Judicial text evaluation system optimization method, judicial text evaluation method, judicial text evaluation device and judicial text evaluation system

Abstract

The invention relates to the technical field of text data, and provides a judicial text evaluation system optimization method, device and system and a judicial text evaluation method, device and system.The optimization method comprises the steps that the initial weight of evaluation parameters of a judicial text evaluation system is determined; identifying a first minority sample class in the first judicial text data set; performing multi-dimensional synthesis minority class oversampling processing on the first minority sample class to fill the sample quantity of the first minority sample class to form a second judicial text data set; identifying a second minority sample class in the second judicial text data set; determining an influence probability value of a weight influenced by the second minority sample class in the evaluation parameter; and adjusting the weight corresponding to the influence probability value greater than the probability threshold. According to the embodiment of the invention, the judicial text evaluation accuracy can be improved.

Description

Technical Field

[0001] The present specification relates to the technical field of text data processing, and in particular to a judicial text evaluation system optimization, a judicial text evaluation method, a device and a system. Background Art

[0002] In many industries (such as industrial informatization, judicial administration informatization, etc.), the evaluation of business processing results based on text information is often involved in order to comprehensively and systematically assess whether the business processing meets the expected requirements.

[0003] However, existing business evaluation systems have some limitations. These systems usually rely on evaluation parameters. However, the weights of evaluation parameters are generally set manually or determined based on general weight calculation methods (such as principal component analysis, entropy method, etc.). Compared with manually setting the weights of evaluation parameters, the method of determining the weights of evaluation parameters based on general weight calculation methods has obvious advantages in terms of efficiency and accuracy. However, this application study found that in the case of data imbalance, the accuracy of weight assignment of principal component analysis, entropy method, etc. is poor, which in turn affects the accuracy of business evaluation. Summary of the invention

[0004] The purpose of the embodiments of this specification is to provide a judicial text evaluation system optimization, judicial text evaluation method, device and equipment to improve the accuracy of judicial text evaluation.

[0005] To achieve the above objectives, on the one hand, the embodiments of this specification provide a method for optimizing a judicial text evaluation system, including:

[0006] Determine the initial weights of the evaluation parameters of the judicial text evaluation system;

[0007] Identify the first minority sample class in the first judicial text dataset;

[0008] Performing multi-dimensional synthetic minority class oversampling processing on the first minority sample class to fill the number of samples of the first minority sample class to form a second judicial text dataset;

[0009] identifying a second minority sample class in the second judicial text dataset;

[0010] Determining an influence probability value of a weight in the evaluation parameter that is influenced by the second minority sample class;

[0011] The weights corresponding to the impact probability values ​​greater than the probability threshold are adjusted.

[0012] In the judicial text evaluation system optimization method of the embodiment of this specification, the identifying of the first minority sample class in the first judicial text data set includes:

[0013] Clustering the first judicial text data set to obtain multiple judicial text classes;

[0014] Calculate the number of samples of each of the judicial text categories;

[0015] The judicial text class whose sample number is lower than the first target threshold is identified as the first minority sample class.

[0016] In the judicial text evaluation system optimization method of the embodiment of this specification, multi-dimensional synthetic minority class oversampling processing is performed on the first minority sample class, including:

[0017] Selecting a judicial text sample from the first minority sample class as a target sample;

[0018] Determine the nearest neighbor of the target sample in the three-dimensional space dimension, where the nearest neighbor is the judicial text sample in the first minority sample class;

[0019] A random function is used to add judicial text between the target sample and its neighbors as a new sample of the first minority sample class.

[0020] In the judicial text evaluation system optimization method of the embodiment of this specification, the use of a random function to add a judicial text between the target sample and its neighbors includes:

[0021]

[0022] Among them, X i is the text feature of the i-th newly added judicial text, N is the number of neighbors, X j is the text feature of the jth neighbor, rand(0,1) is a random function, and X' is the text feature of the target sample selected from the first minority sample class.

[0023] In the judicial text evaluation system optimization method of the embodiment of this specification, the identifying of the second minority sample class in the second judicial text data set includes:

[0024] Clustering the second judicial text data set to obtain multiple judicial text classes;

[0025] Calculate the number of samples of each of the judicial text categories;

[0026] A judicial text class whose sample quantity is lower than a second target threshold is identified as a first minority sample class; the second target threshold is greater than the first target threshold.

[0027] In the judicial text evaluation system optimization method of the embodiment of this specification, determining the influence probability value of the weight affected by the second minority sample class in the evaluation parameter includes:

[0028] According to the formula Calculating an influence probability value of a weight in the evaluation parameter that is influenced by the second minority sample class;

[0029] Among them, P(R i =k|x i ) is the input x i The corresponding output R i The probability value of the influence on the kth weight, R i is the output feature of the i-th group of judicial texts, x i is the input feature of the i-th group of judicial texts, exp is the exponential function, is the transpose of the i-th weight vector affected by the second minority sample class.

[0030] In the judicial text evaluation system optimization method of the embodiment of this specification, the weight corresponding to the influence probability value greater than the probability threshold is adjusted, including:

[0031] According to the formula Gain(Y,K)=H(Y)-H(Y|K), the conditional entropy gain of the weight corresponding to the influence probability value greater than the probability threshold is calculated;

[0032] Determining whether the conditional entropy gain converges;

[0033] When the conditional entropy gain does not converge, adjusting the weight corresponding to the influence probability value greater than the probability threshold until the currently obtained conditional entropy gain converges;

[0034] Wherein, Gain(Y,K) is the conditional entropy gain of the evaluation value Y under the influence of the Kth weight, H(Y) is the entropy value of the evaluation value Y, and H(Y|K) is the conditional entropy of the evaluation value Y under the influence of the Kth weight.

[0035] On the other hand, the embodiments of this specification also provide a judicial text evaluation method, including:

[0036] Obtain evaluation parameter data of the text to be adjudicated;

[0037] The evaluation parameter data is input into the judicial text evaluation system to obtain the evaluation value corresponding to the judicial text to be processed; wherein the judicial text evaluation system is a judicial text evaluation system optimized according to the above-mentioned optimization method.

[0038] On the other hand, the embodiment of this specification also provides a judicial text evaluation system optimization device, including:

[0039] The first determination module is used to determine the initial weights of the evaluation parameters of the judicial text evaluation system;

[0040] A first recognition module, used to identify a first minority sample class in a first judicial text dataset;

[0041] A sample filling module, used for performing multi-dimensional synthetic minority class oversampling processing on the first minority sample class to fill the number of samples of the first minority sample class to form a second judicial text data set;

[0042] A second recognition module, used for identifying a second minority sample class in the second judicial text dataset;

[0043] A second determination module, used to determine an influence probability value of a weight in the evaluation parameter that is influenced by the second minority sample class;

[0044] The weight adjustment module is used to adjust the weight corresponding to the influence probability value greater than the probability threshold.

[0045] On the other hand, the embodiment of this specification also provides a judicial text evaluation system, including:

[0046] An acquisition module is used to obtain evaluation parameter data of the text to be adjudicated;

[0047] An evaluation module generates an evaluation value corresponding to the judicial text according to the evaluation parameter data; wherein the judicial text evaluation system is a judicial text evaluation system optimized according to the above-mentioned optimization method.

[0048] On the other hand, an embodiment of the present specification further provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein when the computer program is executed by the processor, the instructions of the above method are executed.

[0049] On the other hand, an embodiment of the present specification further provides a computer storage medium on which a computer program is stored. When the computer program is executed by a processor of a computer device, the instructions of the above method are executed.

[0050] On the other hand, an embodiment of the present specification further provides a computer program product, wherein the computer program product includes a computer program, and when the computer program is executed by a processor of a computer device, the computer program executes instructions of the above method.

[0051] It can be seen from the technical solutions provided in the above embodiments of this specification that in the embodiments of this specification, by performing multi-dimensional synthetic minority class oversampling processing on the minority sample classes in the judicial text data set, sample learning of the minority sample classes is achieved, thereby reducing or avoiding the impact of data imbalance problems on the weight adjustment of evaluation parameters, and on this basis, identifying and quantitatively calculating the influence probability values ​​of the weights affected by the newly added minority sample classes, and adjusting the weights corresponding to the influence probability values ​​greater than the probability threshold; in this way, through small sample learning and automatic weight adjustment, the judicial text evaluation system can more accurately capture the key features of the judicial text, reduce errors caused by judicial text data imbalance or improper parameter settings, and thereby improve the judicial text evaluation accuracy of the judicial text evaluation system. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art description. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. In the drawings:

[0053] Figure 1 A schematic diagram showing the application environment of judicial text evaluation in some embodiments of this specification;

[0054] Figure 2 A flowchart showing a method for optimizing a judicial text evaluation system in some embodiments of this specification is shown;

[0055] Figure 3 Shows Figure 2 Flowchart of the method for identifying minority sample classes in a judicial text dataset;

[0056] Figure 4 A flowchart of performing multi-dimensional synthetic minority class oversampling processing on a minority sample class in an exemplary embodiment of the present specification is shown;

[0057] Figure 5 A schematic diagram showing a target sample and its neighbors in an exemplary embodiment of the present specification is shown;

[0058] Figure 6 Shows Figure 2 A schematic diagram of adjusting the weight corresponding to the influence probability value greater than the probability threshold in the method shown;

[0059] Figure 7 A flowchart showing a judicial text evaluation method in some embodiments of this specification is shown;

[0060] Figure 8It shows a structural block diagram of a judicial text evaluation system optimization device in some embodiments of this specification;

[0061] Fig. 9 A structural block diagram of a judicial text evaluation system in some embodiments of this specification is shown;

[0062] Fig.10 The structure block diagram of the computer device in some embodiments of the present specification is shown.

[0063] [Description of Reference Numerals]

[0064] 10. Client;

[0065] 20. Server;

[0066] 81. A first determination module;

[0067] 82. A first identification module;

[0068] 83. Sample filling module;

[0069] 84. A second identification module;

[0070] 85. A second determination module;

[0071] 86. Weight adjustment module;

[0072] 91. Get module;

[0073] 92. Evaluation module;

[0074] 1002. Computer equipment;

[0075] 1004. Processor;

[0076] 1006. Memory;

[0077] 1008. Driving mechanism;

[0078] 1010, input / output interface;

[0079] 1012. Input device;

[0080] 1014. Output device;

[0081] 1016. Presentation equipment;

[0082] 1018. Graphical user interface;

[0083] 1020. Network interface;

[0084] 1022. Communication link;

[0085] 1024. Communication bus. DETAILED DESCRIPTION

[0086] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0087] It should be noted that in the embodiments of this specification, the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved are all information and data that have been authorized by the user and fully authorized by all parties, that is, the acquisition, transmission, storage, use, and processing of data in the technical solution of this application are in compliance with the relevant provisions of relevant laws and regulations.

[0088] It should be noted that in the embodiments of the present application, certain software, components, models and other existing solutions in the industry may be mentioned, and they should be regarded as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of the present application, but it does not mean that the applicant has or will necessarily use the solution.

[0089] Figure 1 , a schematic diagram of the application environment in some embodiments of the present specification is shown; the application environment includes a client 10 and a server 20. The server 20 can obtain the evaluation parameter data of the judicial text to be processed from the client 10, input the evaluation parameter data into the judicial text evaluation system, and obtain the evaluation value corresponding to the judicial text to be processed; wherein the judicial text evaluation system is an optimized judicial text evaluation system based on artificially set evaluation parameter weights, or on the basis of setting evaluation parameter weights based on a general weight calculation method, so as to improve the accuracy of judicial text evaluation.

[0090] In some embodiments of the present specification, the client 10 may be a self-service terminal device, a mobile terminal (i.e., a smart phone), a display, a desktop computer, a tablet computer, a laptop computer, a digital assistant, or a smart wearable device, etc. Among them, the smart wearable device may include a smart bracelet, a smart watch, smart glasses, or a smart helmet, etc. Of course, the client 10 is not limited to the above-mentioned electronic devices with a certain entity, and it may also be software running in the above-mentioned electronic devices.

[0091] In some embodiments of the present specification, the server 20 may be an electronic device with computing and network interaction functions; or may be software running in the electronic device and providing business logic for data processing and network interaction.

[0092] In addition, it should be noted that Figure 1 What is shown is only an application environment provided by this specification. In actual application, there may be multiple clients 10 and multiple servers 20, and this specification does not impose any limitation.

[0093] This specification embodiment provides a judicial text evaluation system optimization method, which can be applied to the above-mentioned service side, referring to Figure 2 As shown, in some embodiments of this specification, the judicial text evaluation system optimization method may include the following steps:

[0094] Step 201: Determine the initial weights of the evaluation parameters of the judicial text evaluation system.

[0095] Step 202: Identify a first minority sample class in a first judicial text dataset.

[0096] Step 203: Perform a multidimensional synthetic minority over-sampling technique (M-SMOTE) on the first minority sample class to fill the number of samples of the first minority sample class to form a second judicial text dataset.

[0097] Step 204: Identify a second minority sample class in the second judicial text dataset.

[0098] Step 205: Determine the influence probability value of the weight of the evaluation parameter affected by the second minority sample class.

[0099] Step 206: Adjust the weights corresponding to the influence probability values ​​that are greater than the probability threshold.

[0100] In the embodiments of the present specification, by performing multi-dimensional synthetic minority class oversampling processing on the minority sample classes in the judicial text data set, sample learning of the minority sample classes is achieved, thereby reducing or avoiding the impact of data imbalance problems on the weight adjustment of evaluation parameters, and on this basis, identifying and quantitatively calculating the influence probability values ​​of the weights affected by the newly added minority sample classes, and adjusting the weights corresponding to the influence probability values ​​greater than the probability threshold; in this way, through small sample learning and automatic weight adjustment, the judicial text evaluation system can more accurately capture the key features of the judicial text, reduce errors caused by judicial text data imbalance or improper parameter settings, and thereby improve the judicial text evaluation accuracy of the judicial text evaluation system.

[0101] Moreover, since the various evaluation parameters of the judicial text evaluation system are closely correlated with each other, and as time goes by and the external environment changes, some evaluation parameters may lose their original relevance and importance, by utilizing the optimization method of the embodiment of this specification to regularly or dynamically optimize the judicial text evaluation system, not only can the judicial text evaluation accuracy of the judicial text evaluation system be improved, but also the performance stability of the judicial text evaluation system can be maintained.

[0102] It should be noted that the overall concept of the judicial text evaluation system optimization method of the embodiment of this specification can also be applied to the optimization of evaluation systems in other application fields to improve the evaluation accuracy of the corresponding evaluation system.

[0103] In some embodiments of this specification, an evaluation index system of a judicial text evaluation system can be constructed according to the primary evaluation parameters. For example, in an exemplary embodiment of this specification, a weighted sum judicial text evaluation system can be constructed according to the accuracy of fact finding, procedural compliance, and fairness of substantive results, and the initial weights of the accuracy of fact finding, procedural compliance, and fairness of substantive results are set. The evaluation parameters are rating indicators, which can also be referred to as indicators.

[0104] In other embodiments of this specification, an evaluation index system of a judicial text evaluation system can be constructed according to multi-level evaluation parameters. For example, in an exemplary embodiment of this specification, the accuracy of fact finding, procedural compliance, and fairness of substantive results can be used as first-level evaluation parameters, and on this basis, second-level evaluation parameters and third-level evaluation parameters are constructed and initial weights are set (for example, as shown in Table 1 below).

[0105] Table 1

[0106]

[0107] In some embodiments of the present specification, the initial weights of the evaluation parameters of the judicial text evaluation system can be obtained by parsing the pre-built judicial text evaluation system; wherein, the initial weights of the evaluation parameters can be considered to set the evaluation parameter weights, or to set the evaluation parameter weights based on a general weight calculation method; however, no matter which weight assignment method is adopted, there may be unreasonable weight assignments, especially in data imbalance scenarios, such unreasonable weight assignments may be more prominent. Among them, data imbalance refers to the significant difference in the number of samples of different categories in the data set, usually the number of samples of one category is much larger than that of another category; this imbalance may cause the model to be biased towards the majority class, and therefore, data imbalance will affect the model's generalization ability and recognition ability of the minority class.

[0108] refer to Figure 3As shown, in some embodiments of the present specification, identifying a first minority sample class in a first judicial text dataset may include:

[0109] Step 301: cluster the first judicial text data set to obtain multiple judicial text classes.

[0110] In some embodiments of the present specification, the first judicial text dataset is a judicial text dataset. In order to distinguish it from the judicial text dataset after multi-dimensional synthetic minority class oversampling processing, it is referred to as the first judicial text dataset, and the judicial text dataset after multi-dimensional synthetic minority class oversampling processing is referred to as the second judicial text dataset.

[0111] In some embodiments of the present specification, any suitable clustering algorithm (such as a partition-based clustering algorithm, a hierarchy-based clustering algorithm, a model-based clustering algorithm, etc.) can be used to cluster the first judicial text data set to obtain multiple judicial text classes.

[0112] Step 302: Calculate the number of samples of each judicial text category.

[0113] Step 303: Identify the judicial text class whose sample quantity is lower than the first target threshold as the first minority sample class.

[0114] Among them, the first target threshold can be customized as needed.

[0115] In some scenarios, small sample data may have a greater impact on the overall situation. For example, in the industrial field, normal vibration data indicates the stability of the mechanical system, but sudden small sample noise may also indicate some potential fault hazards. Therefore, it is important to identify a small number of sample classes (i.e., small sample classes).

[0116] refer to Figure 4 As shown, in some embodiments of the present specification, performing multi-dimensional synthetic minority class oversampling processing on the first minority sample class may include the following steps:

[0117] Step 401: Select a judicial text sample from the first minority sample class as a target sample.

[0118] In some embodiments of the present specification, a judicial text sample may be randomly selected from the first minority sample class as a target sample. Figure 5 In the exemplary embodiment shown, node K is the selected target sample.

[0119] Step 402: Determine the nearest neighbors of the target sample in the three-dimensional space dimension, where the nearest neighbors are judicial text samples in the first minority sample class.

[0120] Generally speaking, after clustering, the first judicial text dataset will form multiple different clusters (i.e., multiple judicial text classes), and different clusters will form adjacent relationships, and different nodes (i.e., judicial texts) in the same cluster will also form adjacent relationships; since the cluster is a three-dimensional structure (i.e., a three-dimensional structure), any node in the same cluster has adjacent nodes (i.e., neighbors). For example, in Figure 5 In the exemplary embodiment shown, the node K as the target sample has multiple neighbors (eg Figure 5 In some embodiments of the present specification, it is also possible to determine which nodes in the same cluster are used as neighbors of the target sample by using a distance threshold. For example, neighbors that are not less than a first number (e.g., 6) and not more than a second number (e.g., 12) can be selected to ensure feature diversity of the newly added samples while taking into account processing overhead.

[0121] Step 403: Use a random function to add judicial text between the target sample and its neighbors as a new sample of the first minority sample class.

[0122] In some embodiments of the present specification, adding a judicial text between the target sample and its neighbors using a random function may include:

[0123]

[0124] Among them, X i is the text feature of the i-th newly added judicial text, N is the number of neighbors, X j is the text feature of the jth neighbor, rand(0,1) is a random function and its value is 0 or 1, and X' is the text feature of the target sample selected from the first minority sample class.

[0125] In some embodiments of the present specification, when there are multiple minority sample classes obtained by clustering, multi-dimensional synthetic minority class oversampling processing can be performed for each minority sample class. After completing the multi-dimensional synthetic minority class oversampling processing for each minority sample class, a new judicial text data set (i.e., the second judicial text data set) can be formed. By performing multi-dimensional synthetic minority class oversampling processing on minority sample classes, the impact of data imbalance on the weight adjustment of evaluation parameters can be reduced or avoided.

[0126] In some embodiments of the present specification, the processing logic for identifying the second minority sample class in the second judicial text dataset may include the following steps:

[0127] (1) clustering the second judicial text data set to obtain multiple judicial text clusters;

[0128] (2) Calculate the number of samples for each of the judicial text categories;

[0129] (3) Identify the judicial text class whose sample number is lower than the second target threshold as the first minority sample class; the second target threshold is greater than the first target threshold.

[0130] After adding new samples, the difference in sample quantity between the minority sample class and the majority sample class decreases, but the difference between the minority sample class and the minority sample class may increase, that is, a new minority sample class may be formed due to the relative change in the sample quantity. Therefore, the minority sample class can be identified again; moreover, after adding new samples, the sample quantity of the minority sample class increases. On this basis, in order to facilitate obtaining a more accurate minority sample class, a larger target threshold can be used (that is, the second target threshold is greater than the first target threshold).

[0131] It can be seen that the processing logic for identifying the second minority sample class in the second judicial text data set is the same as the processing logic for identifying the first minority sample class in the first judicial text data set. Therefore, the specific technical details of identifying the second minority sample class in the second judicial text data set will not be repeated.

[0132] In some embodiments of the present specification, determining the influence probability value of the weight affected by the second minority sample class in the evaluation parameter may include: according to the formula Calculating an influence probability value of a weight in the evaluation parameter that is influenced by the second minority sample class;

[0133] Among them, P(R i =k|x i ) is the input x i The corresponding output R i The probability value of the influence on the kth weight, R i is the output feature of the i-th group of judicial texts, x i is the input feature of the i-th group of judicial texts, exp is the exponential function, is the transpose of the i-th weight vector affected by the second minority sample class.

[0134] During the weight assignment process, the large sample class (i.e., the majority sample class) and the small sample class (i.e., the minority sample class) may affect the indicator weight at the same time. Since the large sample class is insensitive to the amount of new data, its impact on the weight assignment is relatively stable; the small sample class is more affected by the new data; based on the impact probability value, the degree of influence of the determined small sample class on the weights of different evaluation parameters can be quantitatively identified, thereby providing an objective basis for the subsequent selection of which evaluation parameters to adjust their weights.

[0135] Since the newly added samples processed by multi-dimensional synthetic minority oversampling do not have the characteristics of linear distribution or continuous distribution, the above multi-classification logistic regression formula is used This enables a more efficient calculation of the impact probability.

[0136] refer to Figure 6 As shown, in some embodiments of the present specification, adjusting the weight corresponding to the influence probability value greater than the probability threshold may include the following steps:

[0137] Step 601, calculate the conditional entropy gain of the weight corresponding to the influence probability value greater than the probability threshold according to the formula Gain(Y,K)=H(Y)-H(Y|K); wherein Gain(Y,K) is the conditional entropy gain of the evaluation value Y under the influence of the Kth weight, H(Y) is the entropy value of the evaluation value Y, and H(Y|K) is the conditional entropy of the evaluation value Y under the influence of the Kth weight.

[0138] Step 602, determine whether the conditional entropy gain converges; if the conditional entropy gain does not converge, execute step 603; otherwise, end the weight adjustment process.

[0139] Step 603: Adjust the weight corresponding to the influence probability value greater than the probability threshold.

[0140] That is, when the conditional entropy gain does not converge, the weight corresponding to the influence probability value greater than the probability threshold is adjusted until the currently obtained conditional entropy gain converges; when the conditional entropy gain converges (that is, tends to be stable), it indicates that the uncertainty of the current weight value under the current conditions is the smallest, thereby obtaining the expected weight value under the current conditions.

[0141] This specification also provides a judicial text evaluation method, referring to Figure 7 As shown, in some embodiments of this specification, the judicial text evaluation method may include the following steps:

[0142] Step 701: Obtain evaluation parameter data of the text to be processed.

[0143] Step 702: input the evaluation parameter data into the judicial text evaluation system to obtain the evaluation value corresponding to the judicial text to be processed; wherein the judicial text evaluation system is a judicial text evaluation system optimized according to the above-mentioned judicial text evaluation system optimization method.

[0144] Since the above-mentioned judicial text evaluation system optimization method can improve the evaluation accuracy of the judicial text evaluation system, evaluating judicial texts based on the optimized judicial text evaluation system can improve the evaluation accuracy of judicial text evaluation.

[0145] Although the process flows described above include multiple operations that occur in a particular order, it should be clearly understood that the processes may include more or fewer operations, which may be performed sequentially or in parallel (eg, using parallel processors or a multi-threaded environment).

[0146] Corresponding to the above-mentioned judicial text evaluation system optimization method, this specification embodiment also provides a judicial text evaluation system optimization device, referring to Figure 8 As shown, in some embodiments of this specification, the judicial text evaluation system optimization device may include:

[0147] The first determination module 81 is used to determine the initial weights of the evaluation parameters of the judicial text evaluation system;

[0148] A first identification module 82, used to identify a first minority sample class in a first judicial text data set;

[0149] A sample filling module 83 is used to perform multi-dimensional synthetic minority class oversampling processing on the first minority sample class to fill the number of samples of the first minority sample class to form a second judicial text data set;

[0150] A second identification module 84, used to identify a second minority sample class in the second judicial text dataset;

[0151] A second determination module 85, configured to determine an influence probability value of a weight in the evaluation parameter that is influenced by the second minority sample class;

[0152] The weight adjustment module 86 is used to adjust the weight corresponding to the influence probability value greater than the probability threshold.

[0153] Corresponding to the above-mentioned judicial text evaluation method, the embodiment of this specification also provides a judicial text evaluation system, which is a judicial text evaluation system optimized according to the above-mentioned judicial text evaluation system optimization method, and can be configured on the above-mentioned server, referring to Fig. 9 As shown, in some embodiments of this specification, the judicial text evaluation system may include:

[0154] An acquisition module 91 is used to acquire evaluation parameter data of the text to be adjudicated;

[0155] The evaluation module 92 generates an evaluation value corresponding to the text to be judicially processed according to the evaluation parameter data.

[0156] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0157] The embodiments of this specification also provide a computer device. Fig.10 As shown, in some embodiments of this specification, the computer device 1002 may include one or more processors 1004, such as one or more central processing units (CPUs) or graphics processing units (GPUs), and each processing unit may implement one or more hardware threads. The computer device 1002 may also include any memory 1006, which is used to store any kind of information such as code, settings, data, etc. In a specific embodiment, the computer program on the memory 1006 and can be run on the processor 1004, when the computer program is run by the processor 1004, can execute the instructions of the judicial text evaluation system optimization method or the judicial text evaluation method described in any of the above embodiments. Non-limiting, for example, the memory 1006 may include any one or more combinations of the following: any type of RAM, any type of ROM, flash memory device, hard disk, optical disk, etc. More generally, any memory can use any technology to store information. Further, any memory can provide volatile or non-volatile retention of information. Further, any memory can represent a fixed or removable component of the computer device 1002. In one embodiment, when the processor 1004 executes the associated instructions stored in any memory or combination of memories, the computer device 1002 can perform any operation of the associated instructions. The computer device 1002 also includes one or more drive mechanisms 1008 for interacting with any memory, such as a hard disk drive mechanism, an optical disk drive mechanism, etc.

[0158] The computer device 1002 may also include an input / output interface 1010 (I / O) for receiving various inputs (via input device 1012) and for providing various outputs (via output device 1014). A specific output mechanism may include a presentation device 1016 and an associated graphical user interface 1018 (GUI). In other embodiments, the input / output interface 1010 (I / O), input device 1012, and output device 1014 may not be included, and the computer device 1002 may be used as a computer device in a network. The computer device 1002 may also include one or more network interfaces 1020 for exchanging data with other devices via one or more communication links 1022. One or more communication buses 1024 couple the components described above together.

[0159] The communication link 1022 may be implemented in any manner, for example, via a local area network, a wide area network (e.g., the Internet), a point-to-point connection, etc., or any combination thereof. The communication link 1022 may include any combination of hardwired links, wireless links, routers, gateway functions, name servers, etc. governed by any protocol or combination of protocols.

[0160] The methods, devices (systems), computer-readable storage media, and computer program products of some embodiments of this specification are described with reference to the corresponding flowcharts and / or block diagrams. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processor to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processor generate instructions for implementing the processes in the flowcharts. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0161] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processor to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0162] These computer program instructions can also be loaded onto a computer or other programmable data processor so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0163] In a typical configuration, a computer device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0164] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0165] Computer readable media include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computer device. As defined in this specification, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0166] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems or computer program products. Therefore, the embodiments of this specification may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the embodiments of this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0167] The present specification embodiments may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present specification embodiments may also be practiced in distributed computing environments where tasks are performed by remote processors connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0168] It should also be understood that in the embodiments of this specification, the term "and / or" is only a description of the association relationship of the associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0169] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0170] In the description of this specification, 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 embodiments of this specification. 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, without contradiction.

[0171] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A judicial text evaluation system optimization method, characterized in that: include: Determine the initial weights of the evaluation parameters of the judicial text evaluation system; Identify the first minority sample class in the first judicial text dataset; Performing multi-dimensional synthetic minority class oversampling processing on the first minority sample class to fill the number of samples of the first minority sample class to form a second judicial text dataset; identifying a second minority sample class in the second judicial text dataset; Determining an influence probability value of a weight in the evaluation parameter that is influenced by the second minority sample class; The weights corresponding to the impact probability values ​​greater than the probability threshold are adjusted.

2. The judicial text evaluation system optimization method according to claim 1, characterized in that: The identifying a first minority sample class in a first judicial text dataset comprises: Clustering the first judicial text data set to obtain multiple judicial text classes; Calculate the number of samples of each of the judicial text categories; The judicial text class whose sample number is lower than the first target threshold is identified as the first minority sample class.

3. The judicial text evaluation system optimization method according to claim 1, characterized in that: Performing multi-dimensional synthetic minority class oversampling processing on the first minority sample class includes: Selecting a judicial text sample from the first minority sample class as a target sample; Determine the nearest neighbor of the target sample in the three-dimensional space dimension, where the nearest neighbor is the judicial text sample in the first minority sample class; A random function is used to add judicial text between the target sample and its neighbors as a new sample of the first minority sample class.

4. The judicial text evaluation system optimization method according to claim 3, characterized in that: The step of adding a judicial text between the target sample and its neighbors by using a random function includes: Among them, X i is the text feature of the i-th newly added judicial text, N is the number of neighbors, X j is the text feature of the jth neighbor, rand(0,1) is a random function, and X' is the text feature of the target sample selected from the first minority sample class.

5. The judicial text evaluation system optimization method according to claim 2, characterized in that: The identifying of the second minority sample class in the second judicial text dataset comprises: Clustering the second judicial text data set to obtain multiple judicial text classes; Calculate the number of samples of each of the judicial text categories; A judicial text class whose sample quantity is lower than a second target threshold is identified as a first minority sample class; the second target threshold is greater than the first target threshold.

6. The judicial text evaluation system optimization method according to claim 1, characterized in that: Determining the influence probability value of the weight of the evaluation parameter affected by the second minority sample class includes: According to the formula Calculating an influence probability value of a weight in the evaluation parameter that is influenced by the second minority sample class; Among them, P(R i =k|x i ) is the input x i The corresponding output R i The probability value of the influence on the kth weight, R i is the output feature of the i-th group of judicial texts, x i is the input feature of the i-th group of judicial texts, exp is the exponential function, is the transpose of the i-th weight vector affected by the second minority sample class.

7. The judicial text evaluation system optimization method according to claim 1, characterized in that: The weights corresponding to the impact probability values ​​greater than the probability threshold are adjusted, including: According to the formula Gain(Y,K)=H(Y)-H(Y|K), the conditional entropy gain of the weight corresponding to the influence probability value greater than the probability threshold is calculated; Determining whether the conditional entropy gain converges; When the conditional entropy gain does not converge, adjusting the weight corresponding to the influence probability value greater than the probability threshold until the currently obtained conditional entropy gain converges; Wherein, Gain(Y,K) is the conditional entropy gain of the evaluation value Y under the influence of the Kth weight, H(Y) is the entropy value of the evaluation value Y, and H(Y|K) is the conditional entropy of the evaluation value Y under the influence of the Kth weight.

8. A judicial text evaluation method, characterized in that: include: Obtain evaluation parameter data of the text to be adjudicated; The evaluation parameter data is input into a judicial text evaluation system to obtain an evaluation value corresponding to the judicial text to be processed; wherein the judicial text evaluation system is a judicial text evaluation system optimized according to the method described in any one of claims 1 to 7.

9. A judicial text evaluation system optimization device, characterized in that: include: The first determination module is used to determine the initial weights of the evaluation parameters of the judicial text evaluation system; A first recognition module, used to identify a first minority sample class in a first judicial text dataset; A sample filling module, used for performing multi-dimensional synthetic minority class oversampling processing on the first minority sample class to fill the number of samples of the first minority sample class to form a second judicial text data set; A second recognition module, used for identifying a second minority sample class in the second judicial text dataset; A second determination module, used to determine an influence probability value of a weight in the evaluation parameter that is influenced by the second minority sample class; The weight adjustment module is used to adjust the weight corresponding to the influence probability value greater than the probability threshold.

10. A judicial text evaluation system, characterized in that: include: An acquisition module is used to obtain evaluation parameter data of the text to be adjudicated; An evaluation module generates an evaluation value corresponding to the judicial text according to the evaluation parameter data; wherein the judicial text evaluation system is a judicial text evaluation system optimized according to the method described in any one of claims 1 to 7.

11. A computer device comprising a memory, a processor, and a computer program stored in the memory, characterized in that: When the computer program is executed by the processor, the computer program executes the instructions of the method according to any one of claims 1 to 8.

12. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor of a computer device, the computer program executes the instructions of the method according to any one of claims 1 to 8.

13. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor of a computer device, the computer program executes instructions of the method according to any one of claims 1 to 8.

Images