Financial service risk assessment method and device, computer device and storage medium

Abstract

The application relates to a financial service risk assessment method and device, computer equipment, a storage medium and a computer program product, and relates to the technical field of information security. The method comprises the following steps: acquiring function requirement text information corresponding to a to-be-evaluated financial service; performing keyword extraction on the function requirement text information to obtain an initial function requirement feature set; matching each initial function requirement feature in the initial function requirement feature set with a deviant risk feature in a preset deviant risk knowledge base; in the initial function requirement feature set, determining an initial function requirement feature that is successfully matched with the deviant risk feature as a target function requirement feature; inputting the target function requirement feature into a preset deviant risk assessment model to obtain a deviant risk level, and determining the deviant risk level as a risk assessment result corresponding to the to-be-evaluated financial service. The method can improve the evaluation efficiency of the deviant risk of the financial service.

Description

Technical Field

[0001] This application relates to the field of information security technology, and in particular to a risk assessment method, apparatus, computer equipment, storage medium, and computer program product for financial business. Background Technology

[0002] In the financial sector, when innovating financial services, business personnel often focus solely on the technical implementation of new services, arbitrarily proposing business requirements based on new technologies and urging the technology department to develop and launch them. However, after the financial institution's technology department completes the development and launches the new financial service, it may trigger negative public opinion or even be penalized by regulatory authorities because it does not comply with fintech standards. Therefore, conducting a risk assessment of non-compliance in financial services before development can effectively avoid subsequent losses due to non-compliance issues.

[0003] Traditional techniques typically rely on the professional experience of reviewers to assess whether new financial transactions pose a risk of financial irregularities. However, this traditional, human-based approach to risk assessment becomes inefficient, especially when dealing with large-scale financial transactions. Summary of the Invention

[0004] Therefore, it is necessary to provide a risk assessment method, apparatus, computer equipment, storage medium, and computer program product for financial businesses that can improve the efficiency of risk assessment for irregularities in financial businesses, addressing the aforementioned technical problems.

[0005] Firstly, this application provides a risk assessment method for financial transactions. The method includes:

[0006] Obtain the textual information of the functional requirements corresponding to the financial business to be evaluated;

[0007] Keyword extraction is performed on the functional requirement text information to obtain an initial functional requirement feature set;

[0008] Each initial functional requirement feature in the initial functional requirement feature set is matched with the deviation risk features in the preset deviation risk knowledge base;

[0009] In the initial functional requirement feature set, the initial functional requirement features that successfully match the deviation risk features are determined as target functional requirement features;

[0010] The target functional requirements are input into a preset deviation risk assessment model to obtain a deviation risk level, and the deviation risk level is determined as the risk assessment result corresponding to the financial business to be assessed.

[0011] In one embodiment, the method further includes:

[0012] When the level of deviation risk is non-compliance or suspected non-compliance, the similarity between each initial functional requirement feature in the initial functional requirement feature set and the alternative deviation risk features in the preset alternative deviation risk knowledge base is calculated to obtain the similarity between each initial functional requirement feature and the alternative deviation risk features.

[0013] The similarity between all initial functional requirement features and alternative deviation risk features is summed to obtain the total similarity.

[0014] If the sum of the similarities is greater than or equal to a preset similarity threshold, a potential risk marker is added to the risk assessment result, wherein the potential risk marker is used to indicate that the financial business to be assessed has potential irregularities.

[0015] In one embodiment, the step of inputting the target functional requirement characteristics into a preset deviation risk assessment model to obtain a deviation risk level includes:

[0016] The target functional requirements are input into a pre-defined deviation risk assessment model;

[0017] In the aforementioned deviation risk assessment model, a set of audit indicators is determined based on the target functional requirements characteristics;

[0018] In the deviation risk assessment model, the evaluation result set corresponding to the audit indicator set is obtained, and the evaluation matrix related to the audit indicator set and the evaluation result set is obtained. Based on the evaluation result set and the evaluation matrix, a comprehensive evaluation matrix is ​​calculated.

[0019] In the aforementioned deviation risk assessment model, the deviation risk level is determined based on the deviation risk value calculated from the audit indicator set, the audit indicator weights corresponding to the audit indicator set, and the comprehensive evaluation matrix.

[0020] In one embodiment, the method further includes:

[0021] Obtain information on historical financial irregularities;

[0022] Feature extraction is performed on the information of the historical financial misconduct events to obtain the misconduct risk characteristics corresponding to the information of the historical financial misconduct events.

[0023] Obtain the evaluation results corresponding to the deviation risk characteristics, wherein the evaluation results are used to characterize the degree to which the deviation risk characteristics conform to the norms;

[0024] Obtain the credibility factor corresponding to the evaluation result;

[0025] Based on the characteristics of the deviation risk, the evaluation results, and the credibility factor, the deviation risk knowledge base is constructed.

[0026] In one embodiment, the training process of the deviation risk assessment model includes:

[0027] The characteristics of the deviation risk in the deviation risk knowledge base are used as a sample audit indicator set and input into the initial deviation risk assessment model for training to obtain the deviation risk assessment model.

[0028] The deviation risk assessment model includes an evaluation matrix; the evaluation matrix is ​​obtained iteratively based on the initial evaluation matrix in the initial deviation risk assessment model; the initial evaluation matrix is ​​determined based on the credibility factor in the deviation risk knowledge base; the credibility factor is determined based on the deviation risk characteristics in the sample audit indicator set and the evaluation results in the evaluation result set.

[0029] In one embodiment, the method further includes:

[0030] Based on preset financial keywords, obtain a collection of financial information from the internet;

[0031] The financial information set is classified to obtain a subset of financial information for each category;

[0032] For each category of financial information subset, a preset number of target financial information items are selected from the financial information subset based on the frequency of occurrence of the financial keywords in the financial information subset.

[0033] For each category of target financial information, extract the stem information from the target financial information to determine the alternative deviation risk characteristics;

[0034] Based on the characteristics of the alternative deviation risks, construct the alternative deviation risk knowledge base.

[0035] Secondly, this application also provides a risk assessment device for financial transactions. The device includes:

[0036] The information acquisition module is used to acquire textual information on the functional requirements of the financial business to be evaluated.

[0037] The keyword extraction module is used to extract keywords from the functional requirement text information to obtain an initial functional requirement feature set.

[0038] The feature matching module is used to match each initial functional requirement feature in the initial functional requirement feature set with the deviation risk features in the preset deviation risk knowledge base;

[0039] The feature determination module is used to determine the initial functional requirement features that successfully match the deviation risk features in the initial functional requirement feature set as target functional requirement features;

[0040] The result determination module is used to input the target functional requirement characteristics into a preset deviation risk assessment model to obtain the deviation risk level, and determine the deviation risk level as the risk assessment result corresponding to the financial business to be assessed.

[0041] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to perform the following steps:

[0042] Obtain the textual information of the functional requirements corresponding to the financial business to be evaluated;

[0043] Keyword extraction is performed on the functional requirement text information to obtain an initial functional requirement feature set;

[0044] Each initial functional requirement feature in the initial functional requirement feature set is matched with the deviation risk features in the preset deviation risk knowledge base;

[0045] In the initial functional requirement feature set, the initial functional requirement features that successfully match the deviation risk features are determined as target functional requirement features;

[0046] The target functional requirements are input into a preset deviation risk assessment model to obtain a deviation risk level, and the deviation risk level is determined as the risk assessment result corresponding to the financial business to be assessed.

[0047] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, performs the following steps:

[0048] Obtain the textual information of the functional requirements corresponding to the financial business to be evaluated;

[0049] Keyword extraction is performed on the functional requirement text information to obtain an initial functional requirement feature set;

[0050] Each initial functional requirement feature in the initial functional requirement feature set is matched with the deviation risk features in the preset deviation risk knowledge base;

[0051] In the initial functional requirement feature set, the initial functional requirement features that successfully match the deviation risk features are determined as target functional requirement features;

[0052] The target functional requirements are input into a preset deviation risk assessment model to obtain a deviation risk level, and the deviation risk level is determined as the risk assessment result corresponding to the financial business to be assessed.

[0053] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, performs the following steps:

[0054] Obtain the textual information of the functional requirements corresponding to the financial business to be evaluated;

[0055] Keyword extraction is performed on the functional requirement text information to obtain an initial functional requirement feature set;

[0056] Each initial functional requirement feature in the initial functional requirement feature set is matched with the deviation risk features in the preset deviation risk knowledge base;

[0057] In the initial functional requirement feature set, the initial functional requirement features that successfully match the deviation risk features are determined as target functional requirement features;

[0058] The target functional requirements are input into a preset deviation risk assessment model to obtain a deviation risk level, and the deviation risk level is determined as the risk assessment result corresponding to the financial business to be assessed.

[0059] The aforementioned risk assessment methods, devices, computer equipment, storage media, and computer program products for financial businesses first acquire the functional requirement text information corresponding to the financial business to be assessed. Then, keywords are extracted from the functional requirement text information to obtain an initial set of functional requirement features. Next, each initial functional requirement feature in the initial set is matched with a pre-set knowledge base of misconduct risks. The initial functional requirement features that successfully match the misconduct risk features are identified as target functional requirement features. Finally, the target functional requirement features are input into a pre-set misconduct risk assessment model to assess the misconduct risk and obtain the misconduct risk level, i.e., the risk assessment result. It can be understood that this application first extracts keywords from the functional requirement text of the financial business as initial functional requirement features, then selects target functional requirement features from the initial functional requirement features that are strongly correlated with the pre-set misconduct risk features, and finally processes the target functional requirement features based on the misconduct risk assessment model to achieve the purpose of automatically assessing the misconduct risk result, which is beneficial to improving the efficiency of misconduct risk assessment. Attached Figure Description

[0060] Figure 1 This is a flowchart illustrating a risk assessment method for financial transactions in one embodiment;

[0061] Figure 2 This is a flowchart illustrating a financial transaction with potential irregularities in one embodiment.

[0062] Figure 3 This is a structural block diagram of a risk assessment device for financial transactions in one embodiment;

[0063] Figure 4 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0065] In one embodiment, such as Figure 1 As shown, a risk assessment method for financial transactions is provided. This embodiment illustrates the method applied to a server, but it is understood that the method can also be applied to a terminal, or to a system including both a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0066] Step S102: Obtain the text information of the functional requirements corresponding to the financial business to be evaluated.

[0067] Among them, functional requirement text information refers to functional requirement information presented in text form.

[0068] Specifically, when a financial institution's business personnel propose a new financial service and the technology department undertakes its development, the business personnel must submit a project initiation application. To ensure that the technology department personnel have a deep understanding of the new financial service and can develop it according to the business personnel's ideas, the project document in the initiation application will contain a detailed textual description of the specific functional requirements of the financial service. Based on this, the server can extract the corresponding functional requirement text information of the financial service to be evaluated from the project document.

[0069] Step S104: Extract keywords from the functional requirement text information to obtain an initial functional requirement feature set.

[0070] The initial functional requirement feature set refers to a set consisting of at least one initial functional requirement feature.

[0071] Specifically, the server extracts at least one keyword from the functional requirement text information and uses this keyword as an initial functional requirement feature to obtain an initial functional requirement feature set. Optionally, the server extracts keywords from the functional requirement text information based on the term frequency–inverse document frequency (TF-IDF) method.

[0072] Step S106: Match each initial functional requirement feature in the initial functional requirement feature set with the deviation risk features in the preset deviation risk knowledge base.

[0073] Optionally, a pre-defined knowledge base of misconduct risks can be stored in the server. The characteristics of misconduct risks are determined based on information about historical financial misconduct events.

[0074] Specifically, the server matches each initial functional requirement feature in the initial functional requirement feature set with a pre-defined deviation risk knowledge base. Optionally, feature matching can be implemented by calculating the similarity between the initial functional requirement features and the deviation risk features. In one embodiment, the server uses a semantic similarity calculation method based on a semantic dictionary to calculate the similarity between the initial functional requirement features and the deviation risk features. If the initial functional requirement features and the deviation risk features are exactly the same, their similarity is 1; if the initial functional requirement features and the deviation risk features are completely different concepts, their similarity is close to 0.

[0075] Step S108: In the initial functional requirement feature set, the initial functional requirement features that successfully match the deviation risk features are identified as target functional requirement features.

[0076] Specifically, in the initial set of functional requirement features, the server identifies the initial functional requirement features that successfully match the deviation risk features as target functional requirement features. Optionally, the server identifies initial functional requirement features with a similarity greater than or equal to a similarity threshold as initial functional requirement features that successfully match the deviation risk features.

[0077] Optionally, in one embodiment, the server sums the similarities between all initial functional requirement features and deviation risk features to obtain a total similarity. If the total similarity is greater than or equal to a preset similarity threshold, then the initial functional requirement feature that successfully matches the deviation risk feature in the initial functional requirement feature set is determined as the target functional requirement feature; if the total similarity is less than the preset similarity threshold, then the process is stopped.

[0078] Step S110: Input the target functional requirement characteristics into the preset deviation risk assessment model to obtain the deviation risk level, and determine the deviation risk level as the risk assessment result corresponding to the financial business to be assessed.

[0079] Specifically, the server inputs the target functional requirement features into a pre-trained deviation risk assessment model (e.g., a random forest model, a gradient boosting decision tree model, etc.) to assess the deviation risk, obtains the deviation risk level, and determines the deviation risk level as the risk assessment result corresponding to the financial business to be assessed. Optionally, the deviation risk level includes non-compliance level, suspected non-compliance level, and compliance level.

[0080] The aforementioned risk assessment method for financial businesses first obtains the functional requirement text information corresponding to the financial business to be assessed. Then, keywords are extracted from the functional requirement text information to obtain an initial set of functional requirement features. Next, each initial functional requirement feature in the initial set is matched with a pre-set knowledge base of misconduct risks. The initial functional requirement features that successfully match the misconduct risk features are identified as target functional requirement features. Finally, the target functional requirement features are input into a pre-set misconduct risk assessment model to perform a misconduct risk assessment, obtaining the misconduct risk level, i.e., the risk assessment result. In essence, this method first extracts keywords from the functional requirement text of the financial business as initial functional requirement features, then selects target functional requirement features from the initial functional requirement features that are strongly correlated with pre-set misconduct risk features, and finally processes the target functional requirement features based on the misconduct risk assessment model to achieve the purpose of automatically assessing the misconduct risk result, thus improving the efficiency of misconduct risk assessment.

[0081] Considering that building an expert knowledge base requires analyzing a large amount of historical financial irregularities and summarizing extensive expert experience, both data analysis and expert judgment take time. It's difficult to keep the knowledge in the expert knowledge base up-to-date in real time, making it hard to ensure it reflects the latest social events and scenarios. For example, a new type of mortgage business has emerged—relay loans. This type of personal housing relay loan refers to a borrower applying for a loan term that doesn't match the bank's age requirements (e.g., a 50-year-old applying for a 30-year repayment period, exceeding the maximum age limit of 70). This can be achieved by adding a co-borrower (a co-owner of the purchased property who is the primary borrower's child). This business circumvents purchase and loan restrictions, operating in a gray area and prohibited in some regions. When financial institutions discover competitors offering this business and gradually dominating the market, they may request project development. If the review process is still based on the current expert knowledge base model, the project application is likely to be approved. This is where building a backup risk knowledge base, serving as a pool of potential experts, comes into play. Based on the matching results from this backup knowledge base, the risk assessment results are flagged, informing business personnel that the business is related to current social controversies. Regardless of the specific area of ​​misconduct, launching such a business carries potential fintech misconduct risks and could even result in regulatory penalties. This information is also communicated to relevant management personnel. Consequently, the business department submitting the project application will consider the risks and modify or withdraw the application, thus achieving the goal of risk prevention through this review process.

[0082] Based on this, in one embodiment, such as Figure 2 As shown, the method also includes the following steps:

[0083] Step S112: When the level of non-compliance risk is non-compliant or suspected non-compliant, calculate the similarity between each initial functional requirement feature in the initial functional requirement feature set and the alternative non-compliance risk features in the preset alternative non-compliance risk knowledge base to obtain the similarity between each initial functional requirement feature and the alternative non-compliance risk features.

[0084] Step S114: Sum the similarities between all the initial functional requirement features and the alternative deviation risk features to obtain the total similarity.

[0085] Step S116: If the total similarity is greater than or equal to the preset similarity threshold, then add a potential risk marker to the risk assessment result.

[0086] Among them, the potential risk marker is used to indicate the potential for irregularities in the financial business being evaluated.

[0087] Specifically, when the risk level of non-compliance is at the non-compliant or suspected non-compliant level, the server uses a semantic similarity calculation method based on a semantic dictionary. It calculates the similarity between each initial functional requirement feature in the initial functional requirement feature set and a pre-set list of alternative non-compliant risk features in the alternative non-compliant risk knowledge base, obtaining the similarity between each initial functional requirement feature and the alternative non-compliant risk features. Then, the server sums the similarities between all initial functional requirement features and alternative non-compliant risk features to obtain a total similarity. Finally, the server compares the total similarity with a pre-set similarity threshold. If the total similarity is greater than or equal to the pre-set similarity threshold, a potential risk marker is added to the risk assessment result, and the risk assessment result with the potential risk marker is fed back to business personnel or technology personnel; if the total similarity is less than the pre-set similarity threshold, the risk assessment result is fed back to business personnel or technology personnel.

[0088] In this embodiment, financial transactions that pass the initial risk identification for irregularities are matched again with a candidate risk knowledge base for irregularities. Financial transactions with a high similarity to the candidate risk knowledge base are then alerted to potential risks of irregularities. This is equivalent to a second review of the financial transactions, which helps reduce the possibility of financial transactions from the business departments of financial institutions entering the development stage with risks of irregularities, reduces the risk of financial technology irregularities in newly applied businesses, and reduces the possibility of financial institutions getting caught in a media storm and suffering regulatory penalties. It also avoids the ineffective waste of human resources in the technology departments of financial institutions and promotes the healthy development of the technology work of financial institutions.

[0089] In one embodiment, step S110 includes the following steps:

[0090] Step S1102: Input the target functional requirement characteristics into the preset deviation risk assessment model;

[0091] Step S1104: In the deviation risk assessment model, determine the audit indicator set based on the target functional requirements characteristics;

[0092] Step S1106: In the deviation risk assessment model, obtain the evaluation result set corresponding to the audit indicator set, and obtain the evaluation matrix related to the audit indicator set and the evaluation result set, and calculate the comprehensive evaluation matrix based on the evaluation result set and the evaluation matrix.

[0093] Step S1108: In the deviation risk assessment model, the deviation risk level is determined based on the deviation risk value calculated from the audit indicator set, the audit indicator weights corresponding to the audit indicator set, and the comprehensive evaluation matrix.

[0094] Specifically, the deviation risk assessment model in this embodiment is constructed based on the analytic hierarchy process (AHP). Based on this, the server inputs the target functional requirement characteristics into the preset deviation risk assessment model. In the deviation risk assessment model, based on the target functional requirement characteristics, a set of audit indicators X is determined, where X = (X1, X2, X3, ..., X...). n The target functional requirement characteristics are a subset of the audit indicators in the audit indicator set. Then, in the deviation risk assessment model, the evaluation result set W corresponding to the audit indicator set is obtained, W = (W1, W2, W3, ..., W...). m The process involves obtaining the evaluation matrix E, an m x n matrix, related to the audit indicator set W and the evaluation result set W. Then, in the deviation risk assessment model, the comprehensive evaluation matrix G, G = E × W, is calculated based on the evaluation result set W and the evaluation matrix E. Finally, in the deviation risk assessment model, the deviation risk value C, C = X × A × G, is calculated based on the audit indicator set X, the corresponding audit indicator weights A, and the comprehensive evaluation matrix G. T Based on this deviation risk value C, the deviation risk level is determined, as shown in Table 1.

[0095] Risk value of deviation C Risk level of deviation <![CDATA[0~C1]]> Compliant with regulations <![CDATA[C1~C2]]> Suspected of not conforming to regulations <![CDATA[C2~1]]> Non-compliant with regulations

[0096] Table 1

[0097] In this embodiment, a deviation risk assessment model based on the analytic hierarchy process is used to process the characteristics of target functional requirements to obtain the deviation risk level, which helps to improve the accuracy of the deviation risk level.

[0098] In one embodiment, the training process of the deviation risk assessment model includes the following steps:

[0099] Step S132: The characteristics of the violation risk in the violation risk knowledge base are used as the sample audit indicator set and input into the initial violation risk assessment model for training to obtain the violation risk assessment model.

[0100] The deviation risk assessment model includes an evaluation matrix. This evaluation matrix is ​​obtained iteratively from the initial evaluation matrix in the initial deviation risk assessment model. The initial evaluation matrix is ​​determined based on the credibility factors in the deviation risk knowledge base. The credibility factors are determined based on the deviation risk characteristics in the sample audit indicator set and the evaluation results in the evaluation result set. In this embodiment, the evaluation parameters of the deviation risk assessment model are determined based on the credibility factors in the deviation risk knowledge base. Since the credibility factors are given by experts based on experience, this helps improve the accuracy of model training.

[0101] In some real-world scenarios, fintech misconduct manifests in two ways: First, data ethics violations. For example, some institutions and technologists sacrifice data privacy, leading to the theft or sale of customer information. They lack the fiduciary duty to protect sensitive information such as customers' linked bank cards and accounts, resulting in related unfair transactions, restricted transactions, and bundled sales. Given the digital literacy gap among residents, some technological innovations fail to respect the needs and rights of different groups, leading to unequal distribution of information dividends. Second, algorithmic ethics violations. Algorithms transform people's various footprints and activities into scores and predictions for loan and insurance assessments. However, the codification of rules leads to opacity, inaccuracy, unfairness, and difficulty in auditing. Algorithms can strengthen control over financial consumer behavior, placing consumers in a vulnerable position, deeply embedded in an invisible and unfree environment, creating algorithmic control problems.

[0102] In one embodiment, a knowledge base for the risk of misconduct is constructed. The purpose of this expert knowledge base is to record, based on the analysis of historical financial misconduct events and expert experience, whether financial transactions comply with technological standards in the aforementioned misconduct scenarios. Based on this, in one embodiment, the method further includes the following steps:

[0103] Step S121: Obtain information on historical financial misconduct events;

[0104] Step S122: Extract features from historical financial misconduct information to obtain the misconduct risk features corresponding to the historical financial misconduct information;

[0105] Step S123: Obtain the evaluation results corresponding to the characteristics of deviation risk;

[0106] Step S124: Obtain the credibility factor corresponding to the evaluation result;

[0107] Step S125: Construct a knowledge base for deviation risk based on deviation risk characteristics, evaluation results, and credibility factors.

[0108] The evaluation results are used to characterize the degree to which the characteristics of deviation risk conform to the norms.

[0109] Specifically, firstly, the server acquires information on historical financial misconduct events and extracts features from this information to obtain the corresponding misconduct risk characteristics. Let S represent the misconduct risk characteristics obtained after analyzing and processing the historical financial misconduct event information. Misconduct risk characteristics could include, for example, contact information, transfer amounts, etc.

[0110] The server then retrieves the evaluation results corresponding to the deviation risk characteristics. Let Q be the qualitative evaluation result of the deviation risk characteristics. The evaluation result could be, for example, compliant with regulations, basically compliant with regulations, non-compliant with regulations, etc. The evaluation result is determined by experts based on experience.

[0111] Next, the server obtains the credibility factor corresponding to the evaluation result. The credibility factor can also be called the rule compliance strength. Let f(q / s) represent the credibility factor. The credibility factor is given by experts based on experience. For example, regarding the deviation risk characteristic "metaverse," if experts determine that its corresponding evaluation result is compliant with the norms, then a credibility factor D1 is given; if experts determine that its corresponding evaluation result is basically compliant with the norms, then a credibility factor D2 is given. Obviously, from the perspective of compliance with the norms, D1 > D2.

[0112] Based on this, we assume that R is a set of rules, which is the summarized expert knowledge.

[0113] The construction process of R is as follows:

[0114] R: S→Q, S={s1,s2,…,s n R = {r1, r2, ..., r} n}, Q={q1,q2,…,q n},

[0115] Ri:if s i then q i with f(q / s)

[0116] Finally, the server constructs a knowledge base for deviation risk based on deviation risk characteristics, evaluation results, and credibility factors.

[0117] In this embodiment, a knowledge base for financial misconduct risk is constructed based on historical information on financial misconduct events and expert experience. As an expert knowledge base, the financial business related to financial misconduct can be accurately identified based on the characteristics of misconduct risk in the knowledge base, thereby improving the accuracy of misconduct risk assessment.

[0118] Given the rapid development of society, new scenarios are likely to emerge in the fintech field. Since expert experience has historical limitations, these new scenarios may harbor potential risks of fintech misconduct (such as ethical violations). These new scenarios may involve aspects such as: compliance with legal and ethical standards; honesty and trustworthiness; accountability; respect for privacy and confidentiality; and protection of public interests. If these aspects are not addressed, they will ultimately manifest as misconduct through data or algorithms. Therefore, these aspects must be considered during the review of new scenarios. However, whether new scenarios involving these aspects by experts comply with regulations is still under discussion, and financial institution personnel may propose innovative financial services related to these aspects. Therefore, a candidate misconduct risk knowledge base needs to be built for these new scenarios, serving as a candidate expert knowledge base. Once the official body has qualitatively characterized the potential misconduct risk characteristics in the candidate knowledge base, these characteristics will be added to the misconduct risk knowledge base, i.e., the expert knowledge base, ensuring timely updates to the expert knowledge base.

[0119] Based on this, in one embodiment, the method further includes the following steps:

[0120] Step S141: Based on preset financial keywords, obtain a set of financial information from the network;

[0121] Step S142: Classify the financial information set to obtain a subset of financial information corresponding to each category;

[0122] Step S143: For each category of financial information subset, select a preset number of target financial information items from the financial information subset based on the frequency of occurrence of financial keywords in the financial information subset;

[0123] Step S144: For each category of target financial information, extract the stem information from the target financial information and determine it as a candidate feature of deviant risk.

[0124] Step S145: Construct a knowledge base of alternative deviation risks based on the characteristics of alternative deviation risks.

[0125] Among them, a financial information set refers to a set consisting of at least one piece of financial information.

[0126] Specifically, first, the server runs a web crawler to collect a certain number (potentially over 10,000) of financial information (e.g., news articles) using pre-defined financial keywords (e.g., loans, wealth management, funds). Then, the server uses regular expressions to match and remove special symbols from each piece of financial information. Next, the server uses a machine learning classification model (e.g., Naive Bayes, k-nearest neighbors, support vector machines) to classify the financial information set, obtaining a subset of financial information for each category. Then, for each category's subset of financial information, the server selects a preset number (at least 10) of target financial information articles in descending order of frequency based on the frequency of financial keywords within the subset. Next, for each category's target financial information, the server uses the Porter2 algorithm to extract stemming information from the target financial information, identifying potential deviant risk features. Finally, the server constructs a potential deviant risk knowledge base based on these potential deviant risk features.

[0127] For example, regarding recent hot topics such as mortgage repayment deferrals due to the pandemic, news items such as "how to apply for mortgage repayment deferral," "how to defer mortgage repayments during the pandemic," "can mortgage repayments be deferred during the pandemic," "applying for mortgage repayment deferral during the pandemic," and "is mortgage repayment deferral during the pandemic legal" can be categorized into the same new event. The stem of these topics can be extracted using the Porter2 algorithm, and finally, mortgage repayment deferrals can be included as a candidate feature for irregular risk in the database.

[0128] In this embodiment, financial information disseminated online is acquired, categorized, and selected based on its popularity. The stems of these stems are then extracted and stored as potential risk features for financial technology misconduct. Since these popular financial news items are more likely to harbor potential financial technology misconduct risks, key stem information is extracted as potential risk features. Based on these potential risk features, financial transactions related to these popular financial news items can be accurately identified, thereby alerting business personnel or technology department staff to potential misconduct risks associated with these transactions.

[0129] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0130] Based on the same inventive concept, this application also provides a risk assessment device for financial business to implement the risk assessment method for financial business involved above. The solution provided by this device is similar to the implementation solution described in the above method. Therefore, the specific limitations of one or more embodiments of the risk assessment device for financial business provided below can be found in the limitations of the risk assessment method for financial business above, and will not be repeated here.

[0131] In one embodiment, such as Figure 3 As shown, a risk assessment device for financial operations is provided, comprising:

[0132] Information acquisition module 202 is used to acquire text information of functional requirements corresponding to the financial business to be evaluated;

[0133] Keyword extraction module 204 is used to extract keywords from functional requirement text information to obtain an initial functional requirement feature set;

[0134] The feature matching module 206 is used to match each initial functional requirement feature in the initial functional requirement feature set with the deviation risk features in the preset deviation risk knowledge base;

[0135] The feature determination module 208 is used to determine the initial functional requirement features that successfully match the deviation risk features in the initial functional requirement feature set as target functional requirement features;

[0136] The result determination module 210 is used to input the target functional requirement characteristics into the preset deviation risk assessment model, obtain the deviation risk level, and determine the deviation risk level as the risk assessment result corresponding to the financial business to be assessed.

[0137] In the aforementioned risk assessment device for financial business, the functional requirement text information corresponding to the financial business to be assessed is first obtained. Then, keywords are extracted from the functional requirement text information to obtain an initial set of functional requirement features. Next, each initial functional requirement feature in the initial set is matched with a pre-set knowledge base of irregularity risk features. The initial functional requirement features that successfully match the irregularity risk features are identified as target functional requirement features. Finally, the target functional requirement features are input into a pre-set irregularity risk assessment model to perform an irregularity risk assessment, obtaining the irregularity risk level, i.e., the risk assessment result. In essence, this device first extracts keywords from the functional requirement text of the financial business as initial functional requirement features, then selects target functional requirement features from the initial functional requirement features that are strongly correlated with the pre-set irregularity risk features, and finally processes the target functional requirement features based on the irregularity risk assessment model to achieve the purpose of automatically assessing the irregularity risk result, thus improving the efficiency of irregularity risk assessment.

[0138] In one embodiment, the device further includes:

[0139] The similarity calculation module is used to calculate the similarity between each initial functional requirement feature in the initial functional requirement feature set and the alternative deviation risk features in the preset alternative deviation risk knowledge base when the deviation risk level is non-compliant or suspected non-compliant.

[0140] The similarity accumulation module is used to accumulate the similarity between all the initial functional requirement features and the alternative deviation risk features to obtain the total similarity.

[0141] The tagging module is used to add potential risk tags to the risk assessment results if the sum of similarities is greater than or equal to a preset similarity threshold. The potential risk tags are used to indicate that there is a potential risk of irregularity in the financial business being assessed.

[0142] In one embodiment, the result determination module 210 is specifically used to input the target functional requirement characteristics into a preset deviation risk assessment model; in the deviation risk assessment model, a set of audit indicators is determined based on the target functional requirement characteristics; in the deviation risk assessment model, an evaluation result set corresponding to the audit indicator set is obtained, as well as an evaluation matrix related to the audit indicator set and the evaluation result set, and a comprehensive evaluation matrix is ​​calculated based on the evaluation result set and the evaluation matrix; in the deviation risk assessment model, the deviation risk level is determined based on the deviation risk value calculated from the audit indicator set, the audit indicator weights corresponding to the audit indicator set, and the comprehensive evaluation matrix.

[0143] In one embodiment, the device further includes:

[0144] The knowledge base construction module is used to acquire information on historical financial misconduct events; extract features from the information on historical financial misconduct events to obtain the corresponding misconduct risk features; obtain the evaluation results corresponding to the misconduct risk features, wherein the evaluation results are used to characterize the degree to which the misconduct risk features conform to the norms; obtain the credibility factor corresponding to the evaluation results; and construct a misconduct risk knowledge base based on the misconduct risk features, evaluation results, and credibility factor.

[0145] In one embodiment, the device further includes:

[0146] The alternative knowledge base construction module is used to obtain a set of financial information from the network based on preset financial keywords; classify the set of financial information to obtain a subset of financial information corresponding to each category; for each subset of financial information, select a preset number of target financial information items based on the frequency of occurrence of financial keywords in the subset; extract the stem information from the target financial information for each category to determine the alternative deviation risk characteristics; and construct an alternative deviation risk knowledge base based on the alternative deviation risk characteristics.

[0147] The various modules in the risk assessment device for the aforementioned financial business can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0148] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 4 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a risk assessment method for financial transactions.

[0149] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0150] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0151] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0152] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0153] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0154] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0155] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0156] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A risk assessment method for financial transactions, characterized in that, The method includes: Obtain the functional requirement text information corresponding to the financial business to be evaluated. The functional requirement text information is extracted from the business project document of the project application and is used to describe the functional requirements of the financial business. Keyword extraction is performed on the functional requirement text information to obtain an initial functional requirement feature set; Each initial functional requirement feature in the initial functional requirement feature set is matched with a pre-set deviation risk knowledge base. The deviation risk knowledge base is constructed based on the deviation risk features, the evaluation results corresponding to the deviation risk features, and the credibility factors corresponding to the evaluation results. The deviation risk features corresponding to historical financial deviation event information are obtained by feature extraction from the historical financial deviation event information. The evaluation results are used to characterize the degree to which the deviation risk features conform to the norms. The evaluation results and the credibility factors are determined by experts based on experience. In the initial functional requirement feature set, the initial functional requirement features that successfully match the deviation risk features are determined as target functional requirement features; The target functional requirement characteristics are input into a preset non-compliance risk assessment model. In this model, a set of audit indicators is determined based on the target functional requirement characteristics. The model also obtains an evaluation result set corresponding to the audit indicator set and an evaluation matrix related to the audit indicator set and the evaluation result set. A comprehensive evaluation matrix is ​​calculated based on the evaluation result set and the evaluation matrix. Finally, the non-compliance risk level is determined based on the non-compliance risk value calculated from the audit indicator set, the audit indicator weights corresponding to the audit indicator set, and the comprehensive evaluation matrix. This non-compliance risk level is then used as the risk assessment result for the financial business to be assessed. The non-compliance risk level includes non-compliance level, suspected non-compliance level, and compliance level. When the level of deviation risk is non-compliance or suspected non-compliance, the similarity between each initial functional requirement feature in the initial functional requirement feature set and the alternative deviation risk features in the preset alternative deviation risk knowledge base is calculated to obtain the similarity between each initial functional requirement feature and the alternative deviation risk features. The similarity between all initial functional requirement features and alternative deviation risk features is summed to obtain the total similarity. If the sum of the similarities is greater than or equal to a preset similarity threshold, a potential risk marker is added to the risk assessment result, wherein the potential risk marker is used to indicate that the financial business to be assessed has potential irregularities.

2. The method according to claim 1, characterized in that, The training process of the deviation risk assessment model includes: The characteristics of the deviation risk in the deviation risk knowledge base are used as a sample audit indicator set and input into the initial deviation risk assessment model for training to obtain the deviation risk assessment model. The deviation risk assessment model includes an evaluation matrix; the evaluation matrix is ​​obtained iteratively based on the initial evaluation matrix in the initial deviation risk assessment model; the initial evaluation matrix is ​​determined based on the credibility factor in the deviation risk knowledge base; the credibility factor is determined based on the deviation risk characteristics in the sample audit indicator set and the evaluation results in the evaluation result set.

3. The method according to claim 2, characterized in that, The method further includes: Based on preset financial keywords, obtain a collection of financial information from the internet; The financial information set is classified to obtain a subset of financial information for each category; For each category of financial information subset, a preset number of target financial information items are selected from the financial information subset based on the frequency of occurrence of the financial keywords in the financial information subset. For each category of target financial information, extract the stem information from the target financial information to determine the alternative deviation risk characteristics; Based on the characteristics of the alternative deviation risks, construct the alternative deviation risk knowledge base.

4. A risk assessment device for financial business, characterized in that, The device includes: The information acquisition module is used to acquire the functional requirement text information corresponding to the financial business to be evaluated. The functional requirement text information is extracted from the business project document of the project application and is used to describe the functional requirements of the financial business. The keyword extraction module is used to extract keywords from the functional requirement text information to obtain an initial functional requirement feature set. The feature matching module is used to match each initial functional requirement feature in the initial functional requirement feature set with the deviation risk features in the preset deviation risk knowledge base. The deviation risk knowledge base is constructed based on the deviation risk features, the evaluation results corresponding to the deviation risk features, and the credibility factors corresponding to the evaluation results. The deviation risk features corresponding to historical financial deviation event information are obtained by feature extraction from the historical financial deviation event information. The evaluation results are used to characterize the degree to which the deviation risk features conform to the norms. The evaluation results and the credibility factors are determined by experts based on experience. The feature determination module is used to determine the initial functional requirement features that successfully match the deviation risk features in the initial functional requirement feature set as target functional requirement features; The result determination module is used to determine a set of audit indicators based on the target functional requirements in the non-compliance risk assessment model; to obtain an evaluation result set corresponding to the audit indicator set and an evaluation matrix related to the audit indicator set and the evaluation result set in the non-compliance risk assessment model, and to calculate a comprehensive evaluation matrix based on the evaluation result set and the evaluation matrix; to determine the non-compliance risk level based on the non-compliance risk value calculated by the audit indicator set, the audit indicator weights corresponding to the audit indicator set, and the comprehensive evaluation matrix in the non-compliance risk assessment model, and to determine the non-compliance risk level as the risk assessment result corresponding to the financial business to be assessed, wherein the non-compliance risk level includes a non-compliance level, a suspected non-compliance level, and a compliance level; The similarity calculation module is used to calculate the similarity between each initial functional requirement feature in the initial functional requirement feature set and the alternative deviation risk features in the preset alternative deviation risk knowledge base when the deviation risk level is non-compliant or suspected non-compliant. The similarity accumulation module is used to accumulate the similarity between all the initial functional requirement features and the alternative deviation risk features to obtain the total similarity. The tagging module is used to add a potential risk tag to the risk assessment result if the sum of the similarities is greater than or equal to a preset similarity threshold, wherein the potential risk tag is used to indicate that the financial business to be assessed has potential irregularities.

5. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 3.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 3.

7. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 3.

Images