the artificial intelligence-based enterprise financing demand and financial service matching system may be a server with data processing capabilities.

[0064] Combination Figure 2 The embodiment of the present invention also provides an artificial intelligence-based enterprise financing demand and financial service matching method, which can be applied to the above-mentioned artificial intelligence-based enterprise financing demand and financial service matching system. Among them, the method steps defined in the process related to the artificial intelligence-based enterprise financing demand and financial service matching method can be implemented by the artificial intelligence-based enterprise financing demand and financial service matching system (hereinafter referred to as the financial service matching system).

[0065] The following will Figure 2 The specific process shown is explained in detail.

[0066] Step S110: performing information acquisition operations based on the financing needs of the target enterprise to obtain an ordered set of corresponding financing needs.

[0067] In an embodiment of the present invention, the financial service matching system can perform information acquisition operations based on the financing needs of the target enterprise to obtain a corresponding ordered set of financing needs. For example, the corresponding financing personnel can send financing demand request information to the financial service matching system through corresponding terminal devices such as mobile phones and computers. In this way, after receiving the financing demand request information, the financial service matching system can parse the financing demand request information to obtain the financing demand enterprise data therein, or the financing demand enterprise data of the target enterprise can be obtained from other databases. Then, the corresponding financing demand ordered set can be formed based on the obtained financing demand enterprise data. Among them, the financing demand ordered set includes at least two financing demand enterprise data arranged according to the corresponding time sequence (such as the financing demand enterprise data of the current period and the financing demand enterprise data of the previous period, and the financing demand enterprise data of different time periods can be partially repeated), the financing demand enterprise data includes text data, and the financing demand enterprise data is used to describe the enterprise financing related situation of the target enterprise in the corresponding time period. For example, in an alternative example, the financing demand enterprise data can include, but is not limited to the following:

[0068] 1. Basic information of the enterprise: Enterprise name and registration information: the legal name, registered address, registered capital, legal representative, company type (such as limited liability company, joint stock company, etc.) and establishment time of the enterprise; Organizational structure and shareholder information: the shareholder structure of the enterprise, the proportion of shareholders' holdings, the background of major shareholders and whether there is any control dispute; Business scope and industry background: the main business of the enterprise, industry attributes, market prospects and development trends of the industry in which it is located; Management team and core personnel background: the qualifications, experience, employment history and background of the company's senior management personnel;

[0069] 2. Financial information: Financial statements: including the company's financial statements for the past 3-5 years (income statement, balance sheet, cash flow statement, etc.), as well as the company's financial health, profitability, debt level, etc. determined by analyzing the financial statements, such as profitability: gross profit margin, net profit margin, ROE (return on equity), ROA (return on total assets), etc.; debt repayment ability: asset-liability ratio, current ratio, quick ratio, interest coverage ratio, etc.; operating efficiency: inventory turnover rate, accounts receivable turnover rate, total asset turnover rate, etc.; cash flow status: cash flow from operating activities, cash flow from investment activities, cash flow from financing activities, etc.; financial audit report: the audit opinion of the audit agency on the financial statements;

[0070] 3. Financing needs and purposes: Financing amount and method: the specific amount of financing required by the enterprise, the financing method (equity financing, debt financing, mixed financing, etc.); Financing purpose: the specific purpose of the financing funds, such as expanding production scale, repaying debts, R&D investment, mergers and acquisitions, etc.; Financing term and repayment plan: if it is debt financing, specify the financing term, interest rate, repayment plan, etc.;

[0071] 4. Credit information: Corporate credit record: whether the company has any record of loan default or overdue payment, and whether it has been included in the list of dishonest persons subject to enforcement; Credit status of shareholders, management and affiliated companies: in addition to the company itself, the credit records of its shareholders and senior management and the credit status of related companies need to be reviewed; Debt default record: whether there have been any debt defaults, lawsuits, and other issues in the past, and whether there have been any events that affect the company's reputation and credit;

[0072] 5. Mortgage and asset status: Mortgage status: If the financing requires mortgage or guarantee, the nature, value, ownership and legal effect of the mortgage shall be reviewed in detail; Major asset evaluation: Evaluation report of the company's major fixed assets, current assets, intangible assets, etc., especially factory buildings, land, equipment, patents, brands, etc.; Guarantor status: If the financing is guaranteed by a third party, the financial status and credibility of the guarantor shall also be reviewed;

[0073] 6. Operational and market risks: Market environment analysis: evaluate the industry and market, taking into account factors such as industry cycles, market demand, technological innovation, and competitive landscape; Operational risks: potential operational risks faced by the company (such as management replacement, changes in business models, product market risks, etc.); Macroeconomic factors: the impact of domestic and foreign economic environments, policy changes, monetary policies, exchange rate risks, etc. on corporate operations; Legal risks: whether the company has potential legal or regulatory risks, such as intellectual property disputes, environmental compliance issues, etc.;

[0074] 7. Debt structure and debt repayment capacity: Current debt situation: the structure of the company's current debt, including short-term debt, long-term debt, loans from financial institutions, interest rates and terms of other debts; Debt repayment history: the company's past debt repayment record, whether there is a history of default, and whether repayment obligations are fulfilled in a timely manner; Debt repayment plan: the company's debt repayment plan after financing, whether there is a clear cash flow to guarantee debt repayment capacity;

[0075] 8. Corporate governance and compliance: Corporate governance structure: the composition of the board of directors, board of supervisors, management and their decision-making mechanisms, whether there is a sound internal control and risk management system; Compliance inspection: whether the company complies with relevant laws and regulations, and whether there are undisclosed compliance risks; Audit and legal opinion: review of the company's legal compliance, intellectual property rights, contract management and other aspects, and whether there are major litigation cases involving the company's operations;

[0076] 9. Enterprise development strategy and prospects: Future development plan: enterprise strategic planning, market expansion plan, technology innovation plan, etc.; Potential merger or investment plan: whether the enterprise has merger, acquisition, investment and other related plans, and their impact on the company's operations and financial status; Capital structure optimization: whether the enterprise has plans to improve capital structure and improve capital utilization efficiency;

[0077] 10. External support and cooperation: Partners and customers: whether the main partners and customers are stable, and the customers' payment ability and credit status; Government support: if the company obtains government support (such as subsidies, policy preferences, tax exemptions, etc.), the sustainability of such support; Certification by industry associations or third-party organizations: whether there is industry certification or qualification, and whether it is recognized within the industry.

[0078] Step S120, performing a semantic information mining operation on any piece of financing demand enterprise data in the ordered set of financing demand to form a financing demand semantic vector corresponding to the any piece of financing demand enterprise data.

[0079] In an embodiment of the present invention, after obtaining the ordered set of financing needs, the financial service matching system may perform semantic information mining operations on any one of the financing demand enterprise data in the ordered set of financing needs to form a financing demand semantic vector corresponding to the any one of the financing demand enterprise data. Among them, the financing demand semantic vector is used to reflect the semantic information about financing-related situations in any one of the financing demand enterprise data; illustratively, any one of the financing demand enterprise data may be the financing demand enterprise data of the current time period (i.e., the most recent time period). In addition, the semantic information mining operation may refer to the embedding processing of text data or the convolution processing of image data, etc., and text data is used as an example for explanation. For example, part of the financing demand enterprise data is "the specific use of financing funds includes expanding the scale of production", which can be processed by word segmentation and embedding through the trained word embedding model to obtain the following word embedding vector:

[0080] “Financing” → [0.12, 0.76, 0.83, 0.45, 0.66, -0.23, 0.45, ..., 0.01];

[0081] “Funds” → [-0.05, 0.21, 0.71, 0.13, 0.55, 0.46, -0.34, ..., 0.23];

[0082] “of” → [0.02, 0.15, -0.06, 0.36, 0.43, 0.75, 0.96, ..., -0.10];

[0083] “Specific” → [0.30, -0.14, 0.07, 0.16, 0.84, 0.47, 0.77, ..., 0.25];

[0084] “Purpose” → [0.56, 0.43, 0.48, 0.61, -0.03, 0.27, -0.15, ..., 0.14];

[0085] “include” → [0.19, 0.21, 0.37, 0.78, 0.89, -0.11, 0.22, ..., -0.02];

[0086] “Expand” → [0.40, -0.12, 0.31, 0.33, 0.28, 0.80, 0.35, ..., 0.05];

[0087] “Production” → [0.15, 0.40, -0.21, 0.41, 0.21, 0.34, 0.49, ..., -0.04];

[0088] “scale” → [-0.08, 0.18, 0.20, 0.51, 0.44, 0.78, 0.02, …, 0.10];

[0089] After obtaining the above-mentioned word embedding vectors, the embedding vectors representing the position of each word in the corresponding text can be superimposed to obtain the target vector of each word. Then, the target vectors of each word can be spliced ​​or superimposed to obtain the corresponding financing demand semantic vector, or the splicing or superposition results can be self-attention processed to obtain the corresponding financing demand semantic vector. The specific method can be selected according to actual needs.

[0090] Step S130, based on the financing demand semantic vector corresponding to any one of the financing demand enterprise data and the historical enterprise data corresponding to any one of the financing demand enterprise data, screen out a target financial service product matching any one of the financing demand enterprise data from a plurality of predetermined candidate financial service products.

[0091] In an embodiment of the present invention, after obtaining the financing demand semantic vector, the financial service matching system can screen out a target financial service product that matches any one of the financing demand enterprise data from a plurality of predetermined candidate financial service products based on the financing demand semantic vector corresponding to any one of the financing demand enterprise data and the historical enterprise data corresponding to any one of the financing demand enterprise data. Among them, the historical enterprise data belongs to other financing demand enterprise data whose corresponding time is earlier than the time corresponding to any one of the financing demand enterprise data. In other words, the semantic information of the historical enterprise data can be integrated into the financing demand semantic vector to form a target vector with richer and more reliable semantic representation, and then, based on the target vector, the target financial service product that matches any one of the financing demand enterprise data can be screened out (such as classified output based on the target vector). It should be noted that the multiple candidate financial service products may be multiple financial products of multiple financial companies (such as banks, etc.), or multiple financial products of a financial company. Financial products may refer to loans, etc., and the information of at least one dimension between each two financial products is different, such as the corresponding financial companies are different, the corresponding loan amounts are different, the corresponding loan interest rates are different, the corresponding repayment times are different, the corresponding repayment methods are different (such as installment repayment, one-time repayment upon maturity, etc.), the corresponding mortgage requirements are different, the corresponding late payment fees are different, the corresponding withdrawal times are different, etc.

[0092] Based on the above method, after the financing demand semantic vector corresponding to the financing demand enterprise data is mined, the product matching will not be directly based on the financing demand semantic vector, but the historical enterprise data corresponding to the financing demand enterprise data will be combined to match the product. In this way, when the historical enterprise data has a certain representation effect on the actual needs of the enterprise (the data generated in the course of business operations are mostly dynamically changing, therefore, it is difficult to characterize the actual situation of the enterprise based only on the data of one time period, such as the data of the current time period, and the historical data can actually positively characterize the data of the current time period, that is, there is a correlation between the two, so that the historical data can be combined, thereby improving the richness of the overall semantic information and the representation accuracy), the reliability of the matching analysis can be improved, thereby improving the relatively low reliability of the matching of enterprise financing needs and financial services in the existing technology.

[0093] It should be noted that for step S130, the specific method of screening out the target financial service product matching any one of the financing demand enterprise data is not limited and can be selected according to actual conditions.

[0094] For example, in an alternative example, in order to fully mine the required semantic information so that reliable target financial service products can be screened out based on the mined semantic information, the above-mentioned step S130 can further include step S131, step S132 and step S133, as described below.

[0095] Step S131, based on the historical enterprise data corresponding to any one of the financing demand enterprise data, and in combination with the first direction difference data between the historical enterprise data and any one of the financing demand enterprise data, a first difference demand semantic vector corresponding to any one of the financing demand enterprise data is mined.

[0096] In an embodiment of the present invention, the first difference demand semantic vector corresponding to any one of the financing demand enterprise data can be mined based on the historical enterprise data corresponding to the any one of the financing demand enterprise data and the first direction difference data between the historical enterprise data and the any one of the financing demand enterprise data. The first direction difference data is used to reflect the difference data in the direction from the historical enterprise data to the any one of the financing demand enterprise data.

[0097] Step S132, based on the any one of the financing demand enterprise data and in combination with the second direction difference data between the any one of the financing demand enterprise data and the corresponding historical enterprise data, a second difference demand semantic vector corresponding to the any one of the financing demand enterprise data is mined.

[0098] In an embodiment of the present invention, the second difference demand semantic vector corresponding to the any one enterprise data for financing demand can be mined based on the any one enterprise data for financing demand and in combination with the second direction difference data between the any one enterprise data for financing demand and the corresponding historical enterprise data. The second direction difference data is used to reflect the difference data in the direction from the any one enterprise data for financing demand to the historical enterprise data. In other words, the difference data is determined from two different directions, so that the representation ability of the difference data in the two directions is better. In addition, since the difference data is between the any one enterprise data for financing demand and the previous historical enterprise data, that is, it represents some changes in the target enterprise during its operation, and these changes actually have a representation effect on the financing needs of the target enterprise, and therefore, can be mined.

[0099] Step S133, based on the financing demand semantic vector, the first difference demand semantic vector and the second difference demand semantic vector, a target financial service product matching any one of the financing demand enterprise data is screened out from a plurality of predetermined candidate financial service products.

[0100] In an embodiment of the present invention, after obtaining the financing demand semantic vector, the first difference demand semantic vector, and the second difference demand semantic vector, a target financial service product matching the any one of the financing demand enterprise data is screened out from a plurality of predetermined candidate financial service products based on the financing demand semantic vector, the first difference demand semantic vector, and the second difference demand semantic vector. In other words, matching and screening can be performed based on semantic information of multiple dimensions, so that the basis for matching and screening is more reliable, thereby obtaining reliable results.

[0101] It can be selected that, in the above-mentioned step S131, the specific method of mining out the first difference demand semantic vector corresponding to any one of the financing demand enterprise data is not restricted and can be selected according to actual conditions. For example, in an alternative example, in order to ensure that the semantic representation accuracy of the mined first difference demand semantic vector can be higher, the historical enterprise data of any one of the financing demand enterprise data includes the previous financing demand enterprise data and the previous two financing demand enterprise data of any one of the financing demand enterprise data in the ordered set of financing demands (that is, the financing demand enterprise data of the two most recent historical time periods. For example, in a specific application, any one of the financing demand enterprise data may refer to the financing demand enterprise data in 2024, the previous financing demand enterprise data may refer to the financing demand enterprise data in 2023, and the previous two financing demand enterprise data may refer to the financing demand enterprise data in 2022). Based on this, combined with Figure 3 The above step S131 may further include the following specific implementation process:

[0102] The first step is to merge the previous financing demand enterprise data and the previous two financing demand enterprise data of any one of the financing demand enterprise data to form corresponding merged historical enterprise data, wherein the merging operation at least includes performing mean calculation on the numerical data; illustratively, the mean calculation can be performed on the numerical data, and for other data, the content in the previous financing demand enterprise data can be retained, and the content in the previous two financing demand enterprise data can be discarded; for data that only exists in one of the previous financing demand enterprise data and the previous two financing demand enterprise data, it can be directly retained;

[0103] The second step is to determine data different from any one of the financing demand enterprise data in the merged historical enterprise data to form corresponding first direction difference data; illustratively, data different from the internal data of any one of the financing demand enterprise data can be screened out, and data not included in any one of the financing demand enterprise data can be screened out;

[0104] The third step is to perform a semantic information mining operation on the merged historical enterprise data to form a first merged semantic vector corresponding to the merged historical enterprise data, and to perform a semantic information mining operation on the first direction difference data to form a first difference semantic vector corresponding to the first direction difference data; wherein the method of the semantic information mining operation is as described above, such as word embedding processing, etc.;

[0105] The fourth step is to perform a semantic association aggregation operation on the first merged semantic vector and the first difference semantic vector to form a first difference demand sub-vector corresponding to any one of the financing demand enterprise data; illustratively, the first merged semantic vector and the first difference semantic vector can be vector-producted to obtain a corresponding multi-dimensional vector product, and then the multi-dimensional vector product can be vector-convolved and / or self-attention-operated to obtain a corresponding deep vector product, and then the deep vector product and the first merged semantic vector can be concatenated or mean-calculated to obtain a corresponding first difference demand sub-vector;

[0106] The fifth step is to perform deep semantic mining operations on the first difference demand sub-vector (the specific processing process is described later) to form a first difference demand deep vector corresponding to any one of the financing demand enterprise data, wherein the first difference demand semantic vector includes the first difference demand sub-vector and the first difference demand deep vector.

[0107] It can be selected that, in the above-mentioned step S132, the specific method of mining out the second difference demand semantic vector corresponding to any one of the financing demand enterprise data is not limited and can be selected according to actual conditions. For example, in an alternative example, in order to ensure that the semantic representation accuracy of the mined second difference demand semantic vector can be higher, the historical enterprise data of any one of the financing demand enterprise data includes the previous financing demand enterprise data and the previous two financing demand enterprise data of any one of the financing demand enterprise data in the financing demand ordered set (that is, the financing demand enterprise data of the two most recent historical time periods, for example, in a specific application, any one of the financing demand enterprise data can refer to the financing demand enterprise data of 2024, the previous financing demand enterprise data can refer to the financing demand enterprise data of 2023, and the previous two financing demand enterprise data can refer to the financing demand enterprise data of 2022). Based on this, the above-mentioned step S132 can further include the following specific implementation process:

[0108] The first step is to merge the previous financing demand enterprise data and the previous two financing demand enterprise data of any one financing demand enterprise data to form corresponding merged historical enterprise data, wherein the merging operation at least includes calculating the mean of the numerical data, as described above;

[0109] The second step is to determine data different from the merged historical enterprise data in any one of the financing demand enterprise data to form corresponding second direction difference data, as described above;

[0110] The third step is to perform a semantic information mining operation on the second direction difference data to form a second difference semantic vector corresponding to the second direction difference data, as described above;

[0111] The fourth step is to perform a semantic association aggregation operation on the financing demand semantic vector and the second difference semantic vector to form a second difference demand sub-vector corresponding to any one of the financing demand enterprise data; illustratively, the financing demand semantic vector and the second difference semantic vector can be vector-producted to obtain a corresponding multi-dimensional vector product, and then the multi-dimensional vector product can be vector-convolved and / or self-attention-operated to obtain a corresponding deep vector product, and then the deep vector product and the financing demand semantic vector can be concatenated or mean-calculated to obtain a corresponding second difference demand sub-vector;

[0112] The fifth step is to perform deep semantic mining operations on the second difference demand sub-vector (refer to the relevant description below) to form a second difference demand deep vector corresponding to any one of the financing demand enterprise data, wherein the second difference demand semantic vector includes the second difference demand sub-vector and the second difference demand deep vector.

[0113] Optionally, in the above-mentioned step S133, the specific method of screening out the target financial service products that match any one of the financing demand enterprise data is not restricted and may be selected according to actual conditions. For example, in an alternative example, in order to be able to fully fuse the financing demand semantic vector, the first difference demand semantic vector and the second difference demand semantic vector, so that the fused vector can fully characterize the semantic information of the three vectors, thereby determining a reliable target financial service product, the above-mentioned step S133 may further include step S133a, step S133b, step S133c and step S133d, as described below.

[0114] Step S133a: performing a semantic aggregation operation on the first difference demand semantic vector and the second difference demand semantic vector to form a corresponding aggregated difference demand semantic vector.

[0115] In an embodiment of the present invention, a semantic aggregation operation may be performed on the first difference demand semantic vector and the second difference demand semantic vector to form a corresponding aggregated difference demand semantic vector. In this way, the aggregated difference demand semantic vector may aggregate the semantic information of two dimensions of the first difference demand semantic vector and the second difference demand semantic vector, so that it has a more reliable semantic representation capability. In addition, the semantic aggregation operation may refer to concatenation or mean calculation, etc.

[0116] Step S133b: performing semantic optimization operation on the aggregated difference demand semantic vector and the financing demand semantic vector to form a target financing demand semantic vector corresponding to any one of the financing demand enterprise data.

[0117] In an embodiment of the present invention, after obtaining the aggregated difference demand semantic vector, the aggregated difference demand semantic vector and the financing demand semantic vector may be subjected to semantic optimization operations to form a target financing demand semantic vector corresponding to any one of the financing demand enterprise data. The vector size of the target financing demand semantic vector (i.e., the number of vector parameters included) is the same as the number of a plurality of predetermined candidate financial service products. Exemplarily, after the aggregated difference demand semantic vector and the financing demand semantic vector are subjected to semantic optimization operations, the corresponding output vector may be subjected to full connection processing to obtain a corresponding fully connected vector, which may be used as the target financing demand semantic vector.

[0118] Step S133c, performing product matching operations according to the target financing demand semantic vector, and outputting matching parameter distribution corresponding to any one financing demand enterprise data.

[0119] In an embodiment of the present invention, after obtaining the target financing demand semantic vector, a product matching operation can be performed based on the target financing demand semantic vector, and a matching parameter distribution corresponding to any one of the financing demand enterprise data can be output. Among them, each matching parameter in the matching parameter distribution is used to reflect the possibility of matching between any one of the financing demand enterprise data and each of the multiple candidate financial service products. Exemplarily, the target financing demand semantic vector can be processed by a classification function such as softmax to obtain a corresponding matching parameter distribution, and the sum of the matching parameters in the matching parameter distribution is equal to 1. Exemplarily, the target financing demand semantic vector can be {z1, z2, z3, ..., zn}, and then these vector parameters can be converted into a probability distribution through a softmax function, such as:

[0120] ;

[0121] in, For the vector parameter " "The probability distribution corresponding to (i.e., matching parameters).

[0122] Step S133d, based on the matching parameter distribution corresponding to the any one of the financing demand enterprise data, select from the multiple candidate financial service products a target financial service product that has the greatest possibility of matching with the any one of the financing demand enterprise data.

[0123] In an embodiment of the present invention, after obtaining the matching parameter distribution, the target financial service product with the greatest possibility of matching with any one of the financing demand enterprise data can be screened out from the multiple candidate financial service products according to the matching parameter distribution corresponding to any one of the financing demand enterprise data. Exemplarily, in the matching parameter distribution, there are matching parameters corresponding to each candidate financial service product, so that the candidate financial service product with the largest matching parameter can be determined as the target financial service product with the greatest possibility of matching with any one of the financing demand enterprise data.

[0124] It can be selected that, in the above step S133a, the specific manner of performing semantic aggregation operation on the first difference demand semantic vector and the second difference demand semantic vector is not limited and can be selected according to actual conditions. For example, in an alternative example, in order to ensure the reliability of the semantic aggregation operation, the first difference demand semantic vector includes a first difference demand sub-vector and a first difference demand depth vector, the second difference demand semantic vector includes a second difference demand sub-vector and a second difference demand depth vector, any one of the difference demand depth vectors is formed by deep semantic mining of the corresponding difference demand sub-vector, and the aggregated difference demand semantic vector includes an aggregated difference demand sub-vector and an aggregated difference demand depth vector. In this way, the above step S133a can further include the following specific implementation process:

[0125] The first step is to perform a semantic aggregation operation (as described above) on the first difference demand sub-vector and the second difference demand sub-vector to form a corresponding aggregated difference demand sub-vector;

[0126] In the second step, a semantic aggregation operation is performed on the first difference required depth vector and the second difference required depth vector (as described above) to form a corresponding aggregated difference required depth vector.

[0127] It can be selected that, in the above-mentioned step S133b, the specific manner of performing semantic optimization operations on the aggregated difference demand semantic vector and the financing demand semantic vector is not restricted and can be selected according to actual conditions. For example, in an alternative example, in order to ensure the reliability of the semantic optimization operation, the target financing demand semantic vector can take into account both the deep information and the shallow information in the aggregated difference demand semantic vector and the financing demand semantic vector, thereby ensuring the semantic representation ability of the target financing demand semantic vector. In the case where the financing demand semantic vector includes a financing demand sub-vector and a financing demand deep vector (for example, by performing the above-mentioned semantic information mining operation on any one of the financing demand enterprise data, the corresponding financing demand sub-vector can be obtained, and then, the above-mentioned deep semantic mining operation is performed on the financing demand sub-vector to obtain the corresponding financing demand deep vector), the above-mentioned step S133b can further include steps b1, b2 and b3, as described below.

[0128] Step b1, performing a semantic association aggregation operation on the aggregated difference demand sub-vector and the financing demand sub-vector to form a corresponding associated difference demand sub-vector.

[0129] In an embodiment of the present invention, the aggregated difference demand sub-vector and the financing demand sub-vector may be subjected to a semantic association aggregation operation to form a corresponding associated difference demand sub-vector. In this way, the associated difference demand sub-vector may carry the semantic information of the aggregated difference demand sub-vector and the financing demand sub-vector, respectively, so that the semantic information is richer. Exemplarily, the aggregated difference demand sub-vector and the financing demand sub-vector may be subjected to a vector product operation to obtain a corresponding multi-dimensional vector product, and then, the multi-dimensional vector product may be subjected to a vector convolution operation and / or a self-attention operation to obtain a corresponding deep vector product, and then, the deep vector product and the financing demand sub-vector may be concatenated or mean-calculated to obtain an associated difference demand sub-vector.

[0130] Step b2: performing a deep semantic mining operation on the associated difference requirement sub-vector to form a corresponding associated difference requirement deep vector.

[0131] In an embodiment of the present invention, after obtaining the associated difference requirement sub-vector, the associated difference requirement sub-vector may be subjected to a deep semantic mining operation to form a corresponding associated difference requirement deep vector, so that the associated difference requirement deep vector may have the ability to represent deep semantic information.

[0132] Step b3, performing semantic association aggregation operation on the aggregated difference demand depth vector, the associated difference demand depth vector and the financing demand depth vector to form a target financing demand semantic vector corresponding to any one of the financing demand enterprise data.

[0133] In an embodiment of the present invention, after obtaining the associated difference demand depth vector, the aggregated difference demand depth vector, the associated difference demand depth vector and the financing demand depth vector can be subjected to semantic association aggregation operations to form a target financing demand semantic vector corresponding to any one of the financing demand enterprise data. In this way, the target financing demand semantic vector can represent the semantic information in the three-dimensional vectors.

[0134] Among them, it can be selected that in the above step b2, the specific method of performing the deep semantic mining operation on the associated difference demand sub-vector is not limited and can be selected according to the actual situation. For example, in an alternative example, in order to ensure the reliability of the deep semantic mining operation, the formed associated difference demand deep vector has richer details, thereby capturing more abstract and semantic features, combined with Figure 4 , the above step b2 may further include:

[0135] The first step is to perform sliding window segmentation on the associated difference demand sub-vector to form a plurality of sliding window vectors corresponding to the associated difference demand sub-vector, wherein the window size and step size of the sliding window can be updated as parameters of the corresponding neural network during the training process;

[0136] In the second step, for every two adjacent sliding window vectors (the adjacent ones may be determined according to the order of the sliding windows) among the multiple sliding window vectors, based on the previous sliding window vector, a cross attention process is performed on the latter sliding window vector (the cross attention process may refer to the relevant prior art) to obtain a corresponding cross attention vector;

[0137] The third step is to perform cross-attention processing on the first sliding window vector based on the last sliding window vector to obtain a corresponding cross-attention vector;

[0138] The fourth step is to sort the cross-attention vectors according to the order relationship between the corresponding sliding window vectors, and for every two adjacent cross-attention vectors, based on the previous cross-attention vector, perform cross-attention processing on the latter cross-attention vector to obtain a new cross-attention vector, and then perform cross-attention processing on the first cross-attention vector based on the last cross-attention vector to obtain a corresponding new cross-attention vector;

[0139] The fifth step is to combine the new cross-attention vectors according to the order relationship between the corresponding cross-attention vectors to form a corresponding association difference requirement depth vector, such as {the first new cross-attention vector, the second new cross-attention vector, the third new cross-attention vector, ..., the last new cross-attention vector}, or, the mean of the vector parameters of each new cross-attention vector can be calculated separately, and then the means can be combined to form a corresponding association difference requirement depth vector, such as {the first mean, the second mean, the third mean, ..., the last mean}.

[0140] Optionally, in the above step b3, the specific manner of performing semantic association aggregation operation on the aggregate difference demand depth vector, the association difference demand depth vector and the financing demand depth vector is not limited and can be selected according to actual conditions. For example, in an alternative example, in order to ensure the reliability of the semantic association aggregation operation, the above step b3 can further include the following specific implementation process:

[0141] The first step is to perform a vector product operation on the aggregate difference demand depth vector, the associated difference demand depth vector and the financing demand depth vector to form a corresponding multi-dimensional vector product. In this way, through the vector product operation, when realizing the fusion of multiple vectors, compared with splicing, mean and other fusions, nonlinear mapping of semantic information can be achieved, thereby capturing more abstract features;

[0142] The second step is to perform a vector convolution operation and / or a self-attention operation on the multi-dimensional vector product to form a corresponding depth vector product;

[0143] The third step is to concatenate or average the depth vector product and the financing demand sub-vector to form a target financing demand semantic vector corresponding to any one of the financing demand enterprise data.

[0144] For the above content, it is necessary to further explain that, in order to ensure the reliable implementation of the above process, such as the reliable implementation of step S120 and step S130, the above steps S120 and S130 can be performed through the target matching network formed by training. Based on this, the method for matching corporate financing needs and financial services based on artificial intelligence can also include the step of training to form the target matching network, which can include the following specific implementation process:

[0145] The first step is to determine an ordered set of training financing needs, wherein the ordered set of training financing needs includes at least two training financing demand enterprise data arranged in a corresponding time sequence, i.e., sample data. The specific data content can refer to the financing demand enterprise data in the previous text, and will not be described one by one here;

[0146] The second step is to use the semantic information mining model in the candidate matching network to perform semantic information mining operations on any one of the training financing demand enterprise data in the ordered set of training financing demand (as described above, i.e., refer to the process of performing semantic information mining operations on the financing demand enterprise data), to form a financing demand training vector corresponding to any one of the training financing demand enterprise data, wherein the candidate matching network belongs to a neural network, and the financing demand training vector is used to reflect the semantic information about financing-related situations in any one of the training financing demand enterprise data;

[0147] The third step is to use the demand service matching model in the candidate matching network to select a training financial service product that matches any one of the training financing demand enterprise data from a plurality of predetermined candidate financial service products based on the financing demand training vector corresponding to any one of the training financing demand enterprise data and the training historical enterprise data corresponding to any one of the training financing demand enterprise data (as described above, i.e., referring to the process of selecting a target financial service product that matches any one of the financing demand enterprise data), wherein the training historical enterprise data belongs to other training financing demand enterprise data whose corresponding time is earlier than the corresponding time of any one of the training financing demand enterprise data;

[0148] In the fourth step, based on the difference between the training financial service product and the financial service product label that matches the ordered set of training financing needs (exemplarily, the output of the demand service matching model can be a matching parameter distribution, so the corresponding difference can be characterized by calculating the cross-entropy loss), the parameters in the candidate matching network are updated to form a corresponding target matching network, wherein the target matching network is used to screen out target financial service products that match any one of the financing demand enterprise data; exemplarily, the parameters in the candidate matching network can be updated in the direction of reducing the difference until the difference converges (such as the difference is less than a preset value or the reduction in the difference is less than a preset range) to obtain the corresponding target matching network. For example, the weights and biases and other parameters of the neural network can be continuously updated through a gradient descent algorithm to minimize the loss and thus optimize the performance of the network.

[0149] To summarize, the artificial intelligence-based enterprise financing demand and financial service matching method and system provided by the present invention firstly performs information acquisition operations based on the financing needs of the target enterprise to obtain an ordered set of corresponding financing needs; secondly, performs semantic information mining operations on any financing demand enterprise data in the ordered set of financing needs to form a financing demand semantic vector corresponding to any financing demand enterprise data; then, based on the financing demand semantic vector corresponding to any financing demand enterprise data and the historical enterprise data corresponding to any financing demand enterprise data, screens out a target financial service product that matches any financing demand enterprise data from a plurality of predetermined candidate financial service products. Based on the above method, after the financing demand semantic vector corresponding to the financing demand enterprise data is mined, the product matching will not be directly based on the financing demand semantic vector, but the historical enterprise data corresponding to the financing demand enterprise data will be combined to match the product. In this way, when the historical enterprise data has a certain representation effect on the actual needs of the enterprise (the data generated in the course of business operations are mostly dynamically changing, therefore, it is difficult to characterize the actual situation of the enterprise based only on the data of one time period, such as the data of the current time period, and the historical data can actually positively characterize the data of the current time period, that is, there is a correlation between the two, so that the historical data can be combined, thereby improving the richness of the overall semantic information and the representation accuracy), the reliability of the matching analysis can be improved, thereby improving the relatively low reliability of the matching of enterprise financing needs and financial services in the existing technology.

[0150] In several embodiments provided in the embodiments of the present invention, it should be understood that the disclosed apparatus and method can also be implemented in other ways. The apparatus and method embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the apparatus, method and computer program product according to multiple embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, program segment or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0151] In addition, the functional modules in the various embodiments of the present invention may be integrated together to form an independent part, or each module may exist independently, or two or more modules may be integrated to form an independent part.

[0152] If the function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, an electronic device, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program code. It should be noted that in this article, the term "include", "include" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements that are not explicitly listed, or also includes elements inherent to such process, method, article or device. Without more constraints, an element defined by the phrase "comprising a..." does not exclude the existence of other identical elements in the process, method, article or apparatus comprising the element.

[0153] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for matching enterprise financing needs and financial services based on artificial intelligence, characterized in that: The method includes: Performing information acquisition operations based on the financing needs of the target enterprise to obtain a corresponding ordered set of financing needs, wherein the ordered set of financing needs includes at least two pieces of financing need enterprise data arranged in a corresponding time sequence, the financing need enterprise data includes text data, and the financing need enterprise data is used to describe the corporate financing related situation of the target enterprise in a corresponding time period; Performing semantic information mining operations on any piece of financing demand enterprise data in the ordered set of financing demand to form a financing demand semantic vector corresponding to the any piece of financing demand enterprise data, wherein the financing demand semantic vector is used to reflect semantic information about financing-related situations in the any piece of financing demand enterprise data; Based on the historical enterprise data corresponding to any one of the enterprise data with financing needs, and in combination with the first direction difference data between the historical enterprise data and any one of the enterprise data with financing needs, a first difference demand semantic vector corresponding to the enterprise data with financing needs is mined, wherein the first direction difference data is used to reflect the difference data in the direction from the historical enterprise data to any one of the enterprise data with financing needs; based on the any one of the enterprise data with financing needs, and in combination with the second direction difference data between the any one of the enterprise data with financing needs and the corresponding historical enterprise data, a second difference demand semantic vector corresponding to the enterprise data with financing needs is mined, wherein the second direction difference data is used to reflect the difference data in the direction from any one of the enterprise data with financing needs to the historical enterprise data; semantic aggregation operation is performed on the first difference demand semantic vector and the second difference demand semantic vector to form a corresponding aggregated difference demand semantic vector; the aggregated difference demand semantic vector is used to generate a corresponding aggregated difference demand semantic vector. The vector and the financing demand semantic vector are subjected to semantic optimization operation to form a target financing demand semantic vector corresponding to any one of the financing demand enterprise data; wherein the vector size of the target financing demand semantic vector is the same as the number of a plurality of predetermined candidate financial service products; a product matching operation is performed based on the target financing demand semantic vector, and a matching parameter distribution corresponding to any one of the financing demand enterprise data is output, wherein each matching parameter in the matching parameter distribution is used to reflect the possibility of matching between any one of the financing demand enterprise data and each candidate financial service product in the plurality of candidate financial service products; based on the matching parameter distribution corresponding to any one of the financing demand enterprise data, a target financial service product with the greatest possibility of matching with any one of the financing demand enterprise data is screened out from the plurality of candidate financial service products, wherein the historical enterprise data belongs to other financing demand enterprise data whose corresponding time is earlier than the time corresponding to any one of the financing demand enterprise data.

2. The method for matching enterprise financing needs and financial services based on artificial intelligence as claimed in claim 1, characterized in that: The first difference demand semantic vector includes a first difference demand sub-vector and a first difference demand depth vector, the second difference demand semantic vector includes a second difference demand sub-vector and a second difference demand depth vector, any difference demand depth vector is formed by deep semantic mining of the corresponding difference demand sub-vector, and the aggregated difference demand semantic vector includes an aggregated difference demand sub-vector and an aggregated difference demand depth vector; The step of performing a semantic aggregation operation on the first difference demand semantic vector and the second difference demand semantic vector to form a corresponding aggregated difference demand semantic vector includes: Performing a semantic aggregation operation on the first difference demand sub-vector and the second difference demand sub-vector to form a corresponding aggregated difference demand sub-vector; A semantic aggregation operation is performed on the first difference required depth vector and the second difference required depth vector to form a corresponding aggregated difference required depth vector.

3. The method for matching enterprise financing needs and financial services based on artificial intelligence as claimed in claim 2, characterized in that: The financing demand semantic vector includes a financing demand sub-vector and a financing demand depth vector; The step of performing semantic optimization operation on the aggregated difference demand semantic vector and the financing demand semantic vector to form a target financing demand semantic vector corresponding to any one financing demand enterprise data includes: Performing a semantic association aggregation operation on the aggregated difference demand sub-vector and the financing demand sub-vector to form a corresponding associated difference demand sub-vector; Performing a deep semantic mining operation on the associated difference demand sub-vector to form a corresponding associated difference demand deep vector; The aggregated difference demand depth vector, the associated difference demand depth vector and the financing demand depth vector are subjected to semantic association aggregation operations to form a target financing demand semantic vector corresponding to any one of the financing demand enterprise data.

4. The method for matching enterprise financing needs and financial services based on artificial intelligence as claimed in claim 3, characterized in that: The step of performing semantic association aggregation operation on the aggregated difference demand depth vector, the associated difference demand depth vector and the financing demand depth vector to form a target financing demand semantic vector corresponding to any one financing demand enterprise data includes: Performing a vector product operation on the aggregated difference demand depth vector, the associated difference demand depth vector, and the financing demand depth vector to form a corresponding multi-dimensional vector product; Perform a vector convolution operation and / or a self-attention operation on the multi-dimensional vector product to form a corresponding depth vector product; The depth vector product and the financing demand sub-vector are concatenated or averaged to form a target financing demand semantic vector corresponding to any one of the financing demand enterprise data.

5. The method for matching enterprise financing needs and financial services based on artificial intelligence as claimed in claim 1, characterized in that: The historical enterprise data of any one piece of enterprise data in need of financing includes the previous piece of enterprise data in need of financing and the previous two pieces of enterprise data in need of financing in the ordered set of financing requirements; The step of mining the first difference demand semantic vector corresponding to any one of the financing demand enterprise data based on the historical enterprise data corresponding to any one of the financing demand enterprise data and combining the first direction difference data between the historical enterprise data and any one of the financing demand enterprise data comprises: Performing a merging operation on the previous enterprise data of financing demand and the previous two enterprise data of financing demand of any one enterprise data of financing demand to form corresponding merged historical enterprise data, wherein the merging operation at least includes calculating the mean of the numerical data; In the merged historical enterprise data, data different from any one of the financing demand enterprise data is determined to form corresponding first direction difference data; Performing a semantic information mining operation on the merged historical enterprise data to form a first merged semantic vector corresponding to the merged historical enterprise data, and performing a semantic information mining operation on the first direction difference data to form a first difference semantic vector corresponding to the first direction difference data; Performing a semantic association aggregation operation on the first merged semantic vector and the first difference semantic vector to form a first difference demand sub-vector corresponding to any one of the financing demand enterprise data; The first difference demand sub-vector is subjected to a deep semantic mining operation to form a first difference demand deep vector corresponding to any one of the financing demand enterprise data, wherein the first difference demand semantic vector includes the first difference demand sub-vector and the first difference demand deep vector.

6. The method for matching enterprise financing needs and financial services based on artificial intelligence as claimed in claim 1, characterized in that: The historical enterprise data of any one piece of enterprise data in need of financing includes the previous piece of enterprise data in need of financing and the previous two pieces of enterprise data in need of financing in the ordered set of financing requirements; The step of mining the second difference demand semantic vector corresponding to any one of the financing demand enterprise data based on the any one of the financing demand enterprise data and combining the second direction difference data between the any one of the financing demand enterprise data and the corresponding historical enterprise data comprises: Performing a merging operation on the previous enterprise data of financing demand and the previous two enterprise data of financing demand of any one enterprise data of financing demand to form corresponding merged historical enterprise data, wherein the merging operation at least includes calculating the mean of the numerical data; In any one of the financing demand enterprise data, determine data different from the merged historical enterprise data to form corresponding second direction difference data; Performing a semantic information mining operation on the second direction difference data to form a second difference semantic vector corresponding to the second direction difference data; Performing a semantic association aggregation operation on the financing demand semantic vector and the second difference semantic vector to form a second difference demand sub-vector corresponding to any one of the financing demand enterprise data; The second difference demand sub-vector is subjected to deep semantic mining operation to form a second difference demand deep vector corresponding to any one of the financing demand enterprise data, wherein the second difference demand semantic vector includes the second difference demand sub-vector and the second difference demand deep vector.

7. The method for matching enterprise financing needs and financial services based on artificial intelligence according to any one of claims 1 to 6, characterized in that: The method further includes: Determine an ordered set of training financing needs, wherein the ordered set of training financing needs includes at least two pieces of training financing demand enterprise data arranged in a corresponding time sequence; Using the semantic information mining model in the candidate matching network, a semantic information mining operation is performed on any one of the training financing demand enterprise data in the ordered set of training financing demand, so as to form a financing demand training vector corresponding to the any one of the training financing demand enterprise data, wherein the candidate matching network belongs to a neural network, and the financing demand training vector is used to reflect the semantic information about the financing-related situation in the any one of the training financing demand enterprise data; Utilizing the demand service matching model in the candidate matching network, based on the financing demand training vector corresponding to the any one training financing demand enterprise data and the training historical enterprise data corresponding to the any one training financing demand enterprise data, a training financial service product matching the any one training financing demand enterprise data is screened out from a plurality of predetermined candidate financial service products, wherein the training historical enterprise data belongs to other training financing demand enterprise data corresponding to a time earlier than a time corresponding to the any one training financing demand enterprise data; Based on the difference between the training financial service products and the financial service product labels that match the ordered set of training financing needs, the parameters in the candidate matching network are updated to form a corresponding target matching network, wherein the target matching network is used to screen out target financial service products that match any one of the financing demand enterprise data.

8. An artificial intelligence-based enterprise financing demand and financial service matching system, characterized in that: It includes a processor and a memory, the memory is used to store a computer program, and the processor is used to execute the computer program to implement the method for matching enterprise financing needs and financial services based on artificial intelligence as described in any one of claims 1 to 7.

Images