CN118096452A

CN118096452A - Case auxiliary judgment method, device, terminal equipment and medium

Info

Publication number: CN118096452A
Application number: CN202410494749.XA
Authority: CN
Inventors: 陈晓红; 詹敏; 韩啸; 曹世东; 董玉成
Original assignee: Xiangjiang Laboratory
Current assignee: Xiangjiang Laboratory
Priority date: 2024-04-24
Filing date: 2024-04-24
Publication date: 2024-05-28
Anticipated expiration: 2044-04-24
Also published as: CN118096452B

Abstract

The application is suitable for the technical field of legal case analysis, and provides a case auxiliary judging method, a device, terminal equipment and a medium, wherein the method is used for refining case knowledge labels; according to the text structural characteristics, a document structure extraction rule is constructed, and judge documents are respectively divided to obtain a structured data set; constructing an entity identification rule according to the case knowledge label, and identifying the structured data set to obtain a case entity set; constructing a case knowledge graph according to the case relation set and the case entity set; calculating the influence degree of each case entity in the case knowledge graph, and determining typical case entities from the case entity set; calculating the similarity between the to-be-judged case and the typical case entity, and determining a matched case; and predicting principal penalty of the to-be-judged case according to the judge document of the typical case entity corresponding to the highest matching score, so as to realize case auxiliary judgment. The application can improve the accuracy and the interpretability of the auxiliary judgment of the case.

Description

Case auxiliary judgment method, device, terminal equipment and medium

Technical Field

The application belongs to the technical field of legal case analysis, and particularly relates to a case auxiliary judgment method, device, terminal equipment and medium.

Background

The case auxiliary judging method and system based on the artificial intelligence technology can process and extract key information of massive cases, manage and analyze the key information, and realize the functions of case retrieval, classical case pushing, case matching and the like according to the actual scene requirements so as to improve the operation efficiency and effect of a court. The existing case auxiliary judging method generally carries out large-scale training on the case through a deep learning model to extract information, stores the information in a relational database and carries out traversal analysis through structured query language (SQL, structured Query Language). However, in the aspect of information extraction, as a semi-structured text, the judge document line text usually has certain characteristics and rules, and the key characteristics of case knowledge can be lost only by directly processing the text through a deep learning model, so that the information extraction accuracy and judicial interpretation are not high; in addition, the functions of searching, case matching, typical case pushing and the like in the conventional case auxiliary judging system are generally to store the cases in a relational database management system (RDBMS, relational Database MANAGEMENT SYSTEM), and the functions of searching and the like are realized mainly by traversing each part of case information based on SQL sentences. However, databases for storing cases often have a large scale, and auxiliary judgment systems constructed based on relational databases need to be repeatedly indexed and read in the application process, resulting in low efficiency and accuracy.

Disclosure of Invention

The application provides a case auxiliary judging method, a device, terminal equipment and a medium, which can solve the problems of low accuracy and poor interpretation of the traditional case auxiliary judging method.

In a first aspect, the present application provides a case auxiliary judging method, including:

Extracting a case knowledge label from a case information base constructed in advance; the case information base comprises a plurality of judge documents, the case knowledge labels comprise auxiliary judging knowledge labels and case retrieval management knowledge labels, the auxiliary judging knowledge labels are used for indicating knowledge labels which are applicable to auxiliary judging methods and have legal basis, the case retrieval management knowledge labels are used for indicating multidimensional knowledge labels which are applicable to fine retrieval, and the case knowledge labels are in one-to-one correspondence with the judge documents;

According to the text structure characteristics of the judge documents, a document structure extraction rule for dividing the text structures of the judge documents is constructed, and the judge documents are respectively divided according to the document structure extraction rule to obtain a structured data set;

Constructing an entity identification rule according to the case knowledge label, and identifying the structured data set according to the entity identification rule to obtain a case entity set; the recognition rules comprise text structure entity recognition rules for recognizing text structures, zero-width assertion entity recognition rules for recognizing case knowledge label break words and judgment statement entity recognition rules for recognizing judgment words in case knowledge labels, and the case entity set comprises a plurality of case entities which are in one-to-one correspondence with the judge documents;

Constructing a case knowledge graph according to a case relation set and a case entity set which are designed in advance;

Respectively calculating influence degree of each case entity in the case knowledge graph, and determining at least one typical case entity from the case entity set according to the influence degree;

Respectively calculating the similarity between the to-be-judged case and each typical case entity, and determining at least one matching case according to the similarity; matching cases represent typical case entities similar to the to-be-examined decide a case cases;

and respectively calculating the matching score of each matching case, and predicting principal penalty of the case to be judged according to the judge document of the typical case entity corresponding to the highest matching score, so as to realize the auxiliary judgment of the case.

Optionally, the auxiliary judgment knowledge tag comprises crime facts, secondary factors, judgment basis and judgment results;

the case retrieval management knowledge labels comprise case types, case numbers, specific case details, attribution provinces, court of law, main judges, judgment dates, names of the interviews, professions of the interviews, degrees of humanization of the interviews, ethnicities of the interviews and sexes of the interviews.

Alternatively, the expression of the document structure extraction rule is as follows:

header: matching ' finishing of the examination and the ' controlling ' in the text as a post-assertion, and acquiring the content before the post-assertion as a header part;

facts: the method comprises the steps of matching 'finishing of the examination and the command' as front words of the examination from a text, matching 'thought of the home', 'thought of the court' and 'thought of the conclusion court' as rear words of the conclusion in the text according to the sequence, and obtaining contents between the front and rear words of the examination as a fact part;

The reason is that: matching the text with the text as a front word and the text as a rear word, and acquiring the content between the front word and the rear word as a reason part;

main text: the method comprises the steps of matching ' decision as follows ' from a text as a front word, matching ' if the decision is not taken and ' trial ' as a rear word from the text according to the sequence, and obtaining the content between the front word and the rear word as a main text part;

Tail part: and matching the words such as not taking the present judgment and the trial judgment as the front assertion from the text, and acquiring the content after the front assertion as the tail part.

Optionally, identifying the structured dataset according to an entity identification rule to obtain a case entity set, including:

For text structure entity recognition rules, a list is obtained by converting the document content of each referee document into list form and taking the paragraphs as intervals ，/>For the i-th piece of the current document,/>Is an entity of the court of law,/>For case type entity,/>Is a case number entity;

Aiming at zero width assertion entity identification rules, a regular matching formula corresponding to the case knowledge labels is constructed by acquiring the assertion words in each case knowledge label; specific contents before, after or between two assertion words are extracted through the regular matching formula, such as a concrete case by taking a 'crime' as a front assertion word, and extracting a concrete case by taking a 'crime' as a rear assertion word, namely, a reported crime name.

And aiming at the judgment statement entity recognition rule, mapping the knowledge tag judged as True into the predefined content by predefining the entity content and constructing a knowledge tag judgment word.

Optionally, the calculation formula of the influence degree is as follows:

；

Wherein, Represents the/>Case entity/>At time/>Influence degree of/(I)Represents the damping coefficient of the damping device,Representation of case entity/>Incoming neighbor of/>Representation/>Outgoing neighbor/>，/>Representing the evaluation score, i.e. the number of edges emanating from the current entity.

Optionally, calculating the similarity between the to-be-evaluated case and each typical case entity, and determining at least one matching case according to the similarity, including:

vectorizing the to-be-judged case and each typical case entity respectively;

By calculation formula

；

To get decide a case pieces to be examinedWith typical case entity/>Similarity between/>; Wherein/>Entity object representing computed similarity,/>Entity vector representing calculated similarity,/>Representing the vector dimension of the current entity,/>First/>, representing the current vector matrixA dimension vector;

and determining the typical case entity with the similarity larger than or equal to a preset similarity threshold as a matching case of the to-be-examined decide a case cases.

Optionally, the matching score is calculated as follows:

；

Wherein, Vector/>, representing the case under trialMatching case/>Vector/>Match score between,/>And representing the vector corresponding to the judgment result.

In a second aspect, the present application provides a case auxiliary judging device, including:

The tag module is used for extracting a case knowledge tag from a case information base which is built in advance; the case information base comprises a plurality of judge documents, the case knowledge labels comprise auxiliary judging knowledge labels and case retrieval management knowledge labels, the auxiliary judging knowledge labels are used for indicating knowledge labels which are applicable to auxiliary judging methods and have legal basis, the case retrieval management knowledge labels are used for indicating multidimensional knowledge labels which are applicable to fine retrieval, and the case knowledge labels are in one-to-one correspondence with the judge documents;

The extraction rule module is used for constructing a document structure extraction rule for dividing the document structure of the referee document according to the text structure characteristics of the referee documents, and respectively dividing the referee documents according to the document structure extraction rule to obtain a structured data set;

The recognition rule module is used for constructing an entity recognition rule according to the case knowledge label, and recognizing the structured data set according to the entity recognition rule to obtain a case entity set; the recognition rules comprise text structure entity recognition rules for recognizing text structures, zero-width assertion entity recognition rules for recognizing case knowledge label break words and judgment statement entity recognition rules for recognizing judgment words in case knowledge labels, and the case entity set comprises a plurality of case entities which are in one-to-one correspondence with the judge documents;

the knowledge graph module is used for constructing a case knowledge graph according to a case relation set and a case entity set which are designed in advance;

The typical case determining module is used for respectively calculating the influence degree of each case entity in the case knowledge graph and determining at least one typical case entity from the case entity set according to the influence degree;

The matching case determining module is used for respectively calculating the similarity between the case to be judged and each typical case entity and determining at least one matching case according to the similarity; matching cases represent typical case entities similar to the to-be-examined decide a case cases;

The auxiliary judging module is used for respectively calculating the matching score of each matching case, and predicting principal penalty of the case to be judged according to the judging document of the typical case entity corresponding to the highest matching score so as to realize the auxiliary judging of the case.

In a third aspect, the present application provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor implements the case auxiliary judging method described above when executing the computer program.

In a fourth aspect, the present application provides a computer readable storage medium, where a computer program is stored, where the computer program when executed by a processor implements the case assisted trial method described above.

The scheme of the application has the following beneficial effects:

According to the case auxiliary judging method provided by the application, the document structure extraction rule is constructed according to the text structure characteristics of a plurality of judge documents, the semi-structured characteristics of the judge documents are utilized, the text structures of the judge documents can be accurately divided, the interpretability of structured data is enhanced, and the case auxiliary judging accuracy is improved; by calculating the similarity and the matching score between the to-be-checked decide a case cases and other cases respectively, the case most similar to the to-be-checked decide a case cases can be further screened, the interference caused by other irrelevant cases is reduced, the workload of the process is reduced, and meanwhile, the accuracy of the auxiliary judgment of the cases is improved.

Other advantageous effects of the present application will be described in detail in the detailed description section which follows.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the embodiments or the description of the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a flowchart of a case auxiliary judging method according to an embodiment of the present application;

Fig. 2 is a schematic structural diagram of a case auxiliary judging device according to an embodiment of the present application;

fig. 3 is a schematic structural diagram of a terminal device according to an embodiment of the present application.

Detailed Description

In the following description, for purposes of explanation and not limitation, specific details are set forth such as the particular system architecture, techniques, etc., in order to provide a thorough understanding of the embodiments of the present application. It will be apparent, however, to one skilled in the art that the present application may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary detail.

It should be understood that the terms "comprises" and/or "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.

It should also be understood that the term "and/or" as used in the present specification and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes such combinations.

As used in the present description and the appended claims, the term "if" may be interpreted as "when..once" or "in response to a determination" or "in response to detection" depending on the context. Similarly, the phrase "if a determination" or "if a [ described condition or event ] is detected" may be interpreted in the context of meaning "upon determination" or "in response to determination" or "upon detection of a [ described condition or event ]" or "in response to detection of a [ described condition or event ]".

Furthermore, the terms "first," "second," "third," and the like in the description of the present specification and in the appended claims, are used for distinguishing between descriptions and not necessarily for indicating or implying a relative importance.

Reference in the specification to "one embodiment" or "some embodiments" or the like means that a particular feature, structure, or characteristic described in connection with the embodiment is included in one or more embodiments of the application. Thus, appearances of the phrases "in one embodiment," "in some embodiments," "in other embodiments," and the like in the specification are not necessarily all referring to the same embodiment, but mean "one or more but not all embodiments" unless expressly specified otherwise. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless expressly specified otherwise.

Aiming at the problems of low accuracy and poor interpretation of the traditional case auxiliary judging method, the application provides a case auxiliary judging method, a device, terminal equipment and a medium, wherein the method constructs a document structure extraction rule according to the text structure characteristics of a plurality of judge documents, and utilizes the semi-structured characteristic of the judge documents to accurately divide the text structure of the judge documents, thereby enhancing the interpretation of structured data and further improving the accuracy of case auxiliary judgment; by calculating the similarity and the matching score between the to-be-checked decide a case cases and other cases respectively, the case most similar to the to-be-checked decide a case cases can be further screened, the interference caused by other irrelevant cases is reduced, the workload of the process is reduced, and meanwhile, the accuracy of the auxiliary judgment of the cases is improved.

The case auxiliary judging method provided by the application is exemplified below.

As shown in fig. 1, the case auxiliary judging method provided by the application comprises the following steps:

and step 11, extracting the case knowledge tag from a case information base constructed in advance.

The case information base comprises a plurality of judge documents, the case knowledge labels comprise auxiliary judging knowledge labels and case retrieval management knowledge labels, the auxiliary judging knowledge labels are used for indicating knowledge labels which are applicable to auxiliary judging methods and have legal basis, the case retrieval management knowledge labels are used for indicating multidimensional knowledge labels which are applicable to fine retrieval, and the case knowledge labels correspond to the judge documents one by one.

Specifically, in the embodiment of the application, the auxiliary judgment knowledge label comprises crime facts, secondary light factors, secondary heavy factors, judgment basis and judgment results.

And step 12, constructing a document structure extraction rule for dividing the document structure of the referee document according to the text structure characteristics of the referee documents, and dividing the referee documents according to the document structure extraction rule to obtain a structured data set.

Specifically, the expression and thought of the document structure extraction rule are as follows:

Header: according to the sequence, matching ' finishing of the examination and the ' command ' from the text as a post-assertion, and acquiring the content before the post-assertion as a header part;

facts: the method comprises the steps of matching 'finishing of the examination and the command' as front words from a text according to the sequence, matching 'thought of the home', 'thought of the court' and 'thought of the concierge of the court' as rear words from the text according to the sequence, and obtaining contents between the front and rear words as a fact part;

The reason is that: according to the sequence, matching the text with the terms of "home thinking", "home thinking" and "local conclusive thinking" as front word, and matching the text with the terms of "decision as follows" as rear word, and obtaining the content between the front word and the rear word as reason part;

Main text: according to the sequence, matching the words such as not taking the present judgment and the trial judgment as the rear assertion words in the text, and acquiring the content between the front assertion word and the rear assertion word as a main text part;

Tail part: and matching the words such as not taking the present decision and the trial decision as the front assertion words from the text according to the sequence, and acquiring the content after the front assertion words as the tail part.

The built extraction rules are all built by zero-width assertion (zero-width assertions) method in regular expression, i.e. the object before, after or in between the two specific contents are matched. And (3) processing the XX document of the criminal first word in XX year through the constructed extraction rule, wherein the head extraction rule is matched with the 'approval end' in the document. "that is, the predicate and all contents before the predicate are identified as the header part of the current document; the fact extraction rules are matched to the "end of the trial" in the text. "and" court "is considered" that is, the fact that the content between the two assertions is identified as part of the current document; the reason extraction rule is matched with the "court's belief" and the "decision is as follows", namely the content between the two assertion words is identified as the reason part of the current document; the main text extraction rule is matched with the 'decision as follows' and the 'decision as not taking the present decision', namely, the content between the two assertion words is identified as the main text part of the current document; the tail extraction rule is matched with the text 'if the text decision is not taken', namely the content between the two assertion words is identified as the main text part of the current document;

The text analysis can be used for knowing that the referee document has certain difference in the expression of the line text, such as the fact extraction rule and the reason extraction rule that the assertion word is considered by the "court" in the fact extraction rule, and the individual document takes the "court considered" and the "court considered by the" court considered "as the initial sentence in the part, so that the structured extraction rule is matched through a multi-layer matching model in the implementation process, the first assertion word is tried in the matching process, the first assertion word is stopped if the content is matched, and the next assertion word is tried until all assertion words are tried.

And 13, constructing an entity identification rule according to the case knowledge label, and identifying the structured data set according to the entity identification rule to obtain a case entity set.

The recognition rules comprise a text structure entity recognition rule for recognizing a text structure, a zero-width assertion entity recognition rule for recognizing a word in a case knowledge label, and a judgment statement entity recognition rule for recognizing a judgment word in the case knowledge label, wherein the case entity set comprises a plurality of case entities which are in one-to-one correspondence with the judge document.

And 14, constructing a case knowledge graph according to the case relation set and the case entity set which are designed in advance.

Illustratively, in an embodiment of the present application, the set of case relationships is shown in the following table:

TABLE 1

The process of constructing a case knowledge graph in the present application is exemplarily described as follows.

Firstly, connecting the entities in the case entity set according to the case relation set to obtain a case knowledge graph. The connection process is mainly realized by the following grammar:

LOAD CSV WITH HEADERS FROM "file:///X.csv" AS line

match (from:A{A:line.A}),(to:B{B:line.B})

merge (from)-[r:C{A:line.A,B:line.B}]->(to)

wherein X.csv is CSV file for storing the extracted entity, A is subject entity, B is object entity, and C is specific relation.

The case entity triples are then imported pyneo into a library (a client library and toolkit for use of Neo4j in Python applications) and the modules for retrieval and visualization are configured.

Finally, configuring a search grammar, converting the grammar of Cypher (a query language) into natural language search, and realizing case search and visualization through Chinese keywords.

And 15, respectively calculating the influence degree of each case entity in the case knowledge graph, and determining at least one typical case entity from the case entity set according to the influence degree.

Specifically, in step 15.1, the calculation formula is adopted

；

Obtain the firstCase entity/>At time/>Influence of/>。

Wherein,Represents the damping coefficient,/>Representation of case entity/>Incoming neighbor of/>Representation ofOutgoing neighbor/>，/>Representing the evaluation score, i.e. the number of edges emanating from the current entity.

And 15.2, taking the case entity with the influence degree exceeding the preset influence degree threshold as a typical case entity.

The typical cases represent cases with larger influence in the current case database, so that users can conveniently and deeply understand and understand relevant legal knowledge and legal applicable principles through the typical cases.

And step 16, calculating the similarity between the to-be-judged case and each typical case entity respectively, and determining at least one matching case according to the similarity.

The matching cases described above represent typical case entities similar to the pending decide a case cases.

And step 17, calculating the matching score of each matching case respectively, and predicting principal penalty of the case to be judged according to the judge document of the typical case entity corresponding to the highest matching score, so as to realize the auxiliary judgment of the case.

The calculation formula of the matching score is as follows:

；

Wherein, Representing the pending decide a case pieces/>Vector/>Matching case/>Vector/>Match score between,/>And representing the vector corresponding to the judgment result.

After the matching score is obtained, the judge can refer to the principal penalty and the sentency of the to-be-judged case according to the judgment result of the judge document of the typical case entity corresponding to the highest matching score, and the obtained judgment result accords with the rule and regulations, and has high interpretation degree and accurate sentency.

The following is an exemplary description of the process of step 13 (constructing an entity recognition rule according to the case knowledge tag, and recognizing the structured dataset according to the entity recognition rule, to obtain the case entity set).

For the text structure entity recognition rule, obtaining a list by converting the document content of each referee document into list form and taking the paragraphs as intervals，/>For the i-th piece of content of the current document,Is an entity of the court of law,/>For case type entity,/>Is a case number entity;

aiming at the zero-width assertion entity identification rule, a regular matching formula corresponding to each case knowledge label is constructed by acquiring assertion words in each case knowledge label; specific contents before, after or between two assertion words are extracted through the regular matching formula, such as a concrete case by taking a 'crime' as a front assertion word, and extracting a concrete case by taking a 'crime' as a rear assertion word, namely, a reported crime name.

And aiming at the judgment statement entity recognition rule, mapping the knowledge tag judged as True into the predefined content by predefining the entity content and constructing a knowledge tag judgment word. The light factor content is predefined as the real supply description, the self-first and the none, and the judgment word is the real supply description and the self-first; from the content of heavy factors predefined as recidivism and none, judging that the word is recidivism; the career of the reported person is predefined as a career, no industry, a driver and an individual, and the judgment words are the career, the agriculture, the farmer, the career, the worker, the no industry and the no career; the reported humanization is predefined as male and female, and the judgment words are male and female; the degree of humanization of the words is predefined as primary culture, junior middle culture, college culture, and judgment words are primary, junior middle, proprietary, and family.

Taking a document XX of the first criminal word on (XX year) as an example, the effect of implementing the text structure entity identification rule is as follows:

Reading the content of the XX document of the criminal initial word (XX year) by the python program, converting the content into a list form, and obtaining the list Text with the specific content: text= [ XX county of XX people court, criminal judgement, first criminal word XX number in XX year ], public complaint organ … … ]; wherein the method comprises the steps of (XX county of people court in XX province) is identified as an entity of the court of approval,/>(Criminal decision) is a case type entity,/>((XX year) the criminal initial word XX number) is a case number entity;

Taking a literary style written in the first word of (XX year) as an example, the effect of implementing the zero width assertion entity identification rule is as follows:

the name of the addressee: constructed recognition rules Identifying the predicate words "interviewee" and ",", i.e., "XXX" between the predicate words as interviewee;

the people group: constructed recognition rules Identifying the assertion words in the text as ' and ' families ', namely identifying ' X ' among the assertion words as the alleged nationality;

crime facts: constructed recognition rules The cubic meter 'identifies the assertion word' cubic meter 'in the text, namely, the numerical value' XX illegal degree value 'of which the prepositions before the assertion word are not the same (excluding the contents about legal description in the text, such as' XX illegal behavior ',' XX illegal degree ', starting from twenty to fifty cubic meters') is identified as crime facts;

the judgment basis is as follows: constructed recognition rules Recognizing the assertion word "according to" in the text, namely the first one of the third hundred forty-five of the "XX criminal law" after the assertion word is broken, and recognizing the sixty-seven third one "as a judgment basis;

the specific scheme is as follows: constructed recognition rules Identifying the assertion words 'offence' and 'crime' in the text, namely identifying the 'XX crime' between the assertion words as a basis of judgment;

Judgment result: in the constructed recognition rules The predicate words "period" and "month" in the text are not recognized, i.e., try the next rule/>Successfully identifying the assertion words as "period" and "year" and identifying the three years between assertion words as period imprisonment in the decision result; constructed recognition rulesIdentifying the assertion words ' penalty ' and ' element ' in the text, namely identifying 15000 elements ' between the assertion words as penalty in the judgment result;

The ascription province: taking an aesthetic court entity in a text structure entity recognition rule result as a recognition object, and recognizing the rule Recognizing the assertion word "province" in the court of the examination, namely recognizing the "XX province" before the assertion word as the attribution province;

Main judges: constructed recognition rules Recognizing the predicate "/>" in text"Identifying the first judge" XX "in the people list immediately before the word is broken as the main judge;

Decision date: constructed recognition rules Recognizing the predicate "/>" in text"And" day ", i.e." two good between the disguised words, four years, september, twenty-eight days "is identified as the decision date;

Taking a document of XX number of the first criminal word on (XX year) as an example, the effect of implementing the identification rule of the judgment statement entity is as follows:

the people to be notified are as follows: identifying the judgment word 'man' in the text, and identifying the person to be advertised as 'man' according to the predefined;

Degree of humanization to be reported: identifying the judgment word 'primary school' in the text, and identifying the degree of the humanization of the notice as 'primary school culture' according to the predefining;

occupation of the interviewee: recognizing a judgment word 'farmer' in the text, and recognizing the occupation of the reported person as 'farmer' according to the predefined;

from the light factors: the judgment word "real-life" in the text is recognized, and the real-life "is recognized as" real-life "according to the predefined.

The following exemplary description is given to the process of step 16 (respectively calculating the similarity between the case to be evaluated and each typical case entity, and determining at least one matching case according to the similarity), which specifically includes steps 16.1 to 16.3:

and 16.1, vectorizing the to-be-judged case and each typical case entity respectively.

Step 16.2, by calculation formula

；

And 16.3, determining a typical case entity with similarity larger than or equal to a preset similarity threshold as a matching case of the to-be-examined decide a case cases.

The case auxiliary judging device provided by the application is exemplified below.

As shown in fig. 2, the case auxiliary judging device 200 includes:

A tag module 201, configured to refine a case knowledge tag from a case information base constructed in advance; the case information base comprises a plurality of judge documents, the case knowledge labels comprise auxiliary judging knowledge labels and case retrieval management knowledge labels, the auxiliary judging knowledge labels are used for indicating knowledge labels which are applicable to auxiliary judging methods and have legal basis, the case retrieval management knowledge labels are used for indicating multidimensional knowledge labels which are applicable to fine retrieval, and the case knowledge labels are in one-to-one correspondence with the judge documents;

The extraction rule module 202 is configured to construct a document structure extraction rule for dividing the document structure of the referee document according to the text structure characteristics of the referee documents, and divide the referee documents according to the document structure extraction rule to obtain a structured data set;

The recognition rule module 203 is configured to construct an entity recognition rule according to the case knowledge tag, and recognize the structured dataset according to the entity recognition rule to obtain a case entity set; the recognition rules comprise text structure entity recognition rules for recognizing text structures, zero-width assertion entity recognition rules for recognizing case knowledge label break words and judgment statement entity recognition rules for recognizing judgment words in case knowledge labels, and the case entity set comprises a plurality of case entities which are in one-to-one correspondence with the judge documents;

the knowledge graph module 204 is configured to construct a case knowledge graph according to a case relationship set and a case entity set that are designed in advance;

the typical case determining module 205 is configured to calculate an influence degree of each case entity in the case knowledge graph, and determine at least one typical case entity from the case entity set according to the influence degree;

The matching case determining module 206 is configured to calculate the similarity between the case to be evaluated and each typical case entity, and determine at least one matching case according to the similarity; matching cases represent typical case entities similar to the to-be-examined decide a case cases;

The auxiliary judging module 207 is configured to calculate a matching score of each matching case, and predict principal penalty to-be-judged cases according to the judge document of the typical case entity corresponding to the highest matching score, so as to implement case auxiliary judgment.

It should be noted that, because the content of information interaction and execution process between the above devices/units is based on the same concept as the method embodiment of the present application, specific functions and technical effects thereof may be referred to in the method embodiment section, and will not be described herein.

It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of the functional units and modules is illustrated, and in practical application, the above-described functional distribution may be performed by different functional units and modules according to needs, i.e. the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-described functions. The functional units and modules in the embodiment may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit, where the integrated units may be implemented in a form of hardware or a form of a software functional unit. In addition, the specific names of the functional units and modules are only for distinguishing from each other, and are not used for limiting the protection scope of the present application. The specific working process of the units and modules in the above system may refer to the corresponding process in the foregoing method embodiment, which is not described herein again.

As shown in fig. 3, an embodiment of the present application provides a terminal device, and as shown in fig. 3, a terminal device D10 of the embodiment includes: at least one processor D100 (only one processor is shown in fig. 3), a memory D101 and a computer program D102 stored in the memory D101 and executable on the at least one processor D100, the processor D100 implementing the steps in any of the various method embodiments described above when executing the computer program D102.

Specifically, when the processor D100 executes the computer program D102, a case knowledge tag is extracted from a case information base constructed in advance; according to the text structure characteristics of the judge documents, a document structure extraction rule for dividing the text structures of the judge documents is constructed, and the judge documents are respectively divided according to the document structure extraction rule to obtain a structured data set; constructing an entity identification rule according to the case knowledge label, and identifying the structured data set according to the entity identification rule to obtain a case entity set; constructing a case knowledge graph according to a case relation set and a case entity set which are designed in advance; respectively calculating influence degree of each case entity in the case knowledge graph, and determining at least one typical case entity from the case entity set according to the influence degree; respectively calculating the similarity between the to-be-judged case and each typical case entity, and determining at least one matching case according to the similarity; and respectively calculating the matching score of each matching case, and predicting principal penalty of the case to be judged according to the judge document of the typical case entity corresponding to the highest matching score, so as to realize the auxiliary judgment of the case. According to the text structure characteristics of a plurality of referee documents, a document structure extraction rule is constructed, the text structure of the referee documents can be accurately divided by utilizing the semi-structured characteristics of the referee documents, and the interpretability of structured data is enhanced, so that the accuracy of case auxiliary judgment is improved; by calculating the similarity and the matching score between the to-be-checked decide a case cases and other cases respectively, the case most similar to the to-be-checked decide a case cases can be further screened, the interference caused by other irrelevant cases is reduced, the workload of the process is reduced, and meanwhile, the accuracy of the auxiliary judgment of the cases is improved.

The Processor D100 may be a central processing unit (CPU, central Processing Unit), the Processor D100 may also be other general purpose processors, digital signal processors (DSP, digital Signal processors), application SPECIFIC INTEGRATED integrated circuits (ASICs), off-the-shelf Programmable gate arrays (FPGAs) or other Programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

The memory D101 may in some embodiments be an internal storage unit of the terminal device D10, for example a hard disk or a memory of the terminal device D10. The memory D101 may also be an external storage device of the terminal device D10 in other embodiments, for example, a plug-in hard disk, a smart memory card (SMC, smart Media Card), a Secure Digital (SD) card, a flash memory card (FLASH CARD) or the like, which are provided on the terminal device D10. Further, the memory D101 may also include both an internal storage unit and an external storage device of the terminal device D10. The memory D101 is used for storing an operating system, an application program, a boot loader (BootLoader), data, other programs, etc., such as program codes of the computer program. The memory D101 may also be used to temporarily store data that has been output or is to be output.

Embodiments of the present application also provide a computer readable storage medium storing a computer program which, when executed by a processor, implements steps for implementing the various method embodiments described above.

Embodiments of the present application provide a computer program product enabling a terminal device to carry out the steps of the method embodiments described above when the computer program product is run on the terminal device.

The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the present application may implement all or part of the flow of the method of the above embodiments, and may be implemented by a computer program to instruct related hardware, where the computer program may be stored in a computer readable storage medium, and when the computer program is executed by a processor, the computer program may implement the steps of each of the method embodiments described above. Wherein the computer program comprises computer program code which may be in source code form, object code form, executable file or some intermediate form etc. The computer readable medium may include at least: any entity or device capable of carrying computer program code to case aid trial devices/terminal equipment, recording medium, computer Memory, read-Only Memory (ROM), random access Memory (RAM, random Access Memory), electrical carrier signals, telecommunications signals, and software distribution media. Such as a U-disk, removable hard disk, magnetic or optical disk, etc. In some jurisdictions, computer readable media may not be electrical carrier signals and telecommunications signals in accordance with legislation and patent practice.

In the foregoing embodiments, the descriptions of the embodiments are emphasized, and in part, not described or illustrated in any particular embodiment, reference is made to the related descriptions of other embodiments.

Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.

In the embodiments provided in the present application, it should be understood that the disclosed apparatus/network device and method may be implemented in other manners. For example, the apparatus/network device embodiments described above are merely illustrative, e.g., the division of the modules or units is merely a logical functional division, and there may be additional divisions in actual implementation, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection via interfaces, devices or units, which may be in electrical, mechanical or other forms.

The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

The case auxiliary judging method provided by the application has the following advantages:

1. the method has the advantages that the case related legal laws are analyzed, a case multidimensional knowledge tag system is constructed from the application point of view, the practical value of the extracted knowledge is ensured, and a tag system reference is provided for other methods.

2. Based on the legal document structural features and the constructed knowledge label features, three case entity extraction methods are constructed, and the accuracy and the efficiency of named entity identification are effectively improved.

3. And combining the case knowledge graph and the actual demand, constructing a case multidimensional label retrieval and visualization model, a classical case pushing model, a case matching model and an trial result prediction model by utilizing an intelligent reasoning algorithm, and opening the process of constructing the case knowledge graph to the application.

4. The constructed case auxiliary judging method relies on the database of the non-index adjacency graph, and compared with the existing method, repeated reading and repeated indexing of the database can be avoided, and the accuracy and efficiency of the model can be effectively improved.

While the foregoing is directed to the preferred embodiments of the present application, it will be appreciated by those skilled in the art that various modifications and adaptations can be made without departing from the principles of the present application, and such modifications and adaptations are intended to be comprehended within the scope of the present application.

Claims

1. The case auxiliary judging method is characterized by comprising the following steps of:

Extracting a case knowledge label from a case information base constructed in advance; the case information base comprises a plurality of judge documents, the case knowledge labels comprise auxiliary judging knowledge labels and case retrieval management knowledge labels, the auxiliary judging knowledge labels represent knowledge labels which are applicable to an auxiliary judging method and have legal basis, the case retrieval management knowledge labels are applicable to multi-dimensional knowledge labels of refined retrieval, and the case knowledge labels are in one-to-one correspondence with the judge documents;

Constructing an entity identification rule according to the case knowledge tag, and identifying the structured data set according to the entity identification rule to obtain a case entity set; the recognition rules comprise text structure entity recognition rules for recognizing text structures, zero-width assertion entity recognition rules for recognizing case knowledge label interrupt words and judgment statement entity recognition rules for recognizing judgment words in case knowledge labels, and the case entity set comprises a plurality of case entities which are in one-to-one correspondence with the judge documents;

constructing a case knowledge graph according to a case relation set and the case entity set which are designed in advance;

respectively calculating the influence degree of each case entity in the case knowledge graph, and determining at least one typical case entity from the case entity set according to the influence degree;

Respectively calculating the similarity between the to-be-judged case and each typical case entity, and determining at least one matching case according to the similarity; the matching case represents a typical case entity similar to the pending decide a case cases;

And respectively calculating the matching score of each matching case, and predicting principal penalty of the case to be judged according to the judge document of the typical case entity corresponding to the highest matching score, so as to realize auxiliary judgment of the case.

2. The case auxiliary judging method according to claim 1, wherein the auxiliary judging knowledge tag includes crime facts, secondary light factors, secondary heavy factors, judgment basis and judgment result;

the case retrieval management knowledge label comprises a case type, a case number, a specific case, a attribution province, an inspection court, a main judge, a judging date, a name of a person to be told, a occupation of the person to be told, a degree of humanization of the person to be told, a ethnicity of the person to be told and a sex of the person to be told.

3. The case-aided trial method of claim 1, wherein the expression and the thought of the document structure extraction rule are as follows:

4. The case assistance trial method of claim 1, wherein the identifying the structured dataset according to the entity identification rule to obtain a case entity set comprises:

For the text structure entity recognition rule, obtaining a list by converting the document content of each referee document into list form and taking the paragraphs as intervals ，/>For the i-th piece of the current document,/>Is an entity of the court of law,/>For case type entity,/>Is a case number entity;

Aiming at the zero-width assertion entity identification rule, a regular matching formula corresponding to each case knowledge label is constructed by acquiring assertion words in each case knowledge label;

5. The case-aided judgment method of claim 1, wherein the influence degree is calculated according to the following formula:

；

Wherein, Represents the/>Case entity/>At time/>Influence degree of/(I)Represents the damping coefficient,/>Representation of case entity/>Incoming neighbor of/>Representation/>Outgoing neighbor/>，/>Representing the evaluation score, i.e. the number of edges emanating from the current entity.

6. The case auxiliary judging method according to claim 1, wherein the calculating the similarity between the case to be judged and each typical case entity and determining at least one matching case according to the similarity includes:

Vectorizing the to-be-judged case and each typical case entity respectively;

By calculation formula

；

And determining the typical case entity corresponding to the similarity greater than or equal to a preset similarity threshold as the matching case of the case to be judged.

7. The case-aided trial method of claim 1, wherein the matching score is calculated as follows:

；

Wherein, Vector/>, representing the case under trialMatching case/>Vector/>Match score between,/>And representing the vector corresponding to the decision result.

8. The utility model provides a case auxiliary trial device which characterized in that includes:

The tag module is used for extracting a case knowledge tag from a case information base which is built in advance; the case information base comprises a plurality of judge documents, the case knowledge labels comprise auxiliary judging knowledge labels and case retrieval management knowledge labels, the auxiliary judging knowledge labels represent knowledge labels which are applicable to auxiliary judging methods and have legal basis, the case retrieval management knowledge labels represent multidimensional knowledge labels applicable to refined retrieval, and the case knowledge labels are in one-to-one correspondence with the judge documents;

the identification rule module is used for constructing an entity identification rule according to the case knowledge label, and identifying the structured data set according to the entity identification rule to obtain a case entity set; the recognition rules comprise text structure entity recognition rules for recognizing text structures, zero-width assertion entity recognition rules for recognizing case knowledge label interrupt words and judgment statement entity recognition rules for recognizing judgment words in case knowledge labels, and the case entity set comprises a plurality of case entities which are in one-to-one correspondence with the judge documents;

The knowledge graph module is used for constructing a case knowledge graph according to a case relation set and the case entity set which are designed in advance;

The matching case determining module is used for respectively calculating the similarity between the case to be judged and each typical case entity and determining at least one matching case according to the similarity; the matching case represents a typical case entity similar to the pending decide a case cases;

the auxiliary judging module is used for respectively calculating the matching score of each matching case, and predicting principal penalty of the case to be judged according to the judge document of the typical case entity corresponding to the highest matching score so as to realize the auxiliary judging of the case.

9. Terminal device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the case-aided trial method according to any one of claims 1 to 7 when the computer program is executed by the processor.

10. A computer readable storage medium storing a computer program, wherein the computer program when executed by a processor implements the case assist trial method of any one of claims 1 to 7.