Rule generation method and device, equipment and storage medium

Abstract

The invention discloses a rule generation method and device, equipment and a storage medium, and relates to the technical field of finance. Analyzing the rule generation demand, and determining the field to which the rule belongs; the target rule generation model is called according to the rule generation demand, the target rule data is generated based on the field to which the rule belongs and the target rule generation model, and the field to which the rule belongs is determined, so that the corresponding rule data is generated by using the target rule generation model, and the method has high interpretability and is high in practicability. According to the method, the rule easy to understand is generated by depending on the target rule generation model constructed by automatic learning of historical cases instead of depending on professional experience of manual work, so that the rule development time is greatly shortened, and possible limitation and prejudice during manual rule making are overcome.

Description

Technical Field

[0001] The present application relates to the field of financial technology, and in particular to rule generation methods, devices, equipment and storage media. Background Art

[0002] In the field of financial risk management, rule generation mainly relies on data-driven methods, either through manually set rules or through rule generation models that use integrated decision trees. However, the generated rules may be too complex and difficult to understand and review, resulting in inaccurate rules and difficulty in identifying and preventing risks in a timely and accurate manner. Summary of the invention

[0003] The main purpose of this application is to provide a rule generation method, device, equipment and storage medium, aiming to solve the technical problem that the currently generated rules are not accurate enough.

[0004] To achieve the above objectives, the present application proposes a rule generation method, which includes:

[0005] Responding to rule generation requirements;

[0006] Analyze the rule generation requirements and determine the field to which the rules belong;

[0007] The target rule generation model is called according to the rule generation requirement, and the target rule data is generated based on the field to which the rule belongs and the target rule generation model.

[0008] In one embodiment, the step of calling a target rule generation model according to the rule generation requirement and generating target rule data based on the field to which the rule belongs and the target rule generation model includes:

[0009] Calling a target rule generation model according to the rule generation requirement;

[0010] Determine the corresponding domain database in the target rule generation model according to the domain to which the rule belongs;

[0011] Extracting target data from the domain database, and extracting target features of the target data;

[0012] The target rule generation model is used to generate target rule data according to the target feature.

[0013] In one embodiment, the step of extracting target data from the domain database and extracting target features of the target data includes:

[0014] Extracting target data from the domain database, classifying the target data, and constructing transaction tags;

[0015] Extracting features and constructing profiles of the target data to obtain user profiles and transaction features;

[0016] The transaction features are processed based on the transaction tags and the user portrait to obtain target features.

[0017] In one embodiment, the step of processing the transaction feature based on the transaction tag and the user portrait to obtain the target feature includes:

[0018] Determining the correlation with the transaction feature according to the transaction tag and the user portrait;

[0019] The transaction features whose correlation meets the preset correlation threshold are used as rule construction features;

[0020] The direction of the rule-constructed feature is set according to the type of the transaction tag, and the rule-constructed feature is classified and marked to obtain the target feature.

[0021] In one embodiment, the step of generating target rule data using the target rule generation model according to the target feature includes:

[0022] Generate rules using a plurality of rule generation methods in the target rule generation model according to the target features to obtain initial rule data;

[0023] Eliminating redundancy of the initial rule data to obtain reference rule data;

[0024] The reference rule data is semantically analyzed, and the reference rule data having a semantic similarity greater than a preset similarity threshold is deduplicated to obtain target rule data.

[0025] In one embodiment, before the step of calling the target rule generation model according to the rule generation requirement and generating target rule data based on the field to which the rule belongs and the target rule generation model, the step further includes:

[0026] Acquire transaction case sample data from a historical transaction database, construct a first label for case samples reported in the transaction case sample data, and construct a second label for case samples not reported in the transaction case sample data;

[0027] Analyze the sample data of the transaction cases, and construct user portraits and transaction sample features based on customer information and transaction information;

[0028] Screening the correlation between the transaction feature and the tag according to the first tag, the second tag, and the user portrait to obtain rule construction sample features;

[0029] The direction of the rule-constructed sample features is set according to the field to which the rule belongs, the rule-constructed sample features are classified and marked, and feature thresholds of different types of rule-constructed sample features are set to obtain target sample features;

[0030] A plurality of training modes are set, and the target sample features are used to train the initial rule generation model through the plurality of training modes to obtain the target rule generation model.

[0031] In one embodiment, after the step of calling the target rule generation model according to the rule generation requirement and generating target rule data based on the field to which the rule belongs and the target rule generation model, the step further includes:

[0032] Identify the current user behavior through the target rule data;

[0033] When it is identified that the current user behavior complies with the rules in the target rule data, confirming that the current user behavior is an abnormal transaction behavior and generating a warning message;

[0034] The current user behavior is monitored according to the warning information.

[0035] In addition, to achieve the above purpose, the present application also proposes a rule generation device, which includes:

[0036] A response module, for responding to rule generation requirements;

[0037] An analysis module, used to analyze the rule generation requirements and determine the field to which the rule belongs;

[0038] A generation module is used to call a target rule generation model according to the rule generation requirement, and generate target rule data based on the field to which the rule belongs and the target rule generation model.

[0039] In addition, to achieve the above objectives, the present application also proposes a rule generation device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the rule generation method described above.

[0040] In addition, to achieve the above objectives, the present application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the rule generation method described above are implemented.

[0041] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps of the rule generation method described above are implemented.

[0042] One or more technical solutions proposed in the present application respond to the rule generation demand; analyze the rule generation demand to determine the field to which the rule belongs; call the target rule generation model according to the rule generation demand, and generate target rule data based on the field to which the rule belongs and the target rule generation model. By determining the field to which the rule belongs, the target rule generation model is used to generate the corresponding rule data. It is highly interpretable and does not rely on manual professional experience. Instead, it relies on the target rule generation model constructed by automatic learning of historical cases to generate easy-to-understand rules, which greatly reduces the time for rule development and overcomes the limitations and biases that may arise when manually formulating rules. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0044] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0045] Figure 1 A flowchart of the first embodiment of the rule generation method of the present application is provided;

[0046] Figure 2 A flowchart of the second embodiment of the rule generation method of the present application is provided;

[0047] Figure 3 A schematic diagram of the framework of the rule generation method provided in Example 2 of the present application;

[0048] Figure 4 This is a schematic diagram of the module structure of the rule generation device according to an embodiment of the present application;

[0049] Figure 5 Schematic diagram of the device structure of the hardware operating environment involved in the rule generation method in the embodiment of the present application.

[0050] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0051] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0052] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0053] The main solution of the embodiment of the present application is: responding to rule generation requirements; analyzing the rule generation requirements to determine the field to which the rules belong; calling the target rule generation model according to the rule generation requirements, and generating target rule data based on the field to which the rules belong and the target rule generation model.

[0054] Since the existing technology uses an integrated decision tree rule generation model to generate rules, the current solution is highly dependent on data-driven, which makes it difficult to integrate some hidden expert knowledge or common sense in the field into the model, and conflicts or redundancies between rules may occur, which requires additional logical control or manual intervention to eliminate.

[0055] This application provides a solution and introduces a decision tree-based rule generation method that is highly interpretable and does not rely on manual professional experience. Instead, it relies on automatic learning of historical cases to generate easy-to-understand rules. This greatly reduces the time for rule development and overcomes the limitations and biases that may occur when manual rule making is in place.

[0056] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, a rule generating device, etc. The following takes the rule generating device as an example to illustrate this embodiment and the following embodiments.

[0057] Based on this, the present application embodiment provides a rule generation method, referring to Figure 1 , Figure 1 This is a flowchart of the first embodiment of the rule generation method of the present application.

[0058] In this embodiment, the rule generation method includes steps S10 to S30:

[0059] Step S10: generating requirements in response to the rules.

[0060] It should be noted that the rule generation demand can be obtained by the user through the use of the rule generation device request, and the rule generation demand can also be automatically triggered according to the monitoring situation of the user.

[0061] The application scenario of this embodiment is a financial transaction scenario, so the rules are the rules for users when conducting financial transactions. For example, financial transactions can be money laundering transactions. In the field of anti-money laundering monitoring, the current rule construction process is cumbersome and involves a lot of design, programming and verification work, resulting in a slow response to new needs and an overly long update cycle for old rules. In addition, as money laundering activities become increasingly professional, money laundering methods and their carrier forms vary. Manually summarizing money laundering characteristics through sampling is not only delayed in timeliness, but also difficult to fully cover various characteristics. It is overly dependent on business experience and difficult to identify and prevent risks in a timely and accurate manner. This embodiment uses the generated target rule model to generate corresponding rules in the field of money laundering, so that user transaction behaviors can be monitored according to the rules to determine whether the user has money laundering behavior.

[0062] Step S20: Analyze the rule generation requirements to determine the field to which the rule belongs.

[0063] In the specific implementation, different rule requirements can generate rules in different fields. Therefore, the key fields in the rule generation requirements can be determined by analyzing the rule generation requirements, and then the fields to which the rules to be generated belong can be determined based on the key fields. The fields to which the rules belong can be anti-money laundering, credit, insurance pricing, etc. Different fields have different data to be extracted. Therefore, the corresponding data can be quickly extracted according to the fields to which the rules belong, thereby improving the efficiency of rule generation.

[0064] Step S30: calling a target rule generation model according to the rule generation requirement, and generating target rule data based on the field to which the rule belongs and the target rule generation model.

[0065] In specific implementation, the target rule generation model can be called according to the rule generation requirements. For example, if the field to which the rule belongs is the anti-money laundering field, the target rule generation model is the anti-money laundering rule generation model, and the corresponding rule data can be generated through the field to which the rule belongs and the anti-money laundering rule generation model. The target rule data is the specific rule of the field to which the rule belongs.

[0066] The target rule generation model can be obtained by collecting corresponding sample data in advance to train the machine learning model, and the machine learning model is used to mine and analyze the money laundering behaviors in the historical money laundering case database to automatically generate and launch anti-money laundering rules. This shortens the development cycle of anti-money laundering rules and improves the effectiveness of anti-money laundering rules.

[0067] For example, if the field to which the rule belongs is the field of anti-money laundering, then the target rule data may be the rules defined for the user's money laundering behavior.

[0068] In a feasible implementation manner, step S30 may include:

[0069] The current user behavior is identified through the target rule data; when it is identified that the current user behavior complies with the rules in the target rule data, the current user behavior is confirmed to be an abnormal transaction behavior and a warning message is generated; the current user behavior is monitored according to the warning message.

[0070] It should be noted that after the target rule data is generated, the current user behavior can be identified based on the target rule data, and user information can be obtained to determine whether the user information and current user behavior comply with the rules in the target rule data, thereby determining whether the user behavior is abnormal.

[0071] In a specific implementation, for example, if the target rule data includes "the transaction time is early morning", the current user behavior can be identified to determine whether the user's current behavior is consistent with the transaction time being early morning. If the user's current behavior is consistent with the transaction time being early morning, the user behavior is determined to be abnormal.

[0072] It should be understood that when it is determined that a user has abnormal trading behavior, a warning message may be generated and corresponding monitoring may be performed based on the warning message. For example, the user's recent transaction data may be obtained through the warning message for further analysis, or the user's account may be managed based on the warning message, thereby performing risk control on the user's current behavior.

[0073] The present embodiment provides a rule generation method, which responds to rule generation requirements; analyzes the rule generation requirements to determine the field to which the rule belongs; calls a target rule generation model according to the rule generation requirements, and generates target rule data based on the field to which the rule belongs and the target rule generation model. By determining the field to which the rule belongs, the target rule generation model is used to generate corresponding rule data. The method has high interpretability and does not rely on manual professional experience. Instead, it relies on the target rule generation model constructed by automatic learning of historical cases to generate easy-to-understand rules, which greatly reduces the time for rule development and overcomes the limitations and biases that may occur when manual rule making is performed.

[0074] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can be referred to the above introduction, and will not be repeated in the following. Figure 2 , step S30 includes steps S301 to S304:

[0075] Step S301: calling a target rule generation model according to the rule generation requirement.

[0076] It should be noted that the corresponding rule generation model can be called according to the specific rule generation requirements, and different types of rule generation models can be stored in the database, so that when rule generation is required, different rule generation models can be called according to the rule generation requirements. For example, if the rule generation requirement is to generate anti-money laundering rules, the target rule generation model is the anti-money laundering rule generation model.

[0077] As another example, only one target rule generation model may be trained and generated, and this target rule generation model may generate different rules according to different rule generation requirements.

[0078] Step S302: Determine a corresponding domain database in the target rule generation model according to the domain to which the rule belongs.

[0079] It is understandable that the corresponding domain database can be determined in the target rule generation model according to the domain to which the rule belongs, the target rule generation model can store domain databases of different rules, and the corresponding domain database can also be called through an interface.

[0080] The domain database stores a large amount of data in the corresponding fields. For example, if the field to which the rules belong is the anti-money laundering field, then the domain database is the money laundering case database.

[0081] Step S303: extracting target data from the domain database, and extracting target features of the target data.

[0082] In a specific implementation, corresponding data may be extracted from a domain database. For example, if the domain database is a money laundering case database, the target data may include reported case data and unreported case data.

[0083] In a specific implementation, the features of the target data can be extracted through certain feature extraction rules, thereby obtaining target features, which are features that best represent the rules.

[0084] In a feasible implementation, step S303 may include steps A10 to A30:

[0085] Step A10: extract target data from the domain database, classify the target data, and construct transaction tags.

[0086] It should be noted that the corresponding target data can be obtained from the domain database, and the corresponding transaction label can be constructed according to the type of the target data. For example, if the target data is reported case data, the transaction label is set to 1; if the target data is unreported case data, the transaction label is set to 0. By constructing the transaction label, the characteristics of the target data can be screened later.

[0087] Step A20: Extract features and construct a profile of the target data to obtain user profiles and transaction features.

[0088] It is understandable that feature extraction and user portrait construction can be performed on the customer and transaction information involved in the target data extracted above. The user portrait is constructed based on the user's behavior and preferences, which may include the user's age, gender, geographic location, consumption habits, etc.

[0089] Feature extraction can specifically analyze transaction data to extract features such as transaction amount, transaction frequency, transaction time, transaction location, etc. Abnormal transaction patterns can be identified, such as sudden large transactions or unusual transaction times, to obtain transaction features.

[0090] Step A30: Process the transaction features based on the transaction tags and the user portrait to obtain target features.

[0091] In a specific implementation, there may be a large number of useless features or features irrelevant to rule generation in the transaction features. Therefore, the transaction features may be processed according to the transaction tags and user portraits, such as screening and adding tags, to obtain the target features.

[0092] In a feasible implementation, step A30 may include: determining the correlation with the transaction feature based on the transaction tag and the user portrait; using the transaction feature whose correlation meets a preset correlation threshold as a rule construction feature; setting the direction of the rule construction feature according to the type of the transaction tag, and classifying and marking the rule construction feature to obtain the target feature.

[0093] It should be noted that the correlation between transaction labels and user portraits and transaction features can be determined based on transaction labels and user portraits, so as to screen transaction features and retain only highly correlated features. A preset correlation threshold can be set in advance to retain transaction features with correlations greater than or equal to the preset correlation threshold, and remove transaction features with correlations less than the preset correlation threshold, thereby obtaining rule construction features, that is, features used for rule construction. The rule construction features have a strong correlation with the corresponding scenarios, such as anti-money laundering scenarios, thereby improving the interpretability of the rules.

[0094] In the specific implementation, given that the feature direction in the existing rule generation scheme may not meet expectations, for example, when generating the rule of "the number of transactions is less than 20", in the field of anti-money laundering, the direction of the number of transactions "greater than" may be more explanatory. Therefore, the direction of the rule construction feature can be set according to the type of transaction label. Specifically, the feature direction can be set according to the relative center position of the transaction label to ensure that the generated rules are more in line with common sense and intuitive.

[0095] After setting the direction of the rule construction features, they can also be classified and labeled. Specifically, they can be divided into counting type, category type, amount type, time type and other categories, and appropriate threshold smoothing strategies can be adopted for different types of features. For example, for time type features, 1 minute, 5 minutes, 30 minutes, etc. are selected as thresholds instead of 57 seconds or 345 seconds. This not only improves the readability of the features, but also enhances the generalization ability of the rules, thereby obtaining the target features.

[0096] Step S304: Generate target rule data using the target rule generation model according to the target features.

[0097] It can be understood that after the target features are screened and processed, the target features can be processed using the target rule generation model to generate target rule data. The target rule generation model reveals the internal rules of the data by learning a large amount of data, thereby generating corresponding target rule data based on the target features.

[0098] In a feasible implementation, step S304 may include steps B10 to B30:

[0099] Step B10: Generate rules using a plurality of rule generation methods in the target rule generation model according to the target features to obtain initial rule data.

[0100] It should be noted that the target rule generation model is trained through a variety of training modes. Therefore, after the target features are input into the target rule generation model, rules can be generated through a variety of rule generation methods in the target rule generation model. For example, the boosting learning mode focuses on misclassified points, the sampling learning mode samples positive and negative features according to a certain probability and proportion, and the sequential learning mode excludes covered feature points during the training process to ensure the diversity of generated rules and obtain initial rule data.

[0101] It should be understood that after the initial rule data is generated, the initial rule data can also be analyzed, for example, by using lasso regression (Least absolute shrinkage and selection operator) to select different types of rules to improve the uniqueness of the rules and automatically adjust the type of generated rules according to the analysis results.

[0102] Step B20: Eliminate redundancy from the initial rule data to obtain reference rule data.

[0103] In a specific implementation, after the initial rule data is generated, the initial rule data may contain redundant expressions. For example, the rule "Is it old = True and age>70 years old" can be optimized to "age>70 years old". The initial rule data can be redundantly eliminated, and the reference rule data can be obtained by eliminating the inefficient or invalid expression parts in the rule to ensure the simplicity of the rule expression. The reference rule data is the data without redundant rules.

[0104] Step B30: semantic analysis is performed on the reference rule data, and reference rule data with semantic similarity greater than a preset similarity threshold is deduplicated to obtain target rule data.

[0105] In a batch of rules generated by the target rule generation model, there may be high semantic similarity between the rules. For example, Rule 1 "Age>60 years old and whether the transaction time is late at night = True and the number of counterparties > 100" and Rule 2 "Age>60 years old and whether the transaction time is late at night = True and the transaction amount > 10,000" both contain the same features "age" and "whether the transaction time is late at night". Therefore, semantic analysis can be performed on the reference rule data, the semantic similarity between each reference rule can be calculated, and a preset similarity threshold, such as 0.8, 0.9, etc., can be set to determine that the reference rule data whose semantic similarity between each reference rule is greater than the preset similarity threshold is deduplicated to prevent excessive duplication of rules.

[0106] In the specific implementation, relatively poorly performing rules can also be filtered out to obtain target rule data. The generated rule, such as "a>10and b<30", consists of three elements: feature ("A", "B"), comparison operator (">", "<") and feature value ("10", "30"). In order to improve the interpretability of the feature level, we calculate the importance of each feature in the rule and eliminate the unimportant parts.

[0107] In some embodiments, in order to improve the efficiency and diversity of rule generation, sample data may be collected in advance to train and generate a target rule generation model. The steps of generating the target rule generation model include:

[0108] Acquire transaction case sample data from a historical transaction database, construct a first label for case samples reported in the transaction case sample data, and construct a second label for case samples not reported in the transaction case sample data;

[0109] Analyze the sample data of the transaction cases, and construct user portraits and transaction sample features based on customer information and transaction information;

[0110] Screening the correlation between the transaction feature and the tag according to the first tag, the second tag, and the user portrait to obtain rule construction sample features;

[0111] The direction of the rule-constructed sample features is set according to the field to which the rule belongs, the rule-constructed sample features are classified and marked, and feature thresholds of different types of rule-constructed sample features are set to obtain target sample features;

[0112] A plurality of training modes are set, and the target sample features are used to train the initial rule generation model through the plurality of training modes to obtain the target rule generation model.

[0113] It should be noted that the target rule generation model can be an anti-money laundering rule generation model, which can extract data from the historical money laundering case database, that is, transaction case sample data. For the reported case sample data, a money laundering label 1 is assigned to indicate that it is a black sample. For the unreported case sample data, a money laundering label 0 is assigned to indicate that it is a white sample.

[0114] In specific implementation, the collected transaction case sample data can be analyzed to extract and construct money laundering features and user portraits based on customer information and transaction information, including collecting the customer's basic Western Sydney and constructing the customer's historical transaction data.

[0115] In the specific implementation, the correlation between labels, user portraits and transaction features can be used for screening, and only highly correlated features are retained to obtain rule-constructed sample features. For different fields, the feature direction must meet the field requirements. For example, in the field of anti-money laundering, the direction of the number of transactions should be greater than a certain number, which may be more explanatory. Therefore, the direction of the rule-constructed sample features can be set according to the field to which the rules belong. Specifically, the feature direction can be set according to the relative center position of the black and white samples to ensure that the generated rules are more in line with common sense. At the same time, the rule-constructed sample features are classified and marked into counting type, category type, amount type, time type and other categories, and appropriate threshold smoothing strategies are adopted for different types of features. The feature thresholds of different types of rule-constructed sample features are set. For example, for time-type features, 1 minute, 5 minutes, 30 minutes, etc. are selected as thresholds instead of 57 seconds or 345 seconds. This not only improves the readability of the features, but also enhances the generalization ability of the rules.

[0116] It should be noted that if Figure 3 As shown, Figure 3The schematic diagram of the framework of the rule generation method is divided into a feature layer, a model layer, and a rule optimization layer. The feature layer includes feature screening, feature direction determination, and feature threshold optimization. The cornerstone of the model layer is a decision tree based on information gain. The decision tree model is based on data and reveals the internal laws of the data by learning a large amount of data. Therefore, for different anti-money laundering scenarios, multiple training modes are provided. The boosting learning mode focuses on misclassified points; the sampling learning mode samples positive and negative samples according to a certain probability and proportion; the sequential learning mode excludes covered sample points during the training process, so that the initial rule generation model is trained using the target sample features through multiple training modes to obtain the target rule generation model. At the same time, after obtaining the target rule generation model, it is also allowed to adjust the weights of misclassified and uncovered samples according to the latest generated rules, and reduce the weights of correctly classified and covered samples, so that the model can focus more on those misclassified samples and improve the training effect of the model. Different anti-money laundering scenarios have different requirements for the complexity of rules (depth of rules), ability preference (biased towards recall or precision) and rule composition (the presence of specific features). By setting different training parameters, the present invention can customize the rules to meet the needs of different scenarios.

[0117] Continue as Figure 3 As shown in the figure, the rule optimization layer may include post-pruning of rules, deduplication based on semantic similarity, deduplication based on distribution similarity, and calculation of feature importance, and perform post-pruning optimization, deduplication between semantic rules, and deduplication between distribution rules on the same batch of rules, so as to ensure the simplicity, robustness, and differentiation of the rules launched.

[0118] This embodiment calls the target rule generation model according to the rule generation requirements; determines the corresponding domain database in the target rule generation model according to the domain to which the rule belongs; extracts target data from the domain database, and extracts target features of the target data; generates target rule data using the target rule generation model according to the target features. The corresponding target rule generation model can be quickly called according to the domain to which the rule belongs, so that the corresponding data can be extracted and the target rule data can be quickly generated. Automated rules can help make decisions quickly, reduce manual intervention, and improve efficiency.

[0119] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the rule generation method of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0120] This application also provides a rule generation device, please refer to Figure 4 , the rule generating device comprises:

[0121] The response module 10 is used to respond to the rule generation requirement.

[0122] The analysis module 20 is used to analyze the rule generation requirements and determine the field to which the rule belongs.

[0123] The generation module 30 is used to call the target rule generation model according to the rule generation requirement, and generate target rule data based on the field to which the rule belongs and the target rule generation model.

[0124] The rule generation device provided by the present application adopts the rule generation method in the above embodiment, which can solve the technical problem that the currently generated rules are not accurate enough. Compared with the prior art, the beneficial effects of the rule generation device provided by the present application are the same as the beneficial effects of the rule generation method provided by the above embodiment, and other technical features in the rule generation device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0125] In one embodiment, the generation module 30 is also used to call the target rule generation model according to the rule generation requirements; determine the corresponding domain database in the target rule generation model according to the domain to which the rule belongs; extract target data from the domain database, and extract target features of the target data; and use the target rule generation model to generate target rule data according to the target features.

[0126] In one embodiment, the generation module 30 is further used to extract target data from the domain database, classify the target data, and construct transaction tags; perform feature extraction and profile construction on the target data to obtain user profiles and transaction features; and process the transaction features based on the transaction tags and the user profiles to obtain target features.

[0127] In one embodiment, the generation module 30 is further used to determine the correlation with the transaction feature based on the transaction tag and the user portrait; use the transaction feature whose correlation meets a preset correlation threshold as a rule construction feature; set the direction of the rule construction feature according to the type of the transaction tag, and classify and mark the rule construction feature to obtain the target feature.

[0128] In one embodiment, the generation module 30 is also used to generate rules according to the target features using a plurality of rule generation methods in the target rule generation model to obtain initial rule data; to eliminate redundancy in the initial rule data to obtain reference rule data; to perform semantic analysis on the reference rule data, to deduplicate reference rule data having a semantic similarity greater than a preset similarity threshold, to obtain target rule data.

[0129] In one embodiment, the generation module 30 is also used to obtain transaction case sample data from a historical transaction database, and construct a first label for the case samples reported in the transaction case sample data, and construct a second label for the case samples not reported in the transaction case sample data; analyze the transaction case sample data, and construct a user portrait and transaction sample features based on customer information and transaction information; screen the correlation between the transaction features and the labels based on the first label, the second label and the user portrait to obtain rule-constructed sample features; set the direction of the rule-constructed sample features according to the field to which the rules belong, classify and mark the rule-constructed sample features, and set feature thresholds for different types of rule-constructed sample features to obtain target sample features; set multiple training modes, and use the target sample features to train the initial rule generation model through the multiple training modes to obtain the target rule generation model.

[0130] In one embodiment, the generation module 30 is also used to identify the current user behavior through the target rule data; when it is identified that the current user behavior complies with the rules in the target rule data, confirm that the current user behavior is an abnormal transaction behavior and generate warning information; monitor the current user behavior according to the warning information.

[0131] The present application provides a rule generation device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the rule generation method in the above-mentioned embodiment 1.

[0132] Reference below Figure 5 , which shows a schematic diagram of the structure of a rule generation device suitable for implementing the embodiment of the present application. The rule generation device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 5 The rule generation device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0133] like Figure 5As shown, the rule generation device may include a processing device 1001 (e.g., a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the rule generation device are also stored. The processing device 1001, ROM1002, and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the rule generation device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a rule generation device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have alternatively.

[0134] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0135] The rule generation device provided by the present application adopts the rule generation method in the above embodiment, which can solve the technical problem that the currently generated rules are not accurate enough. Compared with the prior art, the beneficial effects of the rule generation device provided by the present application are the same as the beneficial effects of the rule generation method provided by the above embodiment, and other technical features in the rule generation device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0136] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0137] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0138] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the rule generation method in the above-mentioned embodiment.

[0139] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0140] The computer-readable storage medium may be included in the rule generating device; or may exist independently without being assembled into the rule generating device.

[0141] The above-mentioned computer-readable storage medium carries one or more programs. When the above-mentioned one or more programs are executed by the rule generation device, the rule generation device: responds to the rule generation demand; analyzes the rule generation demand and determines the field to which the rule belongs; calls the target rule generation model according to the rule generation demand, and generates target rule data based on the field to which the rule belongs and the target rule generation model.

[0142] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0143] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession 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 square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0144] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0145] The readable storage medium provided by the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned rule generation method, and can solve the technical problem that the currently generated rules are not accurate enough. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by the present application are the same as the beneficial effects of the rule generation method provided by the above-mentioned embodiment, and will not be repeated here.

[0146] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned rule generation method when executed by a processor.

[0147] The computer program product provided by this application can solve the technical problem that the currently generated rules are not accurate enough. Compared with the prior art, the beneficial effects of the computer program product provided by this application are the same as the beneficial effects of the rule generation method provided by the above embodiment, which will not be repeated here.

[0148] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A rule generation method, characterized in that: The rule generation method comprises: Responding to rule generation requirements; Analyze the rule generation requirements and determine the field to which the rules belong; The target rule generation model is called according to the rule generation requirement, and the target rule data is generated based on the field to which the rule belongs and the target rule generation model.

2. The method according to claim 1, characterized in that The step of calling the target rule generation model according to the rule generation requirement and generating target rule data based on the field to which the rule belongs and the target rule generation model includes: Calling a target rule generation model according to the rule generation requirement; Determine the corresponding domain database in the target rule generation model according to the domain to which the rule belongs; Extracting target data from the domain database, and extracting target features of the target data; The target rule generation model is used to generate target rule data according to the target feature.

3. The method according to claim 2, characterized in that The step of extracting target data from the domain database and extracting target features of the target data comprises: Extracting target data from the domain database, classifying the target data, and constructing transaction tags; Extracting features and constructing profiles of the target data to obtain user profiles and transaction features; The transaction features are processed based on the transaction tags and the user portrait to obtain target features.

4. The method according to claim 3, characterized in that The step of processing the transaction feature based on the transaction tag and the user portrait to obtain the target feature includes: Determining the correlation with the transaction feature according to the transaction tag and the user portrait; The transaction features whose correlation meets the preset correlation threshold are used as rule construction features; The direction of the rule-constructed feature is set according to the type of the transaction tag, and the rule-constructed feature is classified and marked to obtain the target feature.

5. The method according to claim 2, characterized in that The step of generating target rule data using the target rule generation model according to the target feature comprises: Generate rules using a plurality of rule generation methods in the target rule generation model according to the target features to obtain initial rule data; Eliminating redundancy of the initial rule data to obtain reference rule data; The reference rule data is semantically analyzed, and the reference rule data having a semantic similarity greater than a preset similarity threshold is deduplicated to obtain target rule data.

6. The method according to claim 1, characterized in that Before the step of calling the target rule generation model according to the rule generation requirement and generating target rule data based on the field to which the rule belongs and the target rule generation model, the step further includes: Acquire transaction case sample data from a historical transaction database, construct a first label for case samples reported in the transaction case sample data, and construct a second label for case samples not reported in the transaction case sample data; Analyze the sample data of the transaction cases, and construct user portraits and transaction sample features based on customer information and transaction information; Screening the correlation between the transaction feature and the tag according to the first tag, the second tag, and the user portrait to obtain rule construction sample features; The direction of the rule-constructed sample features is set according to the field to which the rule belongs, the rule-constructed sample features are classified and marked, and feature thresholds of different types of rule-constructed sample features are set to obtain target sample features; A plurality of training modes are set, and the target sample features are used to train the initial rule generation model through the plurality of training modes to obtain the target rule generation model.

7. The method according to any one of claims 1 to 6, characterized in that After the step of calling the target rule generation model according to the rule generation requirement and generating target rule data based on the field to which the rule belongs and the target rule generation model, the method further includes: Identify the current user behavior through the target rule data; When it is identified that the current user behavior complies with the rules in the target rule data, confirming that the current user behavior is an abnormal transaction behavior and generating a warning message; The current user behavior is monitored according to the warning information.

8. A rule generating device, characterized in that: The device comprises: A response module, for responding to rule generation requirements; An analysis module, used to analyze the rule generation requirements and determine the field to which the rule belongs; A generation module is used to call a target rule generation model according to the rule generation requirement, and generate target rule data based on the field to which the rule belongs and the target rule generation model.

9. A rule generation device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the rule generation method according to any one of claims 1 to 7.

10. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the rule generation method according to any one of claims 1 to 7 are implemented.

Images