Training method, device, equipment and readable storage medium for state recognition model

Abstract

The present disclosure discloses a method, apparatus, device, and readable storage medium for training a state recognition model. The state recognition model training method includes: obtaining feature data related to a subject and state data related to the subject; determining specific attribute values ​​that match the feature data based on a preset mapping relationship; determining a standardized state index for the subject based on the state data related to the subject and the specific attribute values; and training the state recognition model based on the standardized state index, thereby converting the state data into a standardized state index. The state recognition model trained based on the standardized state index can more accurately identify the subject's state.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular to a training method, device, electronic device, and readable storage medium for a state recognition model. Background Art

[0002] Currently, when identifying a certain state of a subject, one may focus on the characteristic data of the subject itself and the state data related to the subject. However, the inventors have found that due to the complexity of the actual situation, the above content cannot well reflect the true state of the subject, and inaccurate state recognition results may easily occur.

[0003] For example, when identifying corporate risks, one typically focuses on litigation risk. However, while larger companies are more likely to be involved in lawsuits, this doesn't necessarily mean they have a high litigation risk. Conversely, smaller companies may not be involved in many lawsuits, but compared to similar companies, this may indicate a higher litigation risk. Therefore, it's difficult for users to intuitively determine a company's litigation risk based solely on the number of defendant records. Summary of the Invention

[0004] In order to solve the problems in the related art, the embodiments of the present disclosure provide a training method and device for a state recognition model, a state recognition method and device, a device and a medium.

[0005] In a first aspect, an embodiment of the present disclosure provides a method for training a state recognition model.

[0006] Specifically, the training method of the state recognition model includes:

[0007] obtaining characteristic data associated with a subject, and status data associated with the subject;

[0008] Determining a specific attribute value that matches the feature data according to a preset mapping relationship;

[0009] determining a standardized status indicator of the subject based on the status data associated with the subject and the specific attribute value;

[0010] The state recognition model is trained according to the standardized state indicator.

[0011] In conjunction with the first aspect, in a first implementation of the first aspect of the present disclosure, the mapping relationship includes at least:

[0012] The correspondence between feature data and specific attribute values; wherein, feature data with different value ranges correspond to different feature attribute values.

[0013] In combination with the first aspect, in a second implementation of the first aspect of the present disclosure, determining the standardized status indicator of the subject based on the status data related to the subject and the specific attribute value includes:

[0014] determining an original state indicator based on the state data and the specific attribute value;

[0015] According to the maximum and minimum values ​​of the original state indicators of different subjects, the original state indicators are mapped to a predetermined value range to obtain a standardized state indicator.

[0016] In combination with the first aspect, in a third implementation manner of the first aspect of the present disclosure, the feature data includes graph features determined based on association relationships between the subjects in a subject relationship graph.

[0017] In combination with the third implementation manner of the first aspect, in a fourth implementation manner of the first aspect of the present disclosure, training the state recognition model according to the standardized state indicator includes:

[0018] determining a first vector representation of characteristic data of the subject;

[0019] Converting the first vector representation from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation;

[0020] concatenating the second vector representation with a third vector representation of feature data of a neighboring subject of the subject in the subject relationship graph to obtain a fourth vector representation of the feature data of the subject;

[0021] encoding the fourth vector representation into a fifth vector representation of feature data of the subject having a predetermined dimension as training data for the state recognition model;

[0022] The state recognition model is trained using the standardized state indicators as labels.

[0023] In combination with the fourth implementation manner of the first aspect, in a fifth implementation manner of the first aspect of the present disclosure, the first vector representation includes a one-hot feature representation, and converting the first vector representation from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation includes:

[0024] Converting the one-hot feature representation into a sixth vector representation through an embedding operation;

[0025] The low-dimensional dense second vector representation is obtained by weighted summing of the sixth vector representations.

[0026] In combination with the first aspect and any one of the first to fifth implementations of the first aspect, in a sixth implementation of the first aspect of the present disclosure, the method further includes: presenting the standardized status indicator in a graphical manner.

[0027] In a second aspect, an embodiment of the present disclosure provides a state identification method.

[0028] Specifically, the state identification method includes:

[0029] Obtaining feature data of the subject to be identified;

[0030] The feature data is input into the state recognition model as described in any one of the first aspect and the first to sixth aspects of the first aspect to obtain a standardized state index of the subject to be recognized.

[0031] In a third aspect, an embodiment of the present disclosure provides a method for training a risk identification model.

[0032] Specifically, the risk identification model training method includes:

[0033] Obtaining characteristic data of an enterprise and the number of contract litigation cases in which the enterprise is a defendant, wherein the characteristic data includes turnover;

[0034] Determining the number of contracts for the enterprise based on the mapping relationship between the turnover and the number of contracts;

[0035] Determining a standardized risk index for the enterprise based on the number of contract litigation cases and the number of contracts of the enterprise;

[0036] Training data is determined based on the characteristic data of the enterprise, and a risk identification model is trained using the standardized risk indicators as labels.

[0037] In a fourth aspect, an embodiment of the present disclosure provides a training device for a state recognition model.

[0038] Specifically, the training device of the state recognition model includes:

[0039] A first acquisition module is configured to obtain feature data related to a subject and status data related to the subject;

[0040] A first determining module is configured to determine a specific attribute value matching the feature data according to a preset mapping relationship;

[0041] a second determining module configured to determine a standardized status indicator of the subject based on the status data related to the subject and the specific attribute value;

[0042] A training module is configured to train the state recognition model according to the standardized state indicator.

[0043] In conjunction with the fourth aspect, in a first implementation of the fourth aspect of the present disclosure, the mapping relationship includes at least:

[0044] The correspondence between feature data and specific attribute values, where feature data with different value ranges correspond to different feature attribute values.

[0045] In conjunction with the fourth aspect, in a second implementation of the fourth aspect of the present disclosure, determining the standardized status indicator of the subject based on the status data related to the subject and the specific attribute value includes:

[0046] determining an original state indicator based on the state data and the specific attribute value;

[0047] According to the maximum and minimum values ​​of the original state indicators of different subjects, the original state indicators are mapped to a predetermined value range to obtain a standardized state indicator.

[0048] In combination with the fourth aspect, in a third implementation manner of the fourth aspect of the present disclosure, the feature data includes graph features determined based on the association relationship of the subjects in the subject relationship graph.

[0049] In combination with the third implementation manner of the fourth aspect, in a fourth implementation manner of the fourth aspect of the present disclosure, the training of the state recognition model according to the standardized state indicator includes:

[0050] determining a first vector representation of characteristic data of the subject;

[0051] Converting the first vector representation from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation;

[0052] concatenating the second vector representation with a third vector representation of feature data of a neighboring subject of the subject in the subject relationship graph to obtain a fourth vector representation of the feature data of the subject;

[0053] encoding the fourth vector representation into a fifth vector representation of feature data of the subject having a predetermined dimension as training data for the state recognition model;

[0054] The state recognition model is trained using the standardized state indicators as labels.

[0055] In combination with the fourth implementation manner of the fourth aspect, in a fifth implementation manner of the fourth aspect of the present disclosure, the first vector representation includes a one-hot feature representation, and converting the first vector representation from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation includes:

[0056] Converting the one-hot feature representation into a sixth vector representation through an embedding operation;

[0057] The low-dimensional dense second vector representation is obtained by weighted summing of the sixth vector representations.

[0058] In combination with the fourth aspect and any one of the first to fifth implementations of the fourth aspect, in a sixth implementation of the fourth aspect of the present disclosure, the device further includes:

[0059] The display module is configured to present the standardized status indicator in a graphical manner.

[0060] In a fifth aspect, a state recognition device is provided in an embodiment of the present disclosure.

[0061] Specifically, the state recognition device includes:

[0062] A second acquisition module is configured to acquire feature data of a subject to be identified;

[0063] The identification module is configured to input the feature data into a state identification model as described in any one of the first aspect and the first to sixth implementation methods of the first aspect to obtain a standardized state indicator of the subject to be identified.

[0064] In a sixth aspect, an embodiment of the present disclosure provides an electronic device comprising a memory and a processor, wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement a method as described in any one of the first aspect, the first to sixth implementations of the first aspect, the second aspect, or the third aspect.

[0065] In the seventh aspect, a computer-readable storage medium is provided in an embodiment of the present disclosure, on which computer instructions are stored. When the computer instructions are executed by a processor, a method as described in any one of the first aspect, the first to sixth implementation methods of the first aspect, the second aspect or the third aspect is implemented.

[0066] According to the technical solution provided by the embodiment of the present disclosure, feature data related to a subject and status data related to the subject are obtained; specific attribute values ​​matching the feature data are determined according to a preset mapping relationship; standardized status indicators of the subject are determined based on the status data related to the subject and the specific attribute values; the status recognition model is trained according to the standardized status indicators, so that the status data can be converted into standardized status indicators. The status recognition model trained based on the standardized status indicators can more accurately identify the status of the subject.

[0067] According to the technical solution provided by the embodiment of the present disclosure, a mapping relationship includes at least a correspondence between feature data and specific attribute values, wherein feature data with different value ranges correspond to different feature attribute values, so that the number of specific events can be converted into a standardized state indicator based on the mapping relationship, and the state recognition model trained based on the standardized state indicator can more accurately identify the state of the subject.

[0068] According to the technical solution provided by the embodiment of the present disclosure, the standardized state index of the subject is determined based on the state data related to the subject and the specific attribute value, including: determining the original state index based on the state data and the specific attribute value; mapping the original state index to a predetermined value range according to the maximum and minimum values ​​of the original state indicators of different subjects to obtain a standardized state index, so that the obtained standardized state index is more effective, and the state recognition model trained based on the standardized state index can more accurately identify the state of the subject.

[0069] According to the technical solution provided by the embodiment of the present disclosure, the feature data includes graph features determined based on the association relationship of the subject in the subject relationship graph, so that the status of the subject can be better identified in combination with the graph features.

[0070] According to the technical solution provided by the embodiment of the present disclosure, a first vector representation of the feature data of the subject is determined; the first vector representation is converted from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation; the second vector representation is spliced ​​with the third vector representation of the feature data of the adjacent subjects of the subject in the subject relationship graph to obtain a fourth vector representation of the feature data of the subject; the fourth vector representation is encoded into a fifth vector representation of the feature data of the subject with a predetermined dimension as training data for the state recognition model; the state recognition model is trained with the standardized state indicator as a label, so as to better utilize the graph features to identify the state of the subject.

[0071] According to the technical solution provided by the embodiment of the present disclosure, the one-hot feature representation is converted into a sixth vector representation through an embedding operation; and the sixth vector representation is weighted summed to obtain the low-dimensional dense second vector representation, thereby solving the problem of excessive resource occupation caused by sparse features.

[0072] According to the technical solution provided by the embodiment of the present disclosure, the state of the subject can be more intuitively displayed by presenting the standardized state indicators in a graphical manner.

[0073] According to the technical solution provided by the embodiment of the present disclosure, by obtaining the characteristic data of the subject to be identified, the characteristic data is input into a state recognition model trained by the training method of the above-mentioned state recognition model to obtain the standardized state index of the subject to be identified, so that the state of the subject can be more accurately identified through the standardized state index.

[0074] According to the technical solution provided by the embodiments of the present disclosure, characteristic data of an enterprise and the number of contract litigation cases in which the enterprise is the defendant are obtained, wherein the characteristic data includes turnover; the number of contracts of the enterprise is determined based on the mapping relationship between the turnover and the number of contracts; the standardized risk index of the enterprise is determined based on the number of contract litigation cases and the number of contracts of the enterprise; training data is determined based on the characteristic data of the enterprise, and a risk identification model is trained with the standardized risk index as a label, so that the number of contract litigation cases can be converted into a standardized risk index. The risk identification model trained based on the standardized risk index can more accurately identify the risks of the enterprise.

[0075] According to the technical solution provided by the embodiment of the present disclosure, a first acquisition module is configured to obtain feature data related to a subject and status data related to the subject; a first determination module is configured to determine a specific attribute value that matches the feature data according to a preset mapping relationship; a second determination module is configured to determine a standardized status indicator of the subject based on the status data related to the subject and the specific attribute value; a training module is configured to train the status recognition model according to the standardized status indicator, so that the number of specific events can be converted into a standardized status indicator, and the status recognition model trained based on the standardized status indicator can more accurately identify the status of the subject.

[0076] According to the technical solution provided by the embodiment of the present disclosure, the mapping relationship includes at least a correspondence between feature data and specific attribute values, wherein feature data with different value ranges correspond to different feature attribute values, so that the number of specific events can be converted into a standardized state indicator based on the mapping relationship, and the state recognition model trained based on the standardized state indicator can more accurately identify the state of the subject.

[0077] According to the technical solution provided by the embodiment of the present disclosure, the original state indicator is determined based on the state data and the specific attribute value; according to the maximum and minimum values ​​of the original state indicators of different subjects, the original state indicator is mapped to a predetermined value range to obtain a standardized state indicator, so that the obtained standardized state indicator is more effective, and the state recognition model trained based on the standardized state indicator can more accurately identify the state of the subject.

[0078] According to the technical solution provided by the embodiment of the present disclosure, the graph features determined based on the association relationship of the subjects in the subject relationship graph are used as feature data, so that the status of the subject can be better identified in combination with the graph features.

[0079] According to the technical solution provided by the embodiment of the present disclosure, a first vector representation of the feature data of the subject is determined; the first vector representation is converted from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation; the second vector representation is spliced ​​with the third vector representation of the feature data of the adjacent subjects of the subject in the subject relationship graph to obtain a fourth vector representation of the feature data of the subject; the fourth vector representation is encoded into a fifth vector representation of the feature data of the subject with a predetermined dimension as training data for the state recognition model; the state recognition model is trained with the standardized state indicator as a label, so as to better utilize the graph features to identify the state of the subject.

[0080] According to the technical solution provided by the embodiment of the present disclosure, the first vector representation of the one-hot feature representation is converted into a sixth vector representation through an embedding operation; the weighted sum of the sixth vector representation is performed to obtain the low-dimensional dense second vector representation, thereby solving the problem of excessive resource occupation caused by sparse features.

[0081] According to the technical solution provided by the embodiment of the present disclosure, the display module is configured to present the standardized status indicators in a graphical manner, so that the status of the subject can be displayed more intuitively.

[0082] According to the technical solution provided by the embodiment of the present disclosure, the second acquisition module is configured to obtain feature data of the subject to be identified; the identification module is configured to input the feature data into the state recognition model trained by the training method of the state recognition model described above, so as to obtain the standardized state index of the subject to be identified, so that the state of the subject can be more accurately identified through the standardized state index.

[0083] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] Other features, objectives and advantages of the present disclosure will become more apparent through the following detailed description of non-limiting embodiments in conjunction with the accompanying drawings. In the accompanying drawings:

[0085] Figure 1 A flowchart illustrating a method for training a state recognition model according to an embodiment of the present disclosure is shown;

[0086] Figure 2 A flowchart illustrating a mapping relationship between feature data and specific attribute values ​​according to an embodiment of the present disclosure is shown;

[0087] Figure 3 A schematic diagram illustrating a piecewise function according to an embodiment of the present disclosure;

[0088] Figure 4 A flowchart of determining a normalization status indicator according to an embodiment of the present disclosure is shown;

[0089] Figure 5 A schematic diagram illustrating a state prediction based on graph features according to an embodiment of the present disclosure is shown;

[0090] Figure 6 A schematic diagram showing a subject relationship diagram according to an embodiment of the present disclosure;

[0091] Figure 7 A flowchart showing training of the state recognition model according to the standardized state indicator according to an embodiment of the present disclosure is shown;

[0092] Figure 8 A schematic diagram illustrating feature data processing according to an embodiment of the present disclosure;

[0093] Figure 9 A flowchart of determining a second vector representation according to an embodiment of the present disclosure is shown;

[0094] Figure 10 A schematic diagram illustrating a graphical presentation of standardized status indicators according to an embodiment of the present disclosure;

[0095] Figure 11 A flow chart showing a state recognition method according to an embodiment of the present disclosure is shown;

[0096] Figure 12 A flowchart showing a method for identifying enterprise risks according to an embodiment of the present disclosure is shown;

[0097] Figure 13 A block diagram illustrating a training device for a state recognition model according to an embodiment of the present disclosure is shown;

[0098] Figure 14 A block diagram showing a state recognition device according to an embodiment of the present disclosure is shown;

[0099] Figure 15 A block diagram illustrating an electronic device according to an embodiment of the present disclosure is shown;

[0100] Figure 16 A schematic diagram showing the structure of a computer system suitable for implementing the method and apparatus of the embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0101] Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily implement them. In addition, for the sake of clarity, parts not related to the description of the exemplary embodiments are omitted in the accompanying drawings.

[0102] In the present disclosure, it should be understood that terms such as "include" or "have" are intended to indicate the presence of features, numbers, steps, actions, components, parts, or combinations thereof disclosed in the present specification, and are not intended to exclude the possibility that one or more other features, numbers, steps, actions, components, parts, or combinations thereof exist or are added.

[0103] It should also be noted that, in the absence of conflict, the embodiments and features of the embodiments of the present disclosure may be combined with each other. The present disclosure will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0104] As mentioned above, the characteristic data of the subject itself and the events related to the subject cannot well reflect the true state of the subject, which easily leads to inaccurate state recognition results.

[0105] For example, when identifying corporate risks, one typically focuses on litigation risk. While larger companies are more likely to be involved in lawsuits, this doesn't necessarily mean they are high-litigation risk. Conversely, smaller companies may not be involved in many lawsuits, but compared to similar companies, they may still have a higher litigation risk. Therefore, it's difficult for users to intuitively determine a company's litigation risk based solely on the number of defendant records.

[0106] According to the technical solution provided by the embodiment of the present disclosure, feature data related to a subject and status data related to the subject are obtained; specific attribute values ​​matching the feature data are determined according to a preset mapping relationship; standardized status indicators of the subject are determined based on the status data related to the subject and the specific attribute values; the status recognition model is trained according to the standardized status indicators, so that the status data can be converted into standardized status indicators. The status recognition model trained based on the standardized status indicators can more accurately identify the status of the subject.

[0107] Figure 1 A flowchart of a state recognition model training method according to an embodiment of the present disclosure is shown.

[0108] like Figure 1 As shown, the state recognition model training method includes operations S110 to S140.

[0109] In operation S110 , feature data related to a subject and state data related to the subject are obtained;

[0110] In operation S120, a specific attribute value matching the feature data is determined according to a preset mapping relationship;

[0111] In operation S130 , a standardized status indicator of the subject is determined based on the status data related to the subject and the specific attribute value;

[0112] In operation S140 , the state recognition model is trained according to the standardized state index.

[0113] According to an embodiment of the present disclosure, taking the identification of the risk of being a defendant in a corporate contract as an example, the subject may be the enterprise, the characteristic data may be various corporate information including information such as turnover, and the status data associated with the subject may be, for example, the number of specific events associated with the enterprise. Specific events may be, for example, lawsuits, contract lawsuits, or lawsuits or contract lawsuits in which the enterprise is the defendant. That is, operation S110 may involve obtaining the characteristic data of the enterprise and the number of contract lawsuits in which the enterprise is the defendant.

[0114] According to the embodiment of the present disclosure, the specific attribute value is information that is not easy to obtain. For example, in the embodiment of the present disclosure, the specific attribute value can be the number of contracts signed by the enterprise each year. The preset mapping relationship can be a mapping relationship between feature data established in advance and the specific attribute value, so that after obtaining the feature data of the subject, the specific attribute value of the subject can be determined according to the mapping relationship. For example, the inventor found that the number of contracts signed by an enterprise each year has a certain positive correlation with the annual turnover of the enterprise. Based on the data that can be obtained, a mapping relationship between the annual turnover and the number of contracts signed by the enterprise each year can be established, so that when the annual turnover data in the feature data is obtained, the estimated value of the number of contracts signed by the enterprise each year can be determined according to the mapping relationship. The embodiment of the present disclosure uses the relationship between the average number of contracts signed each year by enterprises of different sizes and the scale factor of the enterprise (such as average turnover) to fit a function. For each enterprise, the function can be used to convert the number of contract defendants into a normalized contract litigation risk for subsequent judgment and modeling.

[0115] According to embodiments of the present disclosure, by comprehensively considering status data and specific attribute values, a standardized status indicator for an entity can be determined to better assess the entity's status. For example, since enterprises of different sizes are more likely to be involved in litigation than small and micro enterprises, a standardized status indicator can be determined based on the ratio of contract litigation cases in which the enterprise is the defendant to the number of contracts signed each year.

[0116] According to the disclosed embodiments, based on the state recognition model, all the characteristics of an entity over the previous period can be used to predict the entity's state over the next period. For example, all the characteristics of each enterprise over the previous period can be used to predict the enterprise's contract litigation risk over the next period. These characteristics are likely related to the risk of being a contract defendant, such as the enterprise's industry, region, scale, financial status, etc. This not only allows us to understand the enterprise's past contract litigation risk, but also predict the contract litigation risk that is more likely to be realistic in the future.

[0117] For example, we can use the 2019 records from the obtained judicial documents to calculate the contract litigation risk of each enterprise in 2019 as a label. We can also use various historical characteristics of the enterprise in 2018, such as the number of defendants in 2018 and the number of administrative penalties in 2018, to predict this label. Of course, some fundamental information about the enterprise, such as registered capital, generally does not change, so data from either 2018 or 2019 can be used. In this way, using the trained model, when analyzing a new enterprise on the test set, we can use its historical characteristics to predict the current enterprise's standardized contract litigation risk.

[0118] According to the technical solution provided by the embodiment of the present disclosure, feature data related to a subject and status data related to the subject are obtained; specific attribute values ​​matching the feature data are determined according to a preset mapping relationship; standardized status indicators of the subject are determined based on the status data related to the subject and the specific attribute values; the status recognition model is trained according to the standardized status indicators, so that the status data can be converted into standardized status indicators. The status recognition model trained based on the standardized status indicators can more accurately identify the status of the subject.

[0119] According to an embodiment of the present disclosure, the mapping relationship at least includes a correspondence between characteristic data and characteristic attribute values, wherein characteristic data in different value ranges correspond to different characteristic attribute values.

[0120] Figure 2 A flowchart of determining a mapping relationship between feature data and specific attribute values ​​according to an embodiment of the present disclosure is shown.

[0121] like Figure 2 As shown, the method may further include operations S210 to S230.

[0122] In operation S210, the subject is divided into a plurality of categories based on the feature data, each category corresponding to a different value range of the feature data;

[0123] In operation S220 , an average value of the specific attribute value of the subjects in each category is determined;

[0124] In operation S230 , a piecewise function is constructed as a mapping relationship between the feature data and the specific attribute value based on the value range of the feature data and the average value of the specific attribute value.

[0125] Taking the example of corporate contract litigation risk, we can first extract cases related to corporate contracts from judicial documents. From these cases, we can select records where the defendant is a company. After this extraction, we can count the number of contract-related defendants for each company.

[0126] As mentioned above, the number of defendants alone cannot directly reflect the risk of contractual defendants for an enterprise and needs to be standardized according to certain rules. For example, the standardized status indicator can be defined as the average risk R of each contract signed with the current enterprise. contract =C / N contract , where C is the number of times the current enterprise has been sued historically, N contract is the number of contracts signed by the company historically. The following table shows the relationship between the average annual number of contracts signed and average turnover for domestic companies of different sizes, as determined through a sample survey.

[0127] Table 1

[0128]

[0129] At each enterprise-level interval, we assume that the growth is linear in each individual interval. In this way, we can construct a linear function based on the upper and lower bounds of the current interval’s turnover and the average number of contracts in the interval as an estimate of the actual situation, such as Figure 3 As shown. Therefore, for the subject to be identified, we can Figure 3 The piecewise function shown uses the feature data to estimate the specific attribute value of the subject, thereby obtaining the subject's label, such as the contract defendant risk value R of the enterprise contract .

[0130] According to the technical solution provided by the embodiment of the present disclosure, a mapping relationship includes at least a correspondence between feature data and specific attribute values; wherein, feature attribute values ​​corresponding to feature data in different value ranges are different, so that the number of specific events can be converted into a standardized state indicator based on the mapping relationship, and the state recognition model trained based on the standardized state indicator can more accurately identify the state of the subject.

[0131] Figure 4 A flow chart of determining a normalized status indicator according to an embodiment of the present disclosure is shown.

[0132] like Figure 4As shown, the above operation S130, ie, determining the standardized status indicator of the subject based on the status data and the specific attribute value, may include operations S410 and S420.

[0133] In operation S410, an original state indicator is determined based on the state data and the specific attribute value;

[0134] In operation S420, the original state indicators of different entities are mapped to a predetermined value range according to the maximum and minimum values ​​of the original state indicators to obtain a standardized state indicator.

[0135] According to the embodiment of the present disclosure, it is also possible to utilize the R contract Perform scaling and normalization of the maximum and minimum values ​​so that the overall score range is within a predetermined interval, for example, 0 to 10 points, and then normalize the R′ obtained after normalization. contract as a standardized status indicator.

[0136] According to the technical solution provided by the embodiment of the present disclosure, the original state indicator is determined based on the quantitative state data of the specific event and the specific attribute value of the subject; according to the maximum and minimum values ​​of the original state indicators of different subjects based on different subjects, the original state indicator is mapped to a predetermined value range to obtain a standardized state indicator, so that the obtained standardized state indicator is more effective, and the state recognition model trained based on the standardized state indicator can more accurately identify the state of the subject.

[0137] While current technologies primarily consider the impact of the current subject's historical information on its state, the inventors have discovered that, in practice, information about subjects associated with the current subject can also be very helpful in predicting the current subject's state. Previous solutions have not adequately utilized the various data and records of these associated nodes.

[0138] According to an embodiment of the present disclosure, the feature data includes graph features determined based on association relationships between the subjects in a subject relationship graph.

[0139] For example, using the example of predicting a company's contract defendant risk, these relationships can include investment relationships, equity structures, partnerships, competition, and other potential relationships. These inter-company relationships can also be very helpful in predicting a company's contract defendant risk. The litigation records of a single company may contain accidental factors, but if a more comprehensive contract litigation prediction model is trained using the full data (especially one that combines neighbor node attributes and record graph features), the results will be more accurate.

[0140] Figure 5 A schematic diagram illustrating state prediction based on graph features according to an embodiment of the present disclosure is shown.

[0141] like Figure 5 The left and right figures are exemplary schematic diagrams of the subject relationship diagram. Nodes marked 1 represent samples with the subject in the first state, such as a company with a high-risk contract defendant risk label, while nodes marked 0 represent samples with the second state, such as a company with a low-risk contract defendant risk label. Unmarked nodes represent the subject to be identified. The states of other nodes in the surrounding environment are also helpful in determining the state of the current subject. The state of the subject to be identified can be predicted based on the attributes of the subject itself and the states of the surrounding nodes.

[0142] For example, if a subsidiary invested in by a company is frequently sued, the company's risk of being sued may also increase. Here, we use investment relationships as an example of an association relationship. If a node in the subject relationship graph is surrounded by two or more nodes with high-risk contract defendant records, the company's risk of being sued increases. Therefore, we can hypothetically calculate a feature that represents the number of contract defendant records among the company's neighboring nodes. Alternatively, a node is considered high-risk only if it has been sued five or more times. Such a feature would be the number of neighboring nodes with more than five contract defendant records. Each neighboring node can even have its own attributes, such as "neighboring nodes with contract defendant records and private companies." There can be many such graph features, and the model can automatically learn these graph-related features, thereby improving the accuracy of state recognition.

[0143] According to the technical solution provided by the embodiment of the present disclosure, the graph features determined based on the association relationship of the subjects in the subject relationship graph are used as feature data, so that the status of the subject can be better identified in combination with the graph features.

[0144] According to the embodiments of the present disclosure, the utilization of graph features can adopt the principle of GraphSage, for example. The basic idea of ​​GraphSage is to learn how the information of a node is aggregated through the features of its neighboring nodes. Compared with the shortcomings of methods such as graph convolutional neural networks (GCN), which learn that the graph features of each node are fixed, the graph features of the nodes learned by the GraphSage method change according to the changes in the neighbor relationships of the nodes. That is to say, even for old nodes, if some new connection relationships are established, the corresponding graph features will also change, and they can be learned very conveniently. The following is combined with Figure 6 A brief explanation of the principles of GraphSage.

[0145] Figure 6 A schematic diagram showing a subject relationship diagram according to an embodiment of the present disclosure.

[0146] like Figure 6 As shown in the figure, A's neighbors are B, C, and D. Taking one of the neighbors B as an example, this node also has two neighbors A and C. Represent the initial attributes of nodes A, B, C, and D respectively. In the first round, for node B, first replace the attributes of neighboring nodes A and C. and The vector obtained after weighted aggregation and the current attribute vector of B Splice and get Similarly, we can get Then, in the second round, Perform weighted aggregation to obtain the neighbor node aggregation vector of node A Then compare it with the current A node vector Splice to get the new vector of node A By analogy, we can obtain feature vectors of any order, and the specific order can be set according to actual needs. Graphsage can better utilize the attributes of neighboring nodes to predict the state of the current node.

[0147] However, the inventors have discovered that when traditional Graphsage encounters many enumerated features, the sparsity of the one-hot variables of the enumerated features will lead to a deterioration in the model effect. For example, if all companies are involved in 1,000 industries, then the one-hot vector of the industry attribute will have 1,000 dimensions, and the features will be very sparse. If there are multiple similar enumerated attribute features, the attribute needs to be one-hot processed first and then spliced ​​to other attribute vectors, which will make it even sparser. In addition, many nodes on the graph may not have attributes. In this way, ordinary Graphsage cannot handle the problem of sparsity of this attribute feature vector.

[0148] The disclosed embodiment proposes an improved Graphsage method based on enumerated attribute embedding compression. By learning the average pooling of the attribute embedding of each node, information compression of multiple enumerated attributes is achieved, helping the model to better learn features.

[0149] Figure 7 A flowchart of training the state recognition model according to the standardized state indicators according to an embodiment of the present disclosure is shown.

[0150] like Figure 7 As shown, the method includes operations S710 to S750.

[0151] In operation S710, a first vector representation of feature data of the subject is determined;

[0152] In operation S720, the first vector representation is converted from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation;

[0153] In operation S730, the second vector representation is concatenated with a third vector representation of feature data of a neighboring subject of the subject in the subject relationship graph to obtain a fourth vector representation of feature data of the subject;

[0154] In operation S740, encoding the fourth vector representation into a fifth vector representation of feature data of the subject having a predetermined dimension as training data for the state recognition model;

[0155] In operation S750 , the state recognition model is trained using the standardized state indicators as labels.

[0156] According to the embodiment of the present disclosure, the first vector representation can be, for example, various enumerated one-hot feature representations. By converting these high-dimensional sparse vector representations into low-dimensional dense second vector representations, the problem of vector sparsity is at least partially solved. Through the idea of ​​Graphsage, the converted second vector representation is used to replace the initial attribute of node X. By concatenating the feature vector of the node with the third feature vector of the adjacent node to obtain a fourth feature vector, and re-encoding it into a fifth feature vector of a predetermined dimension, graph features can be integrated into the original attribute features.

[0157] Node embeddings can simultaneously contain two types of information: the aggregation of neighboring node attribute information onto the current node, and information about the inter-node location context. The first type of information can be learned using a supervised loss. Specifically, the state label of the current node is used to aggregate the attributes of its neighboring nodes onto the current node as its features, and the labels are then used to reversely optimize the neighboring node embeddings. This information can be understood as follows: if a certain feature (such as whether a node has lost trust) is highly influential on the state, then aggregating that feature from neighboring nodes can better determine the current node's state. The second type of information can be optimized using an unsupervised loss. The main idea is to bring the embedding distances of adjacent nodes closer, using a method similar to negative sampling in node2vec. This information can be understood as follows: if a large number of neighboring nodes have the state label "1," then the probability that the current node is in the state "1" is also high.

[0158] The following combination Figure 8 The illustrated embodiment is Figure 7 The method is described.

[0159] Figure 8 A schematic diagram illustrating feature data processing according to an embodiment of the present disclosure.

[0160] like Figure 8 As shown, for nodes A, B, and C, they are initially represented by one-hot feature vectors of different attributes, that is, nodes A, B, and C each have their own first vector representation. Still taking the enterprise as an example, the enterprise itself can have a Nodeone-hot, and then each attribute i of the enterprise has a Feat_{i}one-hot. These one-hot vectors are converted into dense vectors of predetermined dimensions, such as embedding vectors. After weighted summation of the attribute's Feat_{i}embedding and the enterprise's own Node embedding (or taking the average), the final embedding vector of the enterprise is obtained, that is, the second vector representation. Among them, if an attribute of an enterprise is empty, then the attribute can be ignored when calculating the second feature vector of the enterprise. Nodes A, B, and C can calculate their own second vector representations respectively, such as Figure 8 As shown. By concatenating and encoding the second vector representations of nodes B and C, we can obtain a third vector representation of node A's neighbor nodes. This is then concatenated with the second vector representation of node A to obtain a fourth vector representation of node A. By encoding to a predetermined dimension, we obtain a fifth vector representation, which serves as a vector representation of node A that incorporates the graph features of its first-order neighbor nodes. The above method can be used to obtain a vector representation of each node that incorporates the graph features of its neighbor nodes of a given order. This vector representation serves as training data for the state recognition model, which can be trained using the aforementioned standardized state indicators as labels.

[0161] According to the technical solution provided by the embodiment of the present disclosure, a first vector representation of the feature data of the subject is determined; the first vector representation is converted from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation; the second vector representation is spliced ​​with the third vector representation of the feature data of the adjacent subjects of the subject in the subject relationship graph to obtain a fourth vector representation of the feature data of the subject; the fourth vector representation is encoded into a fifth vector representation of the feature data of the subject with a predetermined dimension as training data for the state recognition model; the state recognition model is trained with the standardized state indicator as a label, so as to better utilize the graph features to identify the state of the subject.

[0162] Figure 9 A flowchart for determining a second vector representation according to an embodiment of the present disclosure is shown.

[0163] According to an embodiment of the present disclosure, the first vector representation includes a one-hot feature representation, such as Figure 9As shown, the operation S720, that is, converting the first vector representation from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation, may include operations S910 and S920.

[0164] In operation S910, the one-hot feature representation is converted into a sixth vector representation through an embedding operation;

[0165] In operation S920, weighted summation is performed on the sixth vector representation to obtain the low-dimensional dense second vector representation.

[0166] According to an embodiment of the present disclosure, a unique feature representation is a one-hot feature representation. For example, for a subject with 50 non-empty attributes, the 50 attributes can be represented as 50 one-hot feature representations respectively, but this representation is too sparse and not conducive to subsequent processing. The method of the embodiment of the present disclosure can convert the 50 one-hot vectors into 50 low-dimensional dense vector representations through an embedding operation (embedding), which is commonly known as an embedding vector representation, and then aggregate the 50 embedding vectors. For example, weights can be set for each of the 50 attributes, and a vector representation of the subject, i.e., a second vector representation, can be obtained by weighted summation.

[0167] According to the embodiments of the present disclosure, if an attribute is not an enumeration type one-hot but a continuous numerical type, two solutions can be used for processing. One is to directly manually segment the continuous numerical attribute and discretize it into an enumerable type; the other is to normalize the value and splice it into the Mixture embedding as one bit.

[0168] According to the technical solution provided by the embodiment of the present disclosure, the one-hot feature representation is converted into a sixth vector representation through an embedding operation; and the sixth vector representation is weighted summed to obtain the low-dimensional dense second vector representation, thereby solving the problem of excessive resource occupation caused by sparse features.

[0169] According to an embodiment of the present disclosure, the method may further include presenting the standardized status indicator in a graphical manner. For example, Figure 10 A schematic diagram showing a graphical presentation of standardized status indicators according to an embodiment of the present disclosure is shown. Figure 10 As shown, the standardized status indicator can be presented on the user interface in a dashboard style, for example, so as to intuitively prompt the user with the standardized status indicator of the current subject.

[0170] According to the technical solution provided by the embodiment of the present disclosure, the state of the subject can be more intuitively displayed by presenting the standardized state indicators in a graphical manner.

[0171] Figure 11 A flow chart of a state recognition method according to an embodiment of the present disclosure is shown.

[0172] like Figure 11 As shown, the method includes operations S1110 and S1120.

[0173] In operation S1110 , feature data of a subject to be identified is obtained;

[0174] In operation S1120, the feature data is input into Figures 1 to 10 A state recognition model is trained using any of the state recognition model training methods described above to obtain standardized state indicators of the subject to be identified.

[0175] According to the technical solution provided by the embodiment of the present disclosure, the characteristic data of the subject to be identified is obtained; the characteristic data is input into a state recognition model trained by the above-mentioned state recognition model training method to obtain the standardized state index of the subject to be identified, so that the state of the subject can be more accurately identified through the standardized state index.

[0176] Figure 12 A flow chart of a method for identifying enterprise risks according to an embodiment of the present disclosure is shown.

[0177] like Figure 12 As shown, the method includes operations S1210 to S1240.

[0178] In operation S1210, characteristic data of an enterprise and the number of contract litigation cases in which the enterprise is a defendant are obtained, wherein the characteristic data includes turnover;

[0179] In operation S1220, the number of contracts of the enterprise is determined according to the mapping relationship between the turnover and the number of contracts;

[0180] In operation S1230, a standardized risk index of the enterprise is determined based on the number of contract litigation cases and the number of contracts of the enterprise;

[0181] In operation S1240 , training data is determined based on the characteristic data of the enterprise, and a risk identification model is trained using the standardized risk indicators as labels.

[0182] According to the technical solution provided by the embodiments of the present disclosure, characteristic data of an enterprise and the number of contract litigation cases in which the enterprise is the defendant are obtained, wherein the characteristic data includes turnover; the number of contracts of the enterprise is determined based on the mapping relationship between the turnover and the number of contracts; the standardized risk index of the enterprise is determined based on the number of contract litigation cases and the number of contracts of the enterprise; training data is determined based on the characteristic data of the enterprise, and a risk identification model is trained with the standardized risk index as a label, so that the number of contract litigation cases can be converted into a standardized risk index. The risk identification model trained based on the standardized risk index can more accurately identify the risks of the enterprise.

[0183] For example, before an enterprise intends to sign a contract with another entity, the enterprise's corporate legal department can use the method of the embodiment of the present disclosure to evaluate the risk status of the other party as a basis for decision-making.

[0184] Those skilled in the art will understand that the above scheme in the embodiments of the present disclosure can not only be used to determine the risks of the enterprise to be identified, but can also be used to identify the status or risks of organizations and personnel within or outside the enterprise, can also be used to identify the status or risks of living or non-living things in nature, and can also be used to identify the status or risks of the subject to be identified in the form of data.

[0185] For example, the subject of the embodiment of the present disclosure may be a merchant on an e-commerce platform. The standardized status indicators of the merchant can be judged according to the method of the embodiment of the present disclosure, and the standardized status indicators can be used to manage the merchant or to handle complaints against the merchant. For another example, a lawyer can use the method of the embodiment of the present disclosure to determine the standardized status indicators of a certain subject as an evaluation of the subject, which can be used to simulate the views of the court. For another example, the procuratorate can use the method of the real-time example of the present disclosure to make predictions for the purpose of forming a plan. In addition, in various other civil and administrative management, or the handling of civil and administrative disputes, the people's courts, arbitration committees, governments and government departments can use the standardized status indicators generated by the method provided by the embodiment of the present disclosure as a reference. The method of the embodiment of the present disclosure can also be used to check whether the focus of the dispute has changed.

[0186] Figure 13 A block diagram of a state recognition model training device according to an embodiment of the present disclosure is shown. The device can be implemented as part or all of an electronic device through software, hardware, or a combination of both.

[0187] like Figure 13 As shown, the state recognition model training device 1300 includes a first acquisition module 1310 , a first determination module 1320 , a second determination module 1330 and a training module 1340 .

[0188] A first acquisition module 1310 is configured to obtain feature data related to a subject and status data related to the subject;

[0189] The first determining module 1320 is configured to determine a specific attribute value that matches the feature data according to a preset mapping relationship;

[0190] A second determination module 1330 is configured to determine a standardized status indicator of the subject based on the status data related to the subject and the specific attribute value;

[0191] The training module 1340 is configured to train the state recognition model according to the standardized state indicator.

[0192] According to the technical solution provided by the embodiment of the present disclosure, a first acquisition module is configured to obtain feature data related to a subject and status data related to the subject; a first determination module is configured to determine a specific attribute value that matches the feature data according to a preset mapping relationship; a second determination module is configured to determine a standardized status indicator of the subject based on the status data related to the subject and the specific attribute value; a training module is configured to train the status recognition model according to the standardized status indicator, so that the number of specific events can be converted into a standardized status indicator, and the status recognition model trained based on the standardized status indicator can more accurately identify the status of the subject.

[0193] According to an embodiment of the present disclosure, the mapping relationship at least includes a correspondence between characteristic data and specific attribute values, wherein characteristic attribute values ​​corresponding to characteristic data in different value ranges are different.

[0194] According to the technical solution provided by the embodiment of the present disclosure, a mapping relationship includes at least a correspondence between feature data and specific attribute values, wherein feature data with different value ranges correspond to different feature attribute values, so that the number of specific events can be converted into a standardized state indicator based on the mapping relationship, and the state recognition model trained based on the standardized state indicator can more accurately identify the state of the subject.

[0195] According to an embodiment of the present disclosure, determining the standardized status indicator of the subject based on the status data related to the subject and the specific attribute value includes:

[0196] determining an original state indicator based on the state data and the specific attribute value;

[0197] According to the maximum and minimum values ​​of the original state indicators of different subjects, the original state indicators are mapped to a predetermined value range to obtain a standardized state indicator.

[0198] According to the technical solution provided by the embodiment of the present disclosure, the original state indicator is determined based on the state data and the specific attribute value; according to the maximum and minimum values ​​of the original state indicators of different subjects, the original state indicator is mapped to a predetermined value range to obtain a standardized state indicator, so that the obtained standardized state indicator is more effective, and the state recognition model trained based on the standardized state indicator can more accurately identify the state of the subject.

[0199] According to an embodiment of the present disclosure, the feature data includes graph features determined based on association relationships between the subjects in a subject relationship graph.

[0200] According to the technical solution provided by the embodiment of the present disclosure, the graph features determined based on the association relationship of the subjects in the subject relationship graph are used as feature data, so that the status of the subject can be better identified in combination with the graph features.

[0201] According to an embodiment of the present disclosure, training the state recognition model according to the standardized state indicator includes:

[0202] determining a first vector representation of characteristic data of the subject;

[0203] Converting the first vector representation from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation;

[0204] concatenating the second vector representation with a third vector representation of feature data of a neighboring subject of the subject in the subject relationship graph to obtain a fourth vector representation of the feature data of the subject;

[0205] encoding the fourth vector representation into a fifth vector representation of feature data of the subject having a predetermined dimension as training data for the state recognition model;

[0206] The state recognition model is trained using the standardized state indicators as labels.

[0207] According to the technical solution provided by the embodiment of the present disclosure, a first vector representation of the feature data of the subject is determined; the first vector representation is converted from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation; the second vector representation is spliced ​​with the third vector representation of the feature data of the adjacent subjects of the subject in the subject relationship graph to obtain a fourth vector representation of the feature data of the subject; the fourth vector representation is encoded into a fifth vector representation of the feature data of the subject with a predetermined dimension as training data for the state recognition model; the state recognition model is trained with the standardized state indicator as a label, so as to better utilize the graph features to identify the state of the subject.

[0208] According to an embodiment of the present disclosure, the first vector representation includes a one-hot feature representation, and converting the first vector representation from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation includes:

[0209] Converting the one-hot feature representation into a sixth vector representation through an embedding operation;

[0210] The low-dimensional dense second vector representation is obtained by weighted summing of the sixth vector representations.

[0211] According to the technical solution provided by the embodiment of the present disclosure, the first vector representation of the one-hot feature representation is converted into a sixth vector representation through an embedding operation; the weighted sum of the sixth vector representation is performed to obtain the low-dimensional dense second vector representation, thereby solving the problem of excessive resource occupation caused by sparse features.

[0212] According to an embodiment of the present disclosure, the apparatus may further include a display module configured to present the standardized status indicator in a graphical manner, thereby being able to more intuitively display the status of the subject.

[0213] Figure 14 A block diagram of a state recognition device according to an embodiment of the present disclosure is shown, wherein the device can be implemented as part or all of an electronic device through software, hardware, or a combination of both.

[0214] like Figure 14 As shown, the state recognition device 1400 includes a second acquisition module 1410 and a recognition module 1420 .

[0215] The second acquisition module 1410 is configured to acquire feature data of the subject to be identified;

[0216] The identification module 1420 is configured to input the feature data into the state identification model as described above to obtain the standardized state index of the subject to be identified.

[0217] According to the technical solution provided by the embodiment of the present disclosure, the second acquisition module is configured to obtain the characteristic data of the subject to be identified; the identification module is configured to input the characteristic data into the state recognition model trained by the state recognition model training method described above to obtain the standardized state index of the subject to be identified, so that the state of the subject can be more accurately identified through the standardized state index.

[0218] The present disclosure also discloses an electronic device, Figure 15 A block diagram of an electronic device according to an embodiment of the present disclosure is shown.

[0219] like Figure 15As shown, the electronic device 1500 includes a memory 1501 and a processor 1502, wherein the memory 1501 is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor 1502 to implement the following operations:

[0220] obtaining characteristic data associated with a subject, and status data associated with the subject;

[0221] Determining a specific attribute value that matches the feature data according to a preset mapping relationship;

[0222] determining a standardized status indicator of the subject based on the status data associated with the subject and the specific attribute value;

[0223] The state recognition model is trained according to the standardized state indicator.

[0224] According to an embodiment of the present disclosure, the mapping relationship includes at least: a correspondence between characteristic data and specific attribute values; wherein characteristic attribute values ​​corresponding to characteristic data in different value ranges are different.

[0225] According to an embodiment of the present disclosure, determining the standardized status indicator of the subject based on the status data related to the subject and the specific attribute value includes:

[0226] determining an original state indicator based on the state data and the specific attribute value;

[0227] According to the maximum and minimum values ​​of the original state indicators of different subjects, the original state indicators are mapped to a predetermined value range to obtain a standardized state indicator.

[0228] According to an embodiment of the present disclosure, the feature data includes graph features determined based on association relationships between the subjects in a subject relationship graph.

[0229] According to an embodiment of the present disclosure, training the state recognition model according to the standardized state indicator includes:

[0230] determining a first vector representation of characteristic data of the subject;

[0231] Converting the first vector representation from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation;

[0232] concatenating the second vector representation with a third vector representation of feature data of a neighboring subject of the subject in the subject relationship graph to obtain a fourth vector representation of the feature data of the subject;

[0233] encoding the fourth vector representation into a fifth vector representation of feature data of the subject having a predetermined dimension as training data for the state recognition model;

[0234] The state recognition model is trained using the standardized state indicators as labels.

[0235] According to an embodiment of the present disclosure, the first vector representation includes a one-hot feature representation, and converting the first vector representation from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation includes:

[0236] Converting the one-hot feature representation into a sixth vector representation through an embedding operation;

[0237] The low-dimensional dense second vector representation is obtained by weighted summing of the sixth vector representations.

[0238] According to an embodiment of the present disclosure, the processor 1502 is further configured to execute: presenting the standardized status indicator in a graphical manner.

[0239] According to an embodiment of the present disclosure, the one or more computer instructions are executed by the processor 1502 to implement the following operations:

[0240] Obtaining feature data of the subject to be identified;

[0241] The feature data is input into the state recognition model as described above to obtain the standardized state index of the subject to be recognized.

[0242] According to an embodiment of the present disclosure, the one or more computer instructions are executed by the processor 1502 to implement the following operations:

[0243] Obtaining characteristic data of an enterprise and the number of contract litigation cases in which the enterprise is a defendant, wherein the characteristic data includes turnover;

[0244] Determining the number of contracts for the enterprise based on the mapping relationship between the turnover and the number of contracts;

[0245] Determining a standardized risk index for the enterprise based on the number of contract litigation cases and the number of contracts of the enterprise;

[0246] Training data is determined based on the characteristic data of the enterprise, and a risk identification model is trained using the standardized risk indicators as labels.

[0247] Figure 16 A schematic diagram showing the structure of a computer system suitable for implementing the method and apparatus of the embodiments of the present disclosure is shown.

[0248] like Figure 16As shown, the computer system 1600 includes a processing unit 1601, which can execute various processes in the above-described embodiments according to a program stored in a read-only memory (ROM) 1602 or a program loaded from a storage unit 1608 into a random access memory (RAM) 1603. Various programs and data required for the operation of the system 1600 are also stored in the RAM 1603. The processing unit 1601, the ROM 1602, and the RAM 1603 are connected to each other via a bus 1604. An input / output (I / O) interface 1605 is also connected to the bus 1604.

[0249] The following components are connected to the I / O interface 1605: an input section 1606 including a keyboard, a mouse, etc.; an output section 1607 including a cathode ray tube (CRT), a liquid crystal display (LCD), a speaker, etc.; a storage section 1608 including a hard disk, etc.; and a communication section 1609 including a network interface card such as a LAN card, a modem, etc. The communication section 1609 performs communication processing via a network such as the Internet. A drive 1610 is also connected to the I / O interface 1605 as needed. Removable media 1611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1610 as needed so that a computer program read therefrom can be installed into the storage section 1608 as needed. Among them, the processing unit 1601 can be implemented as a processing unit such as a CPU, a GPU, a TPU, an FPGA, or an NPU.

[0250] In particular, according to embodiments of the present disclosure, the methods described above can be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program tangibly embodied on a computer-readable medium, the computer program comprising program code for executing the methods described above. In such embodiments, the computer program can be downloaded and installed from a network via the communication portion 1609 and / or installed from the removable medium 1611.

[0251] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment or part of code, and the module, program segment or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two 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 box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or can be implemented using a combination of dedicated hardware and computer instructions.

[0252] The units or modules involved in the embodiments described in this disclosure may be implemented by software or programmable hardware. The units or modules described may also be provided in a processor, and the names of these units or modules do not, in certain circumstances, constitute limitations on the units or modules themselves.

[0253] As another aspect, the present disclosure further provides a computer-readable storage medium. This computer-readable storage medium may be included in the electronic device or computer system described in the above embodiments, or may be a standalone computer-readable storage medium not incorporated into the device. The computer-readable storage medium stores one or more programs, which are used by one or more processors to execute the methods described in the present disclosure.

[0254] The above description is merely a preferred embodiment of the present disclosure and an illustration of the technical principles employed. Those skilled in the art should understand that the scope of the invention herein is not limited to the technical solutions formed by the specific combination of the above-mentioned technical features, but also encompasses other technical solutions formed by any combination of the above-mentioned technical features or their equivalents without departing from the inventive concept. For example, a technical solution formed by replacing the above-mentioned features with (but not limited to) technical features with similar functions disclosed in this disclosure.

Claims

1. A method for training a state recognition model, comprising: Obtaining characteristic data related to a subject and status data related to the subject, wherein the subject includes an enterprise, the characteristic data includes enterprise information, and the status data includes the number of litigation cases related to the enterprise; Determining a specific attribute value that matches the characteristic data according to a preset mapping relationship, wherein the specific attribute value includes the number of contracts signed by the enterprise; determining a standardized status indicator of the subject based on the status data associated with the subject and the specific attribute value, wherein the standardized status indicator is used to evaluate the status of the subject; training the state recognition model according to the standardized state indicator; The training of the state recognition model according to the standardized state indicator includes: determining a first vector representation of the characteristic data of the subject; converting the first vector representation from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation; concatenating the second vector representation with a third vector representation of the characteristic data of adjacent subjects in a subject relationship graph to obtain a fourth vector representation of the characteristic data of the subject; encoding the fourth vector representation into a fifth vector representation of the characteristic data of the subject with a predetermined dimension as training data for the state recognition model; and training the state recognition model using the standardized state indicator as a label.

2. The method according to claim 1, wherein the mapping relationship at least includes: The correspondence between feature data and specific attribute values; wherein, feature data with different value ranges correspond to different feature attribute values.

3. The method according to claim 1, wherein Determining the standardized status indicator of the subject based on the status data related to the subject and the specific attribute value includes: determining an original state indicator based on the state data and the specific attribute value; According to the maximum and minimum values ​​of the original state indicators of different subjects, the original state indicators are mapped to a predetermined value range to obtain a standardized state indicator.

4. The method according to claim 1, wherein The feature data includes graph features determined based on association relationships between the subjects in the subject relationship graph.

5. The method according to claim 1, wherein The first vector representation includes a one-hot feature representation, and converting the first vector representation from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation includes: Converting the one-hot feature representation into a sixth vector representation through an embedding operation; The low-dimensional dense second vector representation is obtained by weighted summing of the sixth vector representations.

6. A state recognition method, comprising: Obtaining feature data of the subject to be identified; The feature data is input into the state recognition model according to any one of claims 1 to 5 to obtain a standardized state index of the subject to be recognized.

7. A method for training a risk identification model, comprising: Obtaining characteristic data of an enterprise and the number of contract litigation cases in which the enterprise is a defendant, wherein the characteristic data includes turnover; Determining the number of contracts for the enterprise based on the mapping relationship between the turnover and the number of contracts; Determining a standardized risk index for the enterprise based on the number of contract litigation cases and the number of contracts of the enterprise; Determine training data based on the characteristic data of the enterprise, and train a risk identification model using the standardized risk indicators as labels; The determining of training data based on the characteristic data of the enterprise includes: determining a first vector representation of the characteristic data of the enterprise; converting the first vector representation from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation; concatenating the second vector representation with a third vector representation of the characteristic data of adjacent enterprises of the enterprise in the enterprise relationship graph to obtain a fourth vector representation of the characteristic data of the enterprise; encoding the fourth vector representation into a fifth vector representation of the characteristic data of the enterprise with a predetermined dimension as training data for the risk identification model.

8. A state recognition model training device, comprising: a first acquisition module configured to obtain characteristic data related to a subject and status data related to the subject, wherein the subject includes an enterprise, the characteristic data includes enterprise information, and the status data includes the number of litigation cases related to the enterprise; a first determining module configured to determine a specific attribute value matching the feature data according to a preset mapping relationship, wherein the specific attribute value includes the number of contracts signed by the enterprise; a second determining module configured to determine a standardized status indicator of the subject based on the status data related to the subject and the specific attribute value, wherein the standardized status indicator is used to evaluate the status of the subject; a training module, configured to train the state recognition model according to the standardized state indicator; The training module is also configured to: determine a first vector representation of the feature data of the subject; convert the first vector representation from a high-dimensional sparse vector representation into a low-dimensional dense second vector representation; concatenate the second vector representation with the third vector representation of the feature data of the adjacent subjects in the subject relationship graph to obtain a fourth vector representation of the feature data of the subject; encode the fourth vector representation into a fifth vector representation of the feature data of the subject with a predetermined dimension as training data for the state recognition model; and train the state recognition model using the standardized state indicator as a label.

9. The device according to claim 8, wherein The mapping relationship includes at least: The correspondence between feature data and specific attribute values, where feature data with different value ranges correspond to different feature attribute values.

10. The device according to claim 8, wherein Determining the standardized status indicator of the subject based on the status data related to the subject and the specific attribute value includes: determining an original state indicator based on the state data and the specific attribute value; According to the maximum and minimum values ​​of the original state indicators of different subjects, the original state indicators are mapped to a predetermined value range to obtain a standardized state indicator.

11. The device according to claim 8, wherein The feature data includes graph features determined based on association relationships between the subjects in the subject relationship graph.

12. The device according to claim 8, wherein The first vector representation includes a one-hot feature representation, and converting the first vector representation from a high-dimensional sparse vector representation to a low-dimensional dense second vector representation includes: Converting the one-hot feature representation into a sixth vector representation through an embedding operation; The low-dimensional dense second vector representation is obtained by weighted summing of the sixth vector representations.

13. A state recognition device comprising: A second acquisition module is configured to acquire feature data of a subject to be identified; The recognition module is configured to input the feature data into the state recognition model according to any one of claims 1 to 5 to obtain a standardized state indicator of the subject to be recognized.

14. An electronic device, characterized in that: Comprising a memory and a processor; wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the method steps described in any one of claims 1 to 7.

15. A readable storage medium having computer instructions stored thereon, characterized in that: When the computer instructions are executed by a processor, the method steps described in any one of claims 1 to 7 are implemented.

Images