WO2021042560A1 - Case auxiliary information prompting method, device, storage medium and server - Google Patents

Abstract

Provided are a case auxiliary information prompting method, a device, a storage medium and a server. The method comprises: obtaining a judgment result text of a case (101); extracting sentence information and applicable laws of the case from the judgment result text (102); extracting case information of the case from a legal document related to the case (103); inputting the case information into a pre-constructed first neural network model, and determining the reasonability of the sentence information according to an output result of the first neural network model (104); inputting the case information into a pre-constructed second neural network model, and determining whether the applicable laws are correct according to an output result of the second neural network model (105); and if the reasonability of the sentence information is lower than a preset threshold or the applicable laws are incorrect, outputting auxiliary information for indicating that the case judgment is wrong (106). With this method, a judge can be assisted in supervising and rechecking a judgment result of a case.

Description

Method, device, storage medium and server for prompting case auxiliary information

This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on September 3, 2019, the application number is 201910829464.6, and the application name is "a method, device, storage medium and server for prompting auxiliary information in a case", all of which The content is incorporated in this application by reference.

Technical field

This application relates to the field of computer technology, and in particular to a method, device, storage medium and server for prompting case auxiliary information.

Background technique

At present, the trial process of a case depends to a large extent on the subjective consciousness and opinions of the judge. If the judge makes a mistake during the trial of the case, it may lead to the occurrence of unjust, false and wrong cases. Therefore, how to assist the judge in supervising and reviewing the judgment results of the case and reducing the occurrence of misjudgment has become a problem for technicians to consider.

technical problem

In view of this, the embodiments of the present application provide a method, device, storage medium and server for prompting case auxiliary information, which can assist judges in supervising and reviewing the judgment results of the case, and reduce the occurrence of misjudgment behaviors.

Technical solutions

The first aspect of the embodiments of the present application provides a method for prompting case auxiliary information, including:

Obtain the text of the judgment result of the case;

Extract the sentencing information and applicable laws and regulations of the case from the text of the judgment result;

Extract information on the merits of the case from legal documents related to the case;

The case information is input into a pre-built first neural network model, and the reasonableness of the sentencing information is determined by the output result of the first neural network model, and the first neural network model is determined by the information corresponding to the sentencing information. The case data of a plurality of first sample cases are trained as a training set, and the first neural network model obtains the first similarity by matching the case information with the case data of the first sample case, and according to The magnitude of the first degree of similarity determines the reasonableness of the sentencing information;

The case information is input to a pre-built second neural network model, and the output result of the second neural network model is used to determine whether the applicable law is correct, and the second neural network model is determined by corresponding to the applicable law The case data of the multiple second sample cases is obtained as a training set training, and the second neural network model obtains the second similarity by matching the case information and the case data of the second sample case. The magnitude of the second degree of similarity determines whether the applicable law is correct;

If the reasonableness of the sentencing information is lower than a preset threshold or the applicable legal provisions are incorrect, output auxiliary information indicating that the case is incorrectly judged.

The second aspect of the embodiments of the present application provides a user portrait device for judges, including:

Judgment result obtaining module, used to obtain the text of the judgment result of the case;

The judgment information extraction module is used to extract the sentencing information and applicable laws and regulations of the case from the text of the judgment result;

The case information extraction module is used to extract the case information of the case from the legal documents related to the case;

The sentencing rationality determination module is configured to input the case information into a pre-built first neural network model, and determine the rationality of the sentencing information through the output result of the first neural network model, the first neural network model The case data of a plurality of first sample cases corresponding to the sentencing information is obtained as a training set for training, and the first neural network model matches the case information with the case data of the first sample case Obtain the first degree of similarity, and determine the reasonableness of the sentencing information according to the magnitude of the first degree of similarity;

The applicable law article determination module is used to input the case information into a pre-built second neural network model, and determine whether the applicable law article is correct according to the output result of the second neural network model, and the second neural network model The case data of a plurality of second sample cases corresponding to the applicable law is obtained as a training set for training, and the second neural network model is obtained by matching the case information with the case data of the second sample case A second degree of similarity, and determine whether the applicable law is correct according to the magnitude of the second degree of similarity;

The auxiliary information output module is configured to output auxiliary information for indicating that the judgment of the case is incorrect if the reasonableness of the sentencing information is lower than a preset threshold or the applicable law is incorrect.

A third aspect of the embodiments of the present application provides a computer non-volatile readable storage medium, the computer non-volatile readable storage medium stores computer readable instructions, and the computer readable instructions are executed by a processor When executed, the steps of the method for prompting case auxiliary information proposed in the first aspect of the embodiments of the present application are implemented.

In a fourth aspect of the embodiments of the present application, a server is provided, including a memory, a processor, and computer-readable instructions stored in the memory and running on the processor, and the processor executes the computer When the instructions are readable, the steps of the method for prompting case auxiliary information as proposed in the first aspect of the embodiments of the present application are implemented.

Beneficial effect

In the method for prompting case auxiliary information proposed in the embodiment of this application, after the judge finishes the trial of a case and produces the corresponding judgment result text, the system can extract the corresponding sentencing information and applicable laws and regulations from the judgment result text, and then retrieve it from The case information of the case is extracted from the relevant legal documents, and the case information is input into the neural network model corresponding to the sentencing information and the neural network model corresponding to the applicable law to determine whether the sentencing of the case is reasonable and applicable Whether the law is correct. If the sentencing is unreasonable or the applicable laws and regulations are incorrect, auxiliary information indicating that the judgment of the case is wrong will be output, so as to assist the judge in supervising and reviewing the judgment of the case, and reducing the occurrence of misjudgment.

Description of the drawings

Fig. 1 is a flowchart of a first embodiment of a method for prompting case auxiliary information provided by an embodiment of the present application;

2 is a flowchart of a second embodiment of a method for prompting case auxiliary information provided by an embodiment of the present application;

FIG. 3 is a structural diagram of an embodiment of a device for prompting case auxiliary information provided by an embodiment of the present application;

Fig. 4 is a schematic diagram of a server provided by an embodiment of the present application.

Implementation of this application

Referring to Fig. 1, a first embodiment of a method for prompting case auxiliary information in an embodiment of the present application includes:

101. Obtain the text of the judgment result of the case;

When a case is completed and the judgment result is produced, the server will obtain the text of the judgment result of the case. The case can be a criminal case, a civil case, and other cases that need to be tried, and the text of the judgment result can be an electronic version of the case judgment, which records the trial result of the case.

102. Extract the sentencing information and applicable laws and regulations of the case from the text of the judgment result;

After obtaining the text of the judgment result, the server will extract the sentencing information and applicable laws of the case from the text of the judgment result. Specifically, sentencing information can include fixed-term imprisonment of less than three years, criminal detention less than three years, surveillance of less than three years, fixed-term imprisonment of more than three years and less than 10 years, fixed-term imprisonment of more than 10 years, life imprisonment, death penalty, etc. The applicable law is a judge's judgment The legal provisions adopted in the case, such as "Article 9 of the Criminal Law of the People's Republic of China...", "Article 6 of the Marriage Law", etc. In actual operation, the method of keyword detection and extraction can be used to extract the sentencing information and applicable laws and regulations of the case from the text of the judgment result. For example, if the keyword "death penalty" is detected, the sentencing information can be extracted as "death penalty", and so on.

103. Extract the case information of the case from legal documents related to the case;

Then, the server extracts the case information of the case from the legal documents related to the case. These legal documents record the specific facts of the case, such as the cause, process and consequences of the case, and the testimony of relevant personnel. In actual operation, keyword detection and extraction can also be used to extract case information from the legal document.

104. Input the case information into a pre-built first neural network model, and determine the reasonableness of the sentencing information based on the output result of the first neural network model;

After extracting the case information, the server inputs the case information into the pre-built first neural network model, and the rationality of the sentencing information is determined by the output result of the first neural network model. The first neural network model is obtained by training the case data of multiple sample cases corresponding to the sentencing information as a training set. For example, if the sentencing information is "life imprisonment", the first neural network model is changed from sentenced to " The case data of multiple historical sample cases of life imprisonment is obtained as training set training. Specifically, after the first neural network model obtains the input case information, it will match the case information with the case data of each historical sample case corresponding to the sentencing information, and obtain a similarity as the output As a result, the reasonableness of the sentencing information can be determined according to the similarity. The greater the similarity, the higher the corresponding reasonableness. For example, if the similarity is 90%, the reasonableness is determined to be 80 (maximum 100); if If the similarity is 80%, the reasonableness is determined to be 60 and so on.

105. Input the case information into a pre-built second neural network model, and determine whether the applicable law is correct according to the output result of the second neural network model;

In addition, the server will also input the case information into a pre-built second neural network model, and determine whether the applicable law is correct through the output result of the second neural network model. The second neural network model is determined by the corresponding The case data of multiple sample cases mentioned in the applicable law are obtained as training set training. For example, if the applicable law is “Article 9 of the Criminal Law”, the second neural network model is determined to be applicable to the “Criminal Law Article 9”. The case data of multiple historical sample cases of "bar" is obtained as training set training.

Specifically, after the second neural network model obtains the input case information, it will match the case information with the case data of each historical sample case corresponding to the applicable law to obtain a similarity as Output results, and then determine whether the applicable law is correct according to the degree of similarity. For example, if the similarity exceeds a certain preset threshold, the applicable law is determined to be correct; if the similarity does not exceed the threshold, then determine The applicable law is wrong.

Preferably, both the first neural network model and the second neural network model can use a two-classifier based on a BP neural network. The network consists of an input layer, at least one hidden layer and an output layer. The sample set (Ai, Bi) is sent to the network, the error D between the actual output O of the network and the sample value Bi is calculated, and the weight matrix W of the network is adjusted according to D, so that the error D does not exceed the specified range through continuous training. For the neural network model, its mathematical expression can be y _i =f(net _i ), where,

W is the weight of the neuron connection, and θ is the preset bias value. Among them, the activation function f can choose the sigmoid function. When training the BP neural network model, the gradient descent BP training function can be used, that is, after the sample is input to the neural network to obtain the output value, the error value between the output value and the expected value is calculated, and the The error value is input to the hidden layer layer by layer through the back-propagation algorithm, and the deviation of the parameters of each layer is calculated, and then the parameters are adjusted by moving a specific step length until the parameters are adjusted to an appropriate level, that is, the error is Within the acceptable range. After the classifier training is completed, input the case information into the classifier, and obtain the output result (0 or 1) of the classifier to determine whether the sentencing information and the applicable laws are reasonable. For example, if the classifier outputs 0, it means The reasonableness of the sentencing information is too low, or the applicable law is wrong; output 1 indicates that the sentencing information is reasonably reasonable or the applicable law is correct.

106. If the reasonableness of the sentencing information is lower than a preset threshold or the applicable law is incorrect, output auxiliary information indicating that the case is incorrectly judged.

If the reasonableness of the sentencing information is lower than a preset threshold or the applicable law is incorrect, the server will output auxiliary information indicating that the judgment of the case is incorrect. For example, it is possible to construct and output auxiliary information of "The sentencing information of this case is too light (too heavy), it is recommended to amend it to..." or "The applicable law of this case is wrong, please refer to the law..." to remind the trial of the case The judge's attention can assist the judge in supervising and reviewing the judgment results of the case, and reducing the occurrence of misjudgment.

In the method for prompting case auxiliary information proposed in the embodiment of this application, after the judge finishes the trial of a case and produces the corresponding judgment result text, the system can extract the corresponding sentencing information and applicable laws and regulations from the judgment result text, and then retrieve it from The case information of the case is extracted from the relevant legal documents, and the case information is input into the neural network model corresponding to the sentencing information and the neural network model corresponding to the applicable law to determine whether the sentencing of the case is reasonable and applicable Whether the law is correct. If the sentencing is unreasonable or the applicable laws and regulations are incorrect, auxiliary information indicating that the judgment of the case is wrong will be output, so as to assist the judge in supervising and reviewing the judgment of the case, and reducing the occurrence of misjudgment.

Referring to Fig. 2, a second embodiment of a method for prompting case auxiliary information in an embodiment of the present application includes:

201. Obtain the text of the judgment result of the case;

Step 201 is the same as step 101. For details, please refer to the related description of step 101.

202. Detect sentencing keywords and legal keywords recorded in a pre-built keyword database from the text of the judgment result;

After obtaining the judgment result text of the case, the server detects from the judgment result text the sentenced sentence keywords and legal clause keywords recorded in the pre-built keyword database. On the server side, you can pre-collect all kinds of commonly used sentencing keywords and legal keywords, such as fixed-term imprisonment of more than three years and less than ten years, fixed-term imprisonment of more than ten years, life imprisonment, death penalty, marriage law article 6 and labor law second Article, Article 9 of the Criminal Law...etc, store these keywords in a keyword database. Then, it is detected from the judgment result text whether the keywords recorded in the keyword database are included.

203. Separately count the word frequency of each sentencing keyword detected and the word frequency of each legal article keyword;

Then, the word frequency of each sentenced keyword and the word frequency of each legal article keyword are counted separately, that is, the number of times each keyword appears in the text of the judgment result.

204. Determine the keyword with the highest word frequency among the detected sentencing keywords as the extracted sentencing information;

205. Determine the keyword with the highest word frequency among the detected legal article keywords as the extracted applicable legal article;

Next, the keyword with the highest word frequency among the detected sentence keywords is determined as the extracted sentence information, and the keyword with the highest word frequency among the detected legal keywords is determined as the extracted applicable legal article. For example, if the sentencing keyword "life imprisonment" is detected once, "imprisonment of more than ten years" once, and "death sentence" 4 times, the sentencing keyword "death penalty" is determined as the extracted sentencing information.

206. Extract case information of the case from legal documents related to the case;

Further, the keyword database also records case keywords, and step 206 may include:

(1) Use the pre-built word segmentation model to segment the content of the legal document to obtain the target phrase set;

(2) The case key words detected from the target phrase set are determined as the extracted case information of the case.

On the server side, various commonly used case keywords such as snatching, theft, and bribery can be collected in advance, and these keywords can be stored in the keyword database. When extracting the case information of the case from a legal document, a pre-built word segmentation model can be used to segment the content of the legal document to obtain a target phrase set, and then the case information detected from the target phrase set The keywords are determined as the extracted case information of the case.

Specifically, for step (1), it may include:

(1.1) Detect preset target keywords from the content of the legal document;

(1.2) Determine the type of the legal document according to the detected target keywords;

(1.3) Select the word segmentation model corresponding to the type of the legal document to segment the content of the legal document.

After obtaining the legal document, the preset target keywords are detected from the content of the legal document. Perform a full-text search on the legal document to determine whether the content of the legal document contains certain preset target keywords. These target keywords can be used to determine the type of the legal document, such as criminal, civil, administrative, and first-instance , Second trial and other keywords. Then, the type of the legal document is determined according to the detected target keywords. For example, if the detected target keyword is the keyword corresponding to the "criminal" type legal document, the type of the legal document is determined to be "criminal"; the detected target keyword is the key corresponding to the "first instance" type legal document Word, the type of the legal document is determined to be "first instance". In addition, the type of legal document can also be determined by identifying the legal reasons in the legal document, that is, the law applicable to the judgment. That is, when searching the full text of a legal document, you can first identify the book title number in the document, extract the legal article name from the book title number, and then determine whether the legal document is a criminal, civil, or other type based on the legal article name.

After determining the type of the legal document, a word segmentation model corresponding to the type of the legal document is selected to segment the content of the legal document to obtain a target phrase set. The terms, document structure, and paragraph composition of different legal document types are quite different. Therefore, different word segmentation models are generated in advance according to different document types, and the appropriate word segmentation model is selected for word segmentation during application, which helps to improve word segmentation. Reasonable, so as to obtain more accurate word segmentation results.

Specifically, for step (1.2), it can include:

(1.2.1) If the number of detected target keywords is one, determine the type of the legal document according to the detected target keywords;

(1.2.2) If the number of detected target keywords is more than two, the detected target keywords are divided into more than one keyword combination, and the legal document is determined according to the keyword combination Each keyword combination contains more than two target keywords.

If the number of detected target keywords is only one, the type of the legal document is determined directly based on the target keyword; if the number of detected target keywords is more than two, these target keywords are divided into more than one Then, the type of the legal document is determined according to the keyword combinations obtained by the division, and each of the keyword combinations includes more than two of the target keywords. For example, if the target keyword is "civil", the type of the legal document is directly determined as "civil"; the keyword combination obtained by dividing is "civil, first instance", then the type of the legal document can be determined as "civil at first instance". "Types of.

Specifically, for step (1.2.2), the determining the type of the legal document according to the combination of keywords may include:

(1.2.2.1) If the number of the keyword combination is one, the type of the legal document is determined according to the keyword combination;

(1.2.2.2) If the number of the keyword combinations is more than two, the text distances in the legal documents of the target keywords contained in each keyword combination are counted separately, and the text distance is calculated according to the text The keyword combination with the smallest distance determines the type of the legal document.

If the number of keyword combinations obtained by the division is one, the type of the legal document is directly determined according to the keyword combination. If the number of the keyword combinations obtained by the division is more than two, then each key combination can be counted separately. The text distance of each target keyword contained in the word combination in the legal document, that is, the number of characters between the two keywords in the content of the legal document, and finally the key word combination with the smallest text distance is determined. State the types of legal documents. Through this setting, it can deal with application scenarios where the semantics are complex and the type of legal document cannot be judged by a single keyword.

For example, in the content of the legal document “Revocation of the Civil Judgment No. 1, 4, 5, and 6 of the Beijing Second Intermediate People’s Court (2017) Jing 02 Min Zhong 10216 and the People’s Court of Dongcheng District, Beijing The court (2016) Jing 0101 Min Chu No. 7939 Civil Judgment No. 1; upholds the Beijing Second Intermediate People's Court (2017) Jing 02 Min Zhong No. 10216 Civil Judgment No. 2 and Beijing Dongcheng District People's Court (2016) Jing 0101 In Minchu No. 7939 Civil Judgment No. 2", "Revocation", "Civil Judgment No. 1, 4, 5, and 6", "Maintenance" and "Civil Judgment No. 2" are selected. The target keyword, the combination is the keyword combination "revoke...the second item of civil judgment", and the keyword combination "maintain...the second item of civil judgment". According to statistics, the text distance of the keyword in the keyword combination "Revoke...Civil Judgment 2" is greater than the text distance of "Maintain...Civil Judgment 2", then select "Maintain...Civil Judgment 2" The combination of keywords determines the type of the legal document.

207. Input the case information into a pre-built first neural network model, and determine the reasonableness of the sentencing information based on the output result of the first neural network model;

208. Input the case information into a pre-built second neural network model, and determine whether the applicable law is correct according to the output result of the second neural network model;

209. If the reasonableness of the sentencing information is lower than a preset threshold or the applicable law is incorrect, output auxiliary information indicating that the judgment of the case is incorrect.

Steps 207-209 are the same as steps 104-106. For details, please refer to the relevant descriptions of steps 104-106.

Compared with the first embodiment of this application, this embodiment proposes a specific method for extracting sentencing information and applying legal provisions from the text of the judgment result of a case.

It should be understood that the size of the sequence number of each step in the foregoing embodiment does not mean the order of execution. The execution sequence of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present application.

Referring to FIG. 3, an embodiment of a device for prompting case auxiliary information in an embodiment of the present application includes:

The judgment result obtaining module 301 is used to obtain the judgment result text of the case;

The judgment information extraction module 302 is used to extract the sentencing information and applicable laws and regulations of the case from the text of the judgment result;

The case information extraction module 303 is used to extract case information of the case from legal documents related to the case;

The sentencing rationality determination module 304 is configured to input the case information into a pre-built first neural network model, and determine the rationality of the sentencing information based on the output result of the first neural network model, the first neural network The model is obtained by training the case data of a plurality of first sample cases corresponding to the sentencing information as a training set, and the first neural network model is performed by combining the case information and the case data of the first sample case The first degree of similarity is obtained by matching, and the reasonable degree of the sentencing information is determined according to the magnitude of the first degree of similarity;

The applicable law clause determining module 305 is configured to input the case information into a pre-built second neural network model, and determine whether the applicable law clause is correct according to the output result of the second neural network model, and the second neural network The model is obtained by training the case data of a plurality of second sample cases corresponding to the applicable law as a training set, and the second neural network model matches the case information with the case data of the second sample case Obtain the second degree of similarity, and determine whether the applicable law is correct according to the magnitude of the second degree of similarity;

The auxiliary information output module 306 is configured to output auxiliary information indicating that the judgment of the case is incorrect if the reasonableness of the sentencing information is lower than a preset threshold or the applicable law is incorrect.

Further, the decision information extraction module may include:

The keyword detection unit is used to detect from the judgment result text the sentencing keywords and legal keywords recorded in the pre-built keyword database;

The word frequency statistics unit is used to separately count the word frequency of each sentenced keyword and the word frequency of each legal article keyword detected;

The sentencing information determining unit is configured to determine the keyword with the highest word frequency among the detected sentencing keywords as the extracted sentencing information;

The applicable legal clause determining unit is used to determine the keyword with the highest word frequency among the detected legal clause keywords as the extracted applicable legal clause.

Further, the keyword database also records keywords on the case, and the case information extraction module may include:

The word segmentation unit is used to segment the content of the legal document using a pre-built word segmentation model to obtain a target phrase set;

The case information determining unit is configured to determine the case key words detected from the target phrase set as the extracted case information of the case.

Furthermore, the word segmentation unit may include:

The keyword detection subunit is used to detect preset target keywords from the content of the legal document;

The legal document type determination subunit is used to determine the type of the legal document according to the detected target keywords;

The word segmentation model selection subunit is used to select a word segmentation model corresponding to the type of the legal document to segment the content of the legal document.

Furthermore, the legal document type determination subunit may include:

The first type determining grandchild unit is configured to determine the type of the legal document according to the detected target keyword if the number of detected target keywords is one;

The second type determines the grandchildren unit, which is used to divide the detected target keyword into more than one keyword combination if the number of detected target keywords is more than two, and determine according to the keyword combination For the type of the legal document, each of the keyword combinations includes more than two of the target keywords.

An embodiment of the present application also provides a server, including a memory, a processor, and computer-readable instructions stored in the memory and capable of running on the processor, and the processor executes the computer-readable instructions to implement Figure 1 or Figure 2 shows the steps of any kind of case auxiliary information presentation method.

Fig. 4 is a schematic diagram of a server provided by an embodiment of the present application. As shown in FIG. 4, the server 4 in this embodiment includes a processor 40, a memory 41, and computer-readable instructions 42 stored in the memory 41 and running on the processor 40. When the processor 40 executes the computer-readable instructions 42, the steps in the above-mentioned application promotion effect evaluation method embodiments, such as steps 101 to 106 shown in Fig. 1, are implemented. Alternatively, when the processor 40 executes the computer-readable instructions 42, the functions of the modules/units in the foregoing device embodiments are implemented, for example, the functions of the modules 301 to 306 shown in FIG. 3.

Exemplarily, the computer-readable instructions 42 may be divided into one or more modules/units, and the one or more modules/units are stored in the memory 41 and executed by the processor 40, To complete this application. The one or more modules/units may be a series of computer-readable instruction segments capable of completing specific functions, and the instruction segments are used to describe the execution process of the computer-readable instructions 42 in the server 4.

The processor 40 may be a central processing unit (Central Processing Unit, CPU), or other general-purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor or the like.

The memory 41 may be an internal storage unit of the server 4, for example, a hard disk or a memory of the server 4. The memory 41 may also be an external storage device of the server 4, such as a plug-in hard disk equipped on the server 4, a smart memory card (Smart Media Card, SMC), and a Secure Digital (SD) card. Flash Card, etc. Further, the memory 41 may also include both an internal storage unit of the server 4 and an external storage device. The memory 41 is used to store the computer-readable instructions and other instructions and data required by the server 4. The memory 41 can also be used to temporarily store data that has been output or will be output.

The functional units in the various embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or software functional unit.

If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer non-volatile readable storage medium. Based on this understanding, the technical solution of the present application essentially or the part that contributes to the existing technology or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium , Including several computer-readable instructions to enable a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disks or optical disks, etc., which can store computer readable instructions. Medium.

A person of ordinary skill in the art can understand that all or part of the processes in the method of the above-mentioned embodiments can be implemented by instructing relevant hardware through computer-readable instructions. The computer-readable instructions can be stored in a non-volatile computer. In a readable storage medium, when the computer-readable instructions are executed, they may include the processes of the above-mentioned method embodiments. Wherein, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and/or volatile memory. Non-volatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not a limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Channel (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

The above-mentioned embodiments are only used to illustrate the technical solutions of the present application, not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, a person of ordinary skill in the art should understand that it can still implement the foregoing The technical solutions recorded in the examples are modified, or some of the technical features are equivalently replaced; these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims (20)

1. A method for prompting case auxiliary information, which is characterized in that it includes:

   Obtain the text of the judgment result of the case;

   Extract the sentencing information and applicable laws and regulations of the case from the text of the judgment result;

   Extract information on the merits of the case from legal documents related to the case;

   The case information is input into a pre-built first neural network model, and the reasonableness of the sentencing information is determined by the output result of the first neural network model, and the first neural network model is determined by the information corresponding to the sentencing information. The case data of a plurality of first sample cases are trained as a training set, and the first neural network model obtains the first similarity by matching the case information with the case data of the first sample case, and according to The magnitude of the first degree of similarity determines the reasonableness of the sentencing information;

   The case information is input to a pre-built second neural network model, and the output result of the second neural network model is used to determine whether the applicable law is correct, and the second neural network model is determined by corresponding to the applicable law The case data of the multiple second sample cases is obtained as a training set training, and the second neural network model obtains the second similarity by matching the case information and the case data of the second sample case. The magnitude of the second degree of similarity determines whether the applicable law is correct;

   If the reasonableness of the sentencing information is lower than a preset threshold or the applicable legal provisions are incorrect, output auxiliary information indicating that the case is incorrectly judged.
2. The method for prompting auxiliary information of a case according to claim 1, wherein said extracting sentencing information and applicable legal provisions of said case from said judgment result text comprises:

   Detect sentencing keywords and legal keywords recorded in the pre-built keyword database from the text of the judgment result;

   Separately count the word frequency of each sentenced keyword detected and the word frequency of each legal article keyword;

   Determine the keyword with the highest word frequency among the detected sentencing keywords as the extracted sentencing information;

   The keyword with the highest word frequency among the detected legal clause keywords is determined as the extracted applicable legal clause.
3. The method for prompting auxiliary information of a case according to claim 2, wherein the keyword database also records keywords of case facts, and said extracting case information of the case from legal documents related to the case includes :

   Use the pre-built word segmentation model to segment the content of the legal document to obtain the target phrase set;

   The case key words detected from the target phrase set are determined as the extracted case information of the case.
4. The method for prompting case auxiliary information according to claim 3, characterized in that said using a pre-built word segmentation model to segment the content of the legal document comprises:

   Detect preset target keywords from the content of the legal document;

   Determine the type of the legal document according to the detected target keywords;

   The word segmentation model corresponding to the type of the legal document is selected to segment the content of the legal document.
5. The method for prompting case auxiliary information according to claim 4, wherein the determining the type of the legal document according to the detected target keyword comprises:

   If the number of detected target keywords is one, determine the type of the legal document according to the detected target keywords;

   If the number of detected target keywords is more than two, the detected target keywords are divided into more than one keyword combination, and the type of the legal document is determined according to the keyword combination, each The keyword combination includes two or more of the target keywords.
6. A prompting device for case auxiliary information, which is characterized in that it comprises:

   Judgment result obtaining module, used to obtain the text of the judgment result of the case;

   The judgment information extraction module is used to extract the sentencing information and applicable laws and regulations of the case from the text of the judgment result;

   The case information extraction module is used to extract the case information of the case from the legal documents related to the case;

   The sentencing rationality determination module is configured to input the case information into a pre-built first neural network model, and determine the rationality of the sentencing information through the output result of the first neural network model, the first neural network model The case data of a plurality of first sample cases corresponding to the sentencing information is obtained as a training set for training, and the first neural network model matches the case information with the case data of the first sample case Obtain the first degree of similarity, and determine the reasonableness of the sentencing information according to the magnitude of the first degree of similarity;

   The applicable law article determination module is used to input the case information into a pre-built second neural network model, and determine whether the applicable law article is correct according to the output result of the second neural network model, and the second neural network model The case data of a plurality of second sample cases corresponding to the applicable law is obtained as a training set for training, and the second neural network model is obtained by matching the case information with the case data of the second sample case A second degree of similarity, and determine whether the applicable law is correct according to the magnitude of the second degree of similarity;

   The auxiliary information output module is configured to output auxiliary information for indicating that the judgment of the case is incorrect if the reasonableness of the sentencing information is lower than a preset threshold or the applicable law is incorrect.
7. The device for prompting case auxiliary information according to claim 6, wherein the judgment information extraction module comprises:

   The keyword detection unit is used to detect from the judgment result text the sentencing keywords and legal keywords recorded in the pre-built keyword database;

   The word frequency statistics unit is used to separately count the word frequency of each sentenced keyword and the word frequency of each legal article keyword detected;

   The sentencing information determining unit is configured to determine the keyword with the highest word frequency among the detected sentencing keywords as the extracted sentencing information;

   The applicable legal clause determining unit is used to determine the keyword with the highest word frequency among the detected legal clause keywords as the extracted applicable legal clause.
8. The device for prompting case auxiliary information according to claim 7, wherein the keyword database also records case keywords, and the case information extraction module includes:

   The word segmentation unit is used to segment the content of the legal document using a pre-built word segmentation model to obtain a target phrase set;

   The case information determining unit is configured to determine the case key words detected from the target phrase set as the extracted case information of the case.
9. The prompting device for case auxiliary information according to claim 8, wherein the word segmentation unit comprises:

   The keyword detection subunit is used to detect preset target keywords from the content of the legal document;

   The legal document type determination subunit is used to determine the type of the legal document according to the detected target keywords;

   The word segmentation model selection subunit is used to select a word segmentation model corresponding to the type of the legal document to segment the content of the legal document.
10. The device for prompting case auxiliary information according to claim 9, wherein the legal document type determining subunit comprises:

    The first type determining grandchild unit is configured to determine the type of the legal document according to the detected target keyword if the number of detected target keywords is one;

    The second type determines the grandchildren unit, which is used to divide the detected target keyword into more than one keyword combination if the number of detected target keywords is more than two, and determine according to the keyword combination For the type of the legal document, each of the keyword combinations includes more than two of the target keywords.
11. A computer non-volatile readable storage medium, the computer non-volatile readable storage medium storing computer readable instructions, characterized in that the computer readable instructions are executed by a processor to implement the following steps:

    Obtain the text of the judgment result of the case;

    Extract the sentencing information and applicable laws and regulations of the case from the text of the judgment result;

    Extract information on the merits of the case from legal documents related to the case;

    The case information is input into a pre-built first neural network model, and the reasonableness of the sentencing information is determined by the output result of the first neural network model, and the first neural network model is determined by the information corresponding to the sentencing information. The case data of a plurality of first sample cases are trained as a training set, and the first neural network model obtains the first similarity by matching the case information with the case data of the first sample case, and according to The magnitude of the first degree of similarity determines the reasonableness of the sentencing information;

    The case information is input to a pre-built second neural network model, and the output result of the second neural network model is used to determine whether the applicable law is correct, and the second neural network model is determined by corresponding to the applicable law The case data of the multiple second sample cases is obtained as a training set training, and the second neural network model obtains the second similarity by matching the case information and the case data of the second sample case. The magnitude of the second degree of similarity determines whether the applicable law is correct;

    If the reasonableness of the sentencing information is lower than a preset threshold or the applicable legal provisions are incorrect, output auxiliary information indicating that the case is incorrectly judged.
12. The computer non-volatile readable storage medium according to claim 11, wherein said extracting sentencing information and applicable laws and regulations of said case from said judgment result text comprises:

    Detect sentencing keywords and legal keywords recorded in the pre-built keyword database from the text of the judgment result;

    Separately count the word frequency of each sentenced keyword detected and the word frequency of each legal article keyword;

    Determine the keyword with the highest word frequency among the detected sentencing keywords as the extracted sentencing information;

    The keyword with the highest word frequency among the detected legal clause keywords is determined as the extracted applicable legal clause.
13. The computer non-volatile readable storage medium according to claim 12, wherein the keyword database also records keywords on the case, and extracts information about the case from legal documents related to the case. The case information includes:

    Use the pre-built word segmentation model to segment the content of the legal document to obtain the target phrase set;

    The case key words detected from the target phrase set are determined as the extracted case information of the case.
14. The computer non-volatile readable storage medium according to claim 13, wherein said segmenting the content of said legal document by using a pre-built word segmentation model comprises:

    Detect preset target keywords from the content of the legal document;

    Determine the type of the legal document according to the detected target keywords;

    The word segmentation model corresponding to the type of the legal document is selected to segment the content of the legal document.
15. The computer non-volatile readable storage medium according to claim 14, wherein the determining the type of the legal document according to the detected target keyword comprises:

    If the number of detected target keywords is one, determine the type of the legal document according to the detected target keywords;

    If the number of detected target keywords is more than two, the detected target keywords are divided into more than one keyword combination, and the type of the legal document is determined according to the keyword combination, each The keyword combination includes two or more of the target keywords.
16. A server includes a memory, a processor, and computer-readable instructions stored in the memory and capable of running on the processor, wherein the processor implements the following steps when the processor executes the computer-readable instructions :

    Obtain the text of the judgment result of the case;

    Extract the sentencing information and applicable laws and regulations of the case from the text of the judgment result;

    Extract information on the merits of the case from legal documents related to the case;

    The case information is input into a pre-built first neural network model, and the reasonableness of the sentencing information is determined by the output result of the first neural network model, and the first neural network model is determined by the information corresponding to the sentencing information. The case data of a plurality of first sample cases are trained as a training set, and the first neural network model obtains the first similarity by matching the case information with the case data of the first sample case, and according to The magnitude of the first degree of similarity determines the reasonableness of the sentencing information;

    The case information is input to a pre-built second neural network model, and the output result of the second neural network model is used to determine whether the applicable law is correct, and the second neural network model is determined by corresponding to the applicable law The case data of the multiple second sample cases is obtained as a training set training, and the second neural network model obtains the second similarity by matching the case information and the case data of the second sample case. The magnitude of the second degree of similarity determines whether the applicable law is correct;

    If the reasonableness of the sentencing information is lower than a preset threshold or the applicable legal provisions are incorrect, output auxiliary information indicating that the case is incorrectly judged.
17. The server according to claim 16, wherein said extracting sentencing information and applicable laws and regulations of said case from said judgment result text comprises:

    Detect sentencing keywords and legal keywords recorded in the pre-built keyword database from the text of the judgment result;

    Separately count the word frequency of each sentenced keyword detected and the word frequency of each legal article keyword;

    Determine the keyword with the highest word frequency among the detected sentencing keywords as the extracted sentencing information;

    The keyword with the highest word frequency among the detected legal clause keywords is determined as the extracted applicable legal clause.
18. The server according to claim 17, wherein the keyword database also records key words of case facts, and said extracting case information of the case from legal documents related to the case comprises:

    Use the pre-built word segmentation model to segment the content of the legal document to obtain the target phrase set;

    The case key words detected from the target phrase set are determined as the extracted case information of the case.
19. The server according to claim 18, wherein said using a pre-built word segmentation model to segment the content of the legal document comprises:

    Detect preset target keywords from the content of the legal document;

    Determine the type of the legal document according to the detected target keywords;

    The word segmentation model corresponding to the type of the legal document is selected to segment the content of the legal document.
20. The server according to claim 19, wherein the determining the type of the legal document according to the detected target keyword comprises:

    If the number of detected target keywords is one, determine the type of the legal document according to the detected target keywords;

    If the number of detected target keywords is more than two, the detected target keywords are divided into more than one keyword combination, and the type of the legal document is determined according to the keyword combination, each The keyword combination includes two or more of the target keywords.

Images