Text quality inspection method and device, equipment and storage medium

Abstract

The invention relates to the technical field of computers, in particular to a text quality inspection method and device, equipment and a storage medium. The method comprises the following steps: acquiring a text to be subjected to quality inspection, and predicting the text to be subjected to quality inspection by adopting an error piece prediction model to obtain a prediction result; if the prediction result indicates that the text to be subjected to quality inspection is an error piece, performing quality inspection on the text to be subjected to quality inspection to obtain a quality inspection result corresponding to the text to be subjected to quality inspection; and if the prediction result indicates that the text to be subjected to quality inspection is a non-error piece, determining a stratified sampling requirement corresponding to the text to be subjected to quality inspection, and performing quality inspection on the text to be subjected to quality inspection under the condition that the text to be subjected to quality inspection meets the stratified sampling requirement to obtain a quality inspection result corresponding to the text to be subjected to quality inspection. By the adoption of the scheme, the defect of sampling randomness can be overcome, and the error piece occurrence probability is reduced.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to a text quality inspection method, device, equipment and storage medium. Background Art

[0002] During the text quality inspection process, simple random sampling is usually used to select texts for the quality inspection process based on the preset sampling ratio. However, simple random sampling samples have poor representativeness and large sampling errors. Some text types with a small number of them are not easy to be selected, and problems cannot be effectively discovered through random inspections. In addition, relying solely on random inspections can easily miss error-prone texts, fail to identify erroneous items, and fail to effectively reduce the risk of erroneous items. Summary of the invention

[0003] The present disclosure aims to solve one of the technical problems in the related art at least to some extent.

[0004] To this end, the first purpose of the present invention is to propose a text quality inspection method to make up for the lack of sampling randomness and reduce the probability of errors.

[0005] The second objective of the present disclosure is to provide a text quality inspection device.

[0006] A third objective of the present disclosure is to provide an electronic device.

[0007] A fourth objective of the present disclosure is to provide a computer-readable storage medium.

[0008] A fifth object of the present disclosure is to provide a computer program product.

[0009] To achieve the above-mentioned purpose, the first embodiment of the present disclosure proposes a text quality inspection method, including:

[0010] Acquire the text to be inspected, and use the error part prediction model to predict the text to be inspected to obtain a prediction result;

[0011] If the prediction result indicates that the text to be inspected is an error, then the text to be inspected is inspected to obtain a quality inspection result corresponding to the text to be inspected;

[0012] If the prediction result indicates that the text to be inspected is a non-error part, the stratified sampling requirements corresponding to the text to be inspected are determined, and if the text to be inspected meets the stratified sampling requirements, the text to be inspected is quality inspected to obtain the quality inspection result corresponding to the text to be inspected.

[0013] Optionally, before using the error part prediction model to predict the text to be inspected, the method further includes:

[0014] Get a sample text collection;

[0015] Based on the random forest algorithm, an error prediction model corresponding to the sample text set is constructed.

[0016] Optionally, the document to be inspected is an insurance claim case to be inspected, and determining the stratified sampling requirements corresponding to the document to be inspected includes:

[0017] Determine the claim type and claim amount corresponding to the insurance claim case to be inspected;

[0018] According to the corresponding relationship between the claim type and the claim amount and the sampling ratio, the sampling ratio corresponding to the insurance claim case to be inspected is determined, and the stratified sampling requirements corresponding to the text to be inspected are obtained.

[0019] Optionally, after determining the stratified sampling requirements corresponding to the text to be quality inspected, the method further includes:

[0020] Determine the interval number threshold corresponding to the sampling ratio;

[0021] Determine the number of non-error parts intervals of the same rule corresponding to the insurance claim case to be inspected;

[0022] If the number of intervals of non-error parts with the same rule is the interval number threshold, it is determined that the text to be inspected meets the stratified sampling requirement, and the number of intervals of non-error parts with the same rule is initialized;

[0023] If the number of intervals of non-error parts with the same rule is less than the interval number threshold, it is determined that the text to be quality inspected does not meet the stratified sampling requirement, and the number of intervals of non-error parts with the same rule is increased by one.

[0024] Optionally, before determining the sampling ratio corresponding to the insurance claim case to be inspected according to the corresponding relationship between the claim type and the claim amount and the sampling ratio, the method further includes:

[0025] Obtain a collection of existing insurance claims cases;

[0026] Dividing the insurance claim stock case set into multiple insurance claim stock case subsets according to claim type and claim amount;

[0027] According to the number of cases corresponding to the subset of insurance claim stock cases and the total number of cases corresponding to the set of insurance claim stock cases, the sampling ratio corresponding to the subset of insurance claim stock cases is determined, and the corresponding relationship between the claim type and claim amount and the sampling ratio is obtained.

[0028] Optionally, obtaining the text to be inspected includes:

[0029] Get the text to be approved;

[0030] If the document to be approved passes the approval, the document to be approved will be used as a document to be quality inspected.

[0031] Optionally, after obtaining the quality inspection result corresponding to the text to be quality inspected, the method further includes:

[0032] If the quality inspection result is that the quality inspection has been passed, the document to be inspected will be closed;

[0033] If the quality inspection result is that the quality inspection fails, the document to be inspected will be returned as the document to be approved, and the document to be inspected will be marked as an error document.

[0034] To achieve the above-mentioned purpose, a second aspect of the present disclosure provides a text quality inspection device, including:

[0035] A text prediction unit, used to obtain the text to be inspected, and use the error prediction model to predict the text to be inspected to obtain a prediction result;

[0036] A text quality inspection unit, configured to perform quality inspection on the text to be inspected if the prediction result indicates that the text to be inspected is an error, and obtain a quality inspection result corresponding to the text to be inspected;

[0037] The text sampling unit is used to determine the stratified sampling requirements corresponding to the text to be inspected if the prediction result indicates that the text to be inspected is a non-error part, and to perform quality inspection on the text to be inspected if the text to be inspected meets the stratified sampling requirements to obtain the quality inspection result corresponding to the text to be inspected.

[0038] To achieve the above-mentioned purpose, a third aspect of the present disclosure provides an electronic device, including: a processor, and a memory communicatively connected to the processor;

[0039] Memory stores computer-executable instructions;

[0040] The processor executes the computer-executable instructions stored in the memory to implement the method shown in any one of the first aspects above.

[0041] To achieve the above-mentioned objectives, an embodiment of the fourth aspect of the present disclosure proposes a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the method shown in any one of the above-mentioned first aspects.

[0042] To achieve the above-mentioned objective, a fifth aspect of the present disclosure provides a computer program product, including a computer program, which, when executed by a processor, implements the method shown in any one of the above-mentioned first aspects.

[0043] In summary, the method, apparatus, device and storage medium provided by the present invention, first, by adopting the error prediction model to predict the text to be inspected, it can make up for the lack of sampling randomness and effectively reduce the probability of error occurrence; then, by adopting stratified sampling requirements for stratified sampling, it can solve the problems of poor representativeness of simple random sampling samples and large sampling errors, and setting different stratified sampling requirements for different types of texts can make the inspected text more representative.

[0044] Additional aspects and advantages of the present disclosure will be given in part in the following description and in part will be obvious from the following description or learned through practice of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The above and / or additional aspects and advantages of the present disclosure will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0046] Figure 1 A flowchart of a text quality inspection method provided by an embodiment of the present disclosure;

[0047] Figure 2 A schematic diagram of a decision tree structure provided by an embodiment of the present disclosure;

[0048] Figure 3 A schematic diagram of the structure of a text quality inspection device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0049] Embodiments of the present disclosure are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present disclosure, and should not be construed as limiting the present disclosure.

[0050] The present disclosure is described in detail below with reference to specific embodiments.

[0051] In the first embodiment, if Figure 1 As shown, Figure 1 This is a flow chart of a text quality inspection method provided by an embodiment of the present disclosure, which can be implemented by a computer program and can be run on a device for text quality inspection. The computer program can be integrated into an application or run as an independent tool application.

[0052] Among them, the text quality inspection device can be an electronic device with a text quality inspection function.

[0053] Among them, the text quality inspection method can be executed by an electronic device.

[0054] For example, the text quality inspection method includes the following steps:

[0055] S101, obtaining a text to be inspected, and using an error part prediction model to predict the text to be inspected, to obtain a prediction result;

[0056] According to some embodiments, the text to be inspected refers to a text that needs to be inspected. The text to be inspected does not specifically refer to a fixed type of text. When the application field changes, the type of the text to be inspected may also change. For example, in the insurance field, the text to be inspected may be an insurance claim case to be inspected.

[0057] In some embodiments, an error component prediction model is used to perform error component prediction.

[0058] According to some embodiments, the prediction result refers to the result obtained after the error part prediction model predicts the text to be inspected.

[0059] S102, if the prediction result indicates that the document to be inspected is an error document, then the document to be inspected is inspected to obtain a quality inspection result corresponding to the document to be inspected;

[0060] According to some embodiments, the quality check is used to verify whether the document to be quality checked is an error document.

[0061] S103, if the prediction result indicates that the text to be inspected is a non-error part, determine the stratified sampling requirements corresponding to the text to be inspected, and if the text to be inspected meets the stratified sampling requirements, perform quality inspection on the text to be inspected to obtain the quality inspection result corresponding to the text to be inspected.

[0062] According to some embodiments, the stratified sampling requirement refers to the sampling requirement used when sampling text of the type to which the quality inspection text belongs. When the text type of the quality inspection text changes, the corresponding stratified sampling requirement will also change.

[0063] In summary, the method provided in this embodiment, first, by adopting the error prediction model to predict the text to be inspected, it can make up for the lack of sampling randomness and effectively reduce the probability of error occurrence; then, by adopting stratified sampling requirements for stratified sampling, it can solve the problems of poor representativeness of simple random sampling samples and large sampling errors. Setting different stratified sampling requirements for different types of texts can make the sampled text more representative.

[0064] This embodiment also provides another text quality inspection method, which can be executed by an electronic device.

[0065] For example, the text quality inspection method may include the following steps:

[0066] S201, obtaining the text to be approved;

[0067] According to some embodiments, the text to be approved refers to a text that needs to be approved.

[0068] S202, if the document to be approved is approved, the document to be approved is used as a document to be inspected;

[0069] It should be noted that if the text to be approved fails to pass the approval, a text modification instruction may be issued to instruct the text to be approved to be modified until the text to be approved passes the approval.

[0070] S203, obtaining a sample text set;

[0071] Taking a scenario as an example, in the insurance field, m error cases and m non-error cases can be randomly selected from the existing cases in the claims system.

[0072] S204, constructing an error prediction model corresponding to the sample text set based on a random forest algorithm;

[0073] According to some embodiments, the random forest algorithm is an integrated learning algorithm that improves prediction accuracy and stability by integrating multiple decision trees. Each decision tree is randomly trained, that is, a portion of samples is randomly selected from the original data for training, and then the final output result is determined by voting. The random forest algorithm can be used for classification problems, and has a high accuracy rate and good generalization performance. Using a trained error part prediction model to predict whether the text to be inspected is an error part is a common classification problem, and the use of a random forest algorithm can well solve the problem of sampling sample selection. Therefore, the model can be trained using sample data and algorithms, and the prediction accuracy of the error part prediction model can reach the use standard by adjusting the sample data and model parameters.

[0074] In some embodiments, when constructing an error part prediction model corresponding to a sample text set, the sample text set can be preprocessed to obtain preprocessed sample data, and the sample data can be split into a training set and a test set; then, a random forest composed of multiple decision trees can be constructed by sampling the training set with replacement to obtain a random forest model; finally, the random forest model can be tested using a test set. If the prediction accuracy of the random forest model does not meet the model evaluation requirements, the model parameters and sample characteristics of the random forest model are adjusted until the prediction accuracy of the random forest model meets the model evaluation requirements to obtain an error part prediction model.

[0075] According to some embodiments, when preprocessing a sample text set, a plurality of original feature data of each sample text may be extracted, and the original feature data may be processed and converted into sample data that can be recognized by the model.

[0076] For example, in the insurance field, n feature values ​​(such as insurance type, bill type (outpatient or inpatient), claim amount, disease type, years of experience of the data entry personnel, etc.) can be extracted for each case in the sample text set to obtain the original feature data; then, the original feature data can be processed and converted into sample data that can be recognized by the model.

[0077] The processed sample data can be formed into a two-dimensional array of 2m rows and (n+1) columns, as shown in the following formula:

[0078]

[0079] Among them, a1(x1) represents the first characteristic value of the first case, y(x1) represents the label value of the first case, and the label value is an error or a non-error. 0 can represent an error and 1 can represent a non-error.

[0080] It should be noted that random forest constructs multiple decision trees. When a sample needs to be predicted, the prediction results of each tree in the forest for the sample are counted, and then the final result is selected from these prediction results through voting. Randomness is reflected in two aspects: one is random feature selection, and the other is random sample selection, so that each tree in the forest has both similarities and differences.

[0081] According to some embodiments, in the process of constructing multiple decision trees by sampling the preprocessed sample data with replacement, a decision tree can be constructed based on the randomly selected samples and features. A decision tree is a tree structure, each leaf node of which corresponds to a label value, and non-leaf nodes correspond to a certain feature, and the sample is divided into several branches according to different values ​​of the sample on the feature.

[0082] In some embodiments, the termination conditions for decision tree bifurcation generally include:

[0083] The samples contained in the current node belong to the same category and do not need to be divided;

[0084] When features are exhausted and no features are available, the node with the largest number of labels in the current sample set is labeled accordingly to make it a leaf node;

[0085] The sample set of the current node is empty and cannot be divided. This happens because the sample data lacks this feature value, and the node is classified according to the label of the parent node.

[0086] Take a scenario as an example. Figure 2 A schematic diagram of a decision tree structure provided by an embodiment of the present disclosure. Figure 2 As shown, it shows the decision tree constructed using the training set corresponding to the insurance field.

[0087] According to some embodiments, in the process of adjusting the model parameters and sample features of the random forest model so that the prediction accuracy of the random forest model meets the model evaluation requirements, the adjustable parameters include but are not limited to at least one of the following:

[0088] n_estimators: The number of trees in the forest.

[0089] max_depth: The maximum depth of a single decision tree, which controls the growth of the tree.

[0090] min_samples_split: The minimum number of samples required to decide whether a node should be split.

[0091] min_samples_leaf: The minimum number of samples required at a leaf node.

[0092] max_features: determines the number of features considered each time a node is split.

[0093] random_state: Random seed, controls randomness.

[0094] In some embodiments, the model evaluation requirement may be, for example, that the model score of the random forest model is greater than a score threshold. The model score may be determined by the number of correct predictions and the number of incorrect predictions after the random forest model predicts the test set.

[0095] In some embodiments, when the sample data is split into a training set and a test set, it can be split according to a preset ratio. The preset ratio does not specifically refer to a fixed ratio, and the preset ratio can be adjusted according to the actual application scenario. For example, in the insurance field, the preset ratio can be 8:2.

[0096] S205, using the error part prediction model to predict the text to be inspected, and obtaining a prediction result;

[0097] S206, if the prediction result indicates that the text to be inspected is an error, then quality inspection is performed on the text to be inspected to obtain a quality inspection result corresponding to the text to be inspected;

[0098] It should be noted that when quality inspection is conducted on the quality inspection text, the quality inspection methods include but are not limited to manual quality inspection, intelligent quality inspection, etc.

[0099] S207, if the prediction result indicates that the document to be inspected is a non-error document, then determining the stratified sampling requirements corresponding to the document to be inspected;

[0100] It should be noted that stratified sampling requires the configuration of a series of sampling requirements, and the sampling requirements need to cover all types of texts to be inspected, that is, any text to be inspected will match a general requirement.

[0101] Taking a scenario as an example, in the insurance field, a collection of existing insurance claims cases can be obtained; the collection of existing insurance claims cases can be divided into multiple insurance claims case subsets according to the claim type and claim amount; based on the number of cases corresponding to the insurance claims case subsets and the total number of cases corresponding to the insurance claims case collection, the sampling ratio corresponding to the insurance claims case subsets is determined, and the corresponding relationship between the claim type and claim amount and the sampling ratio is obtained.

[0102] For example, suppose the number of cases corresponding to a certain claim type and claim amount is a x (x is 1, 2, 3, 4) or b y (y is 1, 2, 3, 4, 5), then the stratified sampling rules and sampling ratios can be shown in Table (1):

[0103]

[0104] It should be noted that the sampling ratio does not refer to a specific fixed ratio. For example, it can be updated at fixed intervals.

[0105] According to some embodiments, when the text to be inspected is an insurance claim case to be inspected, the claim type and claim amount corresponding to the insurance claim case to be inspected can be determined; based on the correspondence between the claim type and claim amount and the sampling ratio, the sampling ratio corresponding to the insurance claim case to be inspected is determined to obtain the stratified sampling requirements corresponding to the text to be inspected.

[0106] It should be noted that after the text to be inspected matches the stratified sampling requirements, equidistant sampling can be carried out according to the corresponding sampling ratio.

[0107] Taking a scenario as an example, the threshold of the number of intervals corresponding to the sampling ratio can be determined; the number of intervals of non-error parts with the same rules corresponding to the insurance claim cases to be inspected can be determined; if the number of intervals of non-error parts with the same rules is the threshold of the number of intervals, it is determined that the text to be inspected meets the requirements of stratified sampling, and the number of intervals of non-error parts with the same rules is initialized; if the number of intervals of non-error parts with the same rules is less than the threshold of the number of intervals, it is determined that the text to be inspected does not meet the requirements of stratified sampling, and the number of intervals of non-error parts with the same rules is increased by one. Therefore, it can prevent the error part prediction model from failing to identify the error parts for multiple consecutive times, resulting in no text to be inspected entering the quality inspection, and the rationality of the sampling inspection can be improved.

[0108] S208, when the text to be inspected meets the stratified sampling requirements, the text to be inspected is quality inspected to obtain a quality inspection result corresponding to the text to be inspected;

[0109] It should be noted that if the text to be inspected meets the requirements of stratified sampling, the text to be inspected can be directly closed.

[0110] For example, if the sampling ratio is 0.001, then every 1,000 cases will be selected for quality inspection. If there are cases with the same requirements within the interval number that have entered quality inspection through the error prediction model, the case will not be selected and will be closed directly; if there are no cases with the same rules within the interval distance that have entered quality inspection through the error prediction model, the case will be selected at the interval point to enter the quality inspection process.

[0111] S209, if the quality inspection result is passed, the quality inspection document is closed;

[0112] S210: If the quality inspection result is that the quality inspection has not been passed, the document to be inspected is returned as a document to be approved, and the document to be inspected is marked as an error document.

[0113] In summary, the method provided in this embodiment first obtains the text to be approved; if the text to be approved is approved, the text to be approved is used as the text to be inspected; therefore, the effect of obtaining the text to be inspected can be improved. Then, a sample text set is obtained; based on the random forest algorithm, an error prediction model corresponding to the sample text set is constructed; therefore, the accuracy of error prediction can be improved. Then, the error prediction model is used to predict the text to be inspected to obtain a prediction result; therefore, the lack of sampling randomness can be compensated and the probability of error occurrence can be effectively reduced. Next, if the prediction result indicates that the text to be inspected is an error, the text to be inspected is inspected to obtain the quality inspection result corresponding to the text to be inspected; if the prediction result indicates that the text to be inspected is a non-error, the stratified sampling requirements corresponding to the text to be inspected are determined; when the text to be inspected meets the stratified sampling requirements, the text to be inspected is inspected to obtain the quality inspection result corresponding to the text to be inspected; therefore, the problem of poor representativeness and large sampling error of simple random sampling can be solved. Different stratified sampling requirements are set for different types of texts, which can make the sampled text more representative and improve the rationality of the sample inspection. Finally, if the quality inspection result is passed, the text to be inspected is closed; if the quality inspection result is not passed, the text to be inspected is returned to the text to be approved, and the text to be inspected is marked as an error; therefore, the risk of error occurrence can be reduced and the quality of the overall text can be improved.

[0114] In order to implement the above embodiment, the present disclosure also proposes a text quality inspection device.

[0115] like Figure 3 As shown, the text quality inspection device 300 includes:

[0116] The text prediction unit 301 is used to obtain the text to be inspected, and use the error prediction model to predict the text to be inspected to obtain a prediction result;

[0117] A text quality inspection unit 302 is used to perform quality inspection on the text to be inspected if the prediction result indicates that the text to be inspected is an error, and obtain a quality inspection result corresponding to the text to be inspected;

[0118] The text sampling unit 303 is used to determine the stratified sampling requirements corresponding to the text to be inspected if the prediction result indicates that the text to be inspected is a non-error part, and if the text to be inspected meets the stratified sampling requirements, perform quality inspection on the text to be inspected to obtain the quality inspection result corresponding to the text to be inspected.

[0119] Optionally, before using the error part prediction model to predict the text to be inspected, the text prediction unit 301 is further used to:

[0120] Get a sample text collection;

[0121] Based on the random forest algorithm, an error prediction model corresponding to the sample text set is constructed.

[0122] Optionally, the text to be inspected is an insurance claim case to be inspected, and the text sampling unit 303 is used to determine the stratified sampling requirements corresponding to the text to be inspected, specifically for:

[0123] Determine the claim type and claim amount corresponding to the insurance claim case to be inspected;

[0124] According to the correspondence between claim type and claim amount and sampling ratio, the sampling ratio corresponding to the insurance claim cases to be inspected is determined, and the stratified sampling requirements corresponding to the texts to be inspected are obtained.

[0125] Optionally, after determining the stratified sampling requirements corresponding to the text to be inspected, the text sampling unit 303 is further used to:

[0126] Determine the interval number threshold corresponding to the sampling ratio;

[0127] Determine the number of non-error parts with the same rules corresponding to the insurance claim cases to be inspected;

[0128] If the number of intervals of non-error parts with the same rule is the interval number threshold, it is determined that the text to be inspected meets the stratified sampling requirements, and the number of intervals of non-error parts with the same rule is initialized;

[0129] If the number of intervals of non-error parts with the same rule is less than the interval number threshold, it is determined that the text to be quality inspected does not meet the stratified sampling requirements, and the number of intervals of non-error parts with the same rule is increased by one.

[0130] Optionally, before determining the sampling ratio corresponding to the insurance claim case to be inspected according to the correspondence between the claim type and the claim amount and the sampling ratio, the text sampling unit 303 is further used to:

[0131] Obtain a collection of existing insurance claims cases;

[0132] According to the claim type and claim amount, the insurance claim stock case set is divided into multiple insurance claim stock case subsets;

[0133] According to the number of cases corresponding to the subset of insurance claims stock cases and the total number of cases corresponding to the set of insurance claims stock cases, the sampling ratio corresponding to the subset of insurance claims stock cases is determined, and the corresponding relationship between the claim type and claim amount and the sampling ratio is obtained.

[0134] Optionally, when the text prediction unit 301 is used to obtain the text to be inspected, it is specifically used to:

[0135] Get the text to be approved;

[0136] If the text to be approved is approved, it will be used as the text to be inspected.

[0137] Optionally, after obtaining the quality inspection result corresponding to the text to be inspected, the text quality inspection unit 302 is further used to:

[0138] If the quality inspection result is passed, the quality inspection document will be closed;

[0139] If the quality inspection result is that it fails the quality inspection, the document to be inspected will be returned to the document to be approved, and the document to be inspected will be marked as an error.

[0140] It should be noted that the above explanation of the text quality inspection method embodiment is also applicable to the text quality inspection device of this embodiment, and will not be repeated here.

[0141] In summary, the device provided by the embodiment of the present disclosure, first, by adopting the error prediction model to predict the text to be inspected, it can make up for the lack of sampling randomness and effectively reduce the probability of error occurrence; then, by adopting stratified sampling requirements to perform stratified sampling, it can solve the problems of poor representativeness of simple random sampling samples and large sampling errors, and setting different stratified sampling requirements for different types of texts can make the inspected text more representative.

[0142] In order to implement the above embodiments, the present disclosure also proposes an electronic device, including: a processor, and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method provided by the above embodiments.

[0143] In order to implement the above embodiments, the present disclosure further proposes a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the methods provided by the above embodiments.

[0144] In order to implement the above embodiments, the present disclosure also proposes a computer program product, including a computer program, which implements the methods provided by the above embodiments when executed by a processor.

[0145] The collection, storage, use, processing, transmission, provision and disclosure of user personal information involved in this disclosure shall comply with the relevant laws and regulations and shall not violate public order and good morals.

[0146] It should be noted that personal information from users should be collected for legitimate and reasonable purposes and should not be shared or sold outside of these legitimate uses. In addition, such collection / sharing should be carried out after receiving the user's informed consent, including but not limited to notifying the user to read the user agreement / user notice and sign the agreement / authorization including authorization of relevant user information before the user uses the function. In addition, any necessary steps should be taken to protect and safeguard access to such personal information data and ensure that others who have access to personal information data comply with its privacy policy and procedures.

[0147] The present disclosure anticipates providing implementation schemes for users to selectively block the use or access of personal information data. That is, the present disclosure anticipates providing hardware and / or software to prevent or block access to such personal information data. Once the personal information data is no longer needed, risks can be minimized by limiting data collection and deleting the data. In addition, when applicable, such personal information is de-identified to protect the privacy of the user.

[0148] In the description of the aforementioned embodiments, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

[0149] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present disclosure, "plurality" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0150] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present disclosure includes additional implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present disclosure belong.

[0151] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute the instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or otherwise processing in a suitable manner if necessary, and then stored in a computer memory.

[0152] It should be understood that the various parts of the present disclosure can be implemented in hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0153] A person skilled in the art may understand that all or part of the steps in the above-mentioned embodiment method may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0154] In addition, each functional unit in each embodiment of the present disclosure may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0155] The storage medium mentioned above may be a read-only memory, a disk or an optical disk, etc. Although the embodiments of the present disclosure have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations of the present disclosure. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present disclosure.

Claims

1. A text quality inspection method, characterized in that: include: Acquire the text to be inspected, and use the error part prediction model to predict the text to be inspected to obtain a prediction result; If the prediction result indicates that the text to be inspected is an error, then the text to be inspected is inspected to obtain a quality inspection result corresponding to the text to be inspected; If the prediction result indicates that the text to be inspected is a non-error part, the stratified sampling requirements corresponding to the text to be inspected are determined, and if the text to be inspected meets the stratified sampling requirements, the text to be inspected is quality inspected to obtain the quality inspection result corresponding to the text to be inspected.

2. The method according to claim 1, characterized in that Before using the error part prediction model to predict the text to be inspected, the method further includes: Get a sample text collection; Based on the random forest algorithm, an error prediction model corresponding to the sample text set is constructed.

3. The method according to claim 1, characterized in that The document to be inspected is an insurance claim case to be inspected, and the stratified sampling requirements corresponding to the document to be inspected include: Determine the claim type and claim amount corresponding to the insurance claim case to be inspected; According to the corresponding relationship between the claim type and the claim amount and the sampling ratio, the sampling ratio corresponding to the insurance claim case to be inspected is determined, and the stratified sampling requirements corresponding to the text to be inspected are obtained.

4. The method according to claim 3, characterized in that After determining the stratified sampling requirements corresponding to the text to be inspected, the method further includes: Determine the interval number threshold corresponding to the sampling ratio; Determine the number of non-error parts intervals of the same rule corresponding to the insurance claim case to be inspected; If the number of intervals of non-error parts with the same rule is the interval number threshold, it is determined that the text to be inspected meets the stratified sampling requirement, and the number of intervals of non-error parts with the same rule is initialized; If the number of intervals of non-error parts with the same rule is less than the interval number threshold, it is determined that the text to be quality inspected does not meet the stratified sampling requirement, and the number of intervals of non-error parts with the same rule is increased by one.

5. The method according to claim 3, characterized in that: Before determining the sampling ratio corresponding to the insurance claim case to be inspected according to the corresponding relationship between the claim type and the claim amount and the sampling ratio, the method further includes: Obtain a collection of existing insurance claims cases; Dividing the insurance claim stock case set into multiple insurance claim stock case subsets according to claim type and claim amount; According to the number of cases corresponding to the subset of insurance claim stock cases and the total number of cases corresponding to the set of insurance claim stock cases, the sampling ratio corresponding to the subset of insurance claim stock cases is determined, and the corresponding relationship between the claim type and claim amount and the sampling ratio is obtained.

6. The method according to claim 1, characterized in that The obtaining of the text to be inspected includes: Get the text to be approved; If the document to be approved passes the approval, the document to be approved will be used as a document to be quality inspected.

7. The method according to claim 6, characterized in that After obtaining the quality inspection result corresponding to the text to be inspected, the method further includes: If the quality inspection result is that the quality inspection has been passed, the document to be inspected will be closed; If the quality inspection result is that the quality inspection fails, the document to be inspected will be returned as the document to be approved, and the document to be inspected will be marked as an error document.

8. A text quality inspection device, characterized in that: include: A text prediction unit, used to obtain the text to be inspected, and use the error prediction model to predict the text to be inspected to obtain a prediction result; A text quality inspection unit, configured to perform quality inspection on the text to be inspected if the prediction result indicates that the text to be inspected is an error, and obtain a quality inspection result corresponding to the text to be inspected; The text sampling unit is used to determine the stratified sampling requirements corresponding to the text to be inspected if the prediction result indicates that the text to be inspected is a non-error part, and to perform quality inspection on the text to be inspected if the text to be inspected meets the stratified sampling requirements to obtain the quality inspection result corresponding to the text to be inspected.

9. An electronic device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed by a processor.

Images