CN112163752A - Auxiliary case division method based on convolutional neural network - Google Patents
Auxiliary case division method based on convolutional neural network Download PDFInfo
- Publication number
- CN112163752A CN112163752A CN202010995994.0A CN202010995994A CN112163752A CN 112163752 A CN112163752 A CN 112163752A CN 202010995994 A CN202010995994 A CN 202010995994A CN 112163752 A CN112163752 A CN 112163752A
- Authority
- CN
- China
- Prior art keywords
- case
- judge
- matching
- neural network
- judges
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 21
- 239000013598 vector Substances 0.000 claims abstract description 28
- 238000011176 pooling Methods 0.000 claims description 6
- 238000012549 training Methods 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 4
- 239000013604 expression vector Substances 0.000 claims description 4
- 238000000638 solvent extraction Methods 0.000 claims 3
- 238000013507 mapping Methods 0.000 abstract description 5
- 238000004364 calculation method Methods 0.000 abstract description 4
- 238000005516 engineering process Methods 0.000 abstract description 2
- 238000013441 quality evaluation Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 235000019580 granularity Nutrition 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06311—Scheduling, planning or task assignment for a person or group
- G06Q10/063112—Skill-based matching of a person or a group to a task
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/18—Legal services
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Entrepreneurship & Innovation (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Educational Administration (AREA)
- Evolutionary Biology (AREA)
- General Business, Economics & Management (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Technology Law (AREA)
- Primary Health Care (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses an auxiliary case division method based on a convolutional neural network, which comprises the following steps: firstly, proposing a judge representation method fusing judge quality, selecting the characteristics of any judge through the judge quality weight of the judge case, and selecting the characteristics of the case with high judge quality of the judge to represent the judge; secondly, a case representation method is provided to obtain a semantic feature representation vector of the case; thirdly, calculating the matching degree of the case and the judge by using a convolutional neural network; and fourthly, outputting the first N judges with high matching values as recommended judges of the case through sorting the matching degrees, and realizing automatic case division. The method comprises the steps of mapping sentence representations to a high-order semantic space by using a word embedding technology, automatically extracting associated semantics between case representations and judge representations through a convolutional neural network, then realizing matching degree calculation of cases and judges in a nonlinear space, obtaining the matching degree of the cases and the judges, obtaining recommended judges through a case classification module, and realizing automatic case classification.
Description
Technical Field
The invention relates to a court case division method, in particular to an auxiliary case division method based on a convolutional neural network, and belongs to the technical field of data mining, natural language processing and deep learning.
Background
Case distribution is an important link of litigation procedures and also an important content of trial management, and has a key effect of moving the whole body in a single action on reasonably allocating court trial resources, exciting the enthusiasm of judges for case handling, promoting justice to be clean and fair, improving the quality of trial and effect. Throughout the historical context of the improvement of the national court split system, manual assigned split cases are approximately developed into computer random split cases. In recent years, the highest people's court proposes to establish a case allocation system which mainly adopts random case allocation and assists in appointed case allocation, and the case allocation needs to consider the factors such as case type, difficulty degree and the like. The division method of court at all levels in China is still simple random division at present, and the concrete performance is balanced division aiming at the equal number of cases distributed by a judge, or completely random division without considering the professional ability and case property of the judge, such as number shaking division.
The traditional manual case distribution and simple random case distribution mechanism has the problems of manual excessive intervention, uncomfortable human case and the like, and the existing case distribution method cannot adapt to a novel case handling mechanism along with the national improvement of establishing a smart court and a staff rating system. Under the background of the reform of the membership system, the judge team is further specialized, elite and occupational. The invention considers that it is reasonable to distribute cases to judges of lawbreakers who are adept at judging cases, and the cases also meet the requirement of the reform of judicial system. In order to solve the problem of discomfortable case in the existing court system, the invention aims to automatically distribute the case to the officers who are good at judging the case, so as to realize special case application, form a specialized case application mode, avoid the disadvantages of personal case, relationship case, money case and the like, and improve the case application quality. However, there are two difficulties in implementing automatic case division. The first difficulty is presentation difficulty. The court system stores basic information and historical trial data of judges. Most of them are text information and metadata. How to fuse abstract semantic features of a judge in the expression of the judge and embody the field of judgment adept by the judge is a difficulty in realizing automatic case division. A second difficulty is matching difficulty. How to automatically map the case representation and the judge representation to a high-price semantic space, automatically acquire the associated semantic information in the case representation and the judge representation, and calculate the matching degree of the case and the judge is another difficulty in realizing automatic case division.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: a Convolution Neural network-based auxiliary case division method is provided, wherein the field of judgment adequacy of a judge is fused in the expression of the judge, then, double-Channel Convolution Neural Networks (DCNN) are adopted to extract abstract semantic features with different granularities in the expression of a case and the expression of the judge, automatic matching of the judge and the case is realized, and according to the matching degree of the case and a plurality of judges, the first N judges with high matching values are output as recommended judges of the case, so that the problems of manual excessive intervention, discomfort of the case and the like existing in the traditional manual case division and simple random case division mechanisms are solved.
The technical scheme of the invention is as follows: an auxiliary case dividing method based on a convolutional neural network, the method comprises the following steps: the judge representation method comprises the following steps of selecting the characteristics of any judge through the judge quality weight of the judge case, selecting the characteristics of the case with higher judge quality of the judge to represent the judge, wherein the judge quality weight is as follows:wherein: w is aijThe judgment quality weight of the judge on any case is represented, i is 1 … n, j is 1 … m, the case number is represented, theta represents the case settlement rate within the legal normal examination limit, alpha represents the rate of sending the first examination and the second examination, and beta represents the trial time of the cases; secondly, a case representation method is provided to obtain a semantic feature representation vector of the case; thirdly, calculating the matching degree of the case and the judge by using a convolutional neural network; fourthly, matching degree is sorted through sortingAnd outputting the first N judges with high matching values as recommended judges of the case, and realizing automatic case division.
In the second step, parameters are set to change case semantic features forming the expression of the judge, when the value is greater than 0.5, cases which show that more than 50% of case semantic features forming the expression of the judge are cases with higher judging quality of the judge, and the more the value is close to 1, the more abstract semantic features expressed by the judge can represent the judging field which is good at the judge.
In the third step, firstly, the expression texts of the judge and the case are mapped into low-dimensional real-value vectors through an embedded layer based on a pre-training word vector table, a model is input, valuable features are gradually extracted through a convolutional layer and a pooling layer to obtain abstract semantic feature expression vectors of the judge and the case, the two feature vectors are spliced and input into a full-connection layer to realize matching, finally probability weights of case distribution and case non-distribution to the judge are obtained through Softmax, the distributed probability is taken as the matching degree of the case and the judge, and the matching degree is output.
In the fourth step, one y can be obtained by matching the case with the officers in the law and the academy one by one, the matching degree of the case and the officers is obtained by extracting the probability distributed in the y, the recommended officers are obtained through the case division module, one case is matched with M officers to obtain M matching degrees, the M values are compared, and the officers with the top N matching values are output as N undertaking officers recommended for the case.
The invention has the beneficial effects that: compared with the prior art, the invention has the following contributions: firstly, a judge representation method fusing case judging quality is provided. And obtaining the judging quality weight of the judge under various cases through judging quality evaluation indexes. And (4) representing the judging field which is good in the judge by using the case semantic features with high judging quality of the judge. Thereby fusing abstract semantic information of the judge adept field in the judge representation. And secondly, the automatic matching of cases and judges is realized by utilizing a convolutional neural network. The method maps sentence expression to a high-order semantic space by using a word embedding technology, automatically extracts the associated semantics between case expression and judge expression through a convolutional neural network, and then realizes the matching degree calculation of the case and the judge in a nonlinear space, thereby obtaining good use effect.
Drawings
FIG. 1 is a sectional flow chart of the present invention;
FIG. 2 is a matching model based on a convolutional neural network of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings.
Example 1: as shown in fig. 1-2, an auxiliary case-splitting method based on a convolutional neural network includes the following steps: the judge representation method comprises the following steps of selecting the characteristics of any judge through the judge quality weight of the judge case, selecting the characteristics of the case with higher judge quality of the judge to represent the judge, wherein the judge quality weight is as follows:wherein: w is aijThe judgment quality weight of the judge on any case is represented, i is 1 … n, j is 1 … m, the case number is represented, theta represents the case settlement rate within the legal normal examination limit, alpha represents the rate of sending the first examination and the second examination, and beta represents the trial time of the cases; secondly, a case representation method is provided to obtain a semantic feature representation vector of the case; thirdly, calculating the matching degree of the case and the judge by using a convolutional neural network; and fourthly, outputting the first N judges with high matching values as recommended judges of the case through sorting the matching degrees, and realizing automatic case division.
And in the second step, setting parameters to change case semantic features forming the expression of the judge, wherein when the value is greater than 0.5, cases which show that more than 50% of case semantic features forming the expression of the judge are cases with higher judging quality of the judge, and the more the value is close to 1, the more abstract semantic features expressed by the judge can embody the judging field which is good at the judge.
Firstly, obtaining the judging quality weight of a judge under various cases through judging quality evaluation indexes; secondly, case semantic features with high judge quality of the judge are used for representing the judge field adept by the judge, so that abstract semantic information of the judge field adept is fused in judge representation.
In the third step, firstly, mapping the expression texts of the judge and the case into low-dimensional real-valued vectors through an embedded layer based on a pre-training word vector table, inputting a model, gradually extracting valuable features through a convolutional layer and a pooling layer to obtain abstract semantic feature expression vectors of the judge and the case, splicing the two feature vectors, inputting the two feature vectors into a full-connection layer to realize matching, finally obtaining probability weights of case distribution and case non-distribution to the judge through Softmax, taking the distributed probability as the matching degree of the case and the judge, and outputting the matching degree.
Mapping sentence expression to a high-order semantic space by utilizing word embedding; and secondly, automatically extracting the associated semantics between the case representation and the judge representation through a convolutional neural network, and then realizing the matching degree calculation of the case and the judge in a nonlinear space.
And in the fourth step, matching the cases and the judges of the law institution one by one to obtain y, obtaining the matching degree of the cases and the judges by extracting the probability distributed in the y, obtaining the recommended judges through a case division module, matching M judges by one case to obtain M matching degrees, comparing the M values, and outputting the judges with the previous N matching values as N undertaking judges recommended for the cases.
The case division flow of the invention is shown in figure 1 and mainly comprises a case judging quality evaluation module, a representation module, a matching degree estimation module and a case division module. The whole working process is as follows: firstly, preprocessing the data and obtaining the representation of cases and judges through a representation module. And secondly, automatically evaluating the matching degree of the case and the judge through a matching degree evaluation module to obtain the matching degree of the case and any judge. And finally, sorting the matching degrees of the cases and all judges according to the value based on a case division module, and outputting the first N judges with high matching values as N undertaking judges recommended for the cases.
To clarify the advantages and technical solutions of the present invention, the following detailed description of the present invention is provided with reference to the accompanying drawings.
1. Judge representation method fusing case judgment quality
The traditional case distribution only takes case law as a unique case distribution standard, ignores the characteristic that a judge is good at judging the field, cannot ensure that the judge distributed to the case is good at judging the case, and often causes the case to be unsuitable. In order to solve the problem, the invention provides a judge representation method fusing case judging quality so as to highlight the expertise of a judge. The criminal has many historical trial cases, and different cases have the trial quality. The case with higher judging quality of the judge is considered to be the case which is good at judging by the judge, and the expertise, case judgment thinking and judging habits of the judge can be reflected by utilizing the case semantic information.
In 2011, the highest people court publishes 31 indexes for evaluating the overall case judgment quality of the court in the 'guidance opinions about the development of case quality evaluation work'. The invention selects three indexes of reexamination rate, case trial time and case rate in legal normal trial limit to evaluate the case trial quality of the judge person. The judgment quality weight calculation formula of the judge on any case is as follows:wherein, wijThe judgment quality weight of the judge on any case is expressed, i is 1 … n and j is 1 … m. Theta represents the case settlement rate within the legal normal trial limit, alpha represents the re-trial rate sent back by the first trial, and beta represents the trial time of all cases.
Under any case, the judge has a judging quality weight w, and the judge quality of case type judges with larger value is higher compared with the w values of the same judge under different types. We consider this type of case to be the case that judges well by law.
Through the analysis of referee documents, the fact description of the case is found to be the main basis for case judgment, so that the fact description of the case is extracted from the document to form the representation of the case. According to the above analysis, the expression of the judge is composed of case characteristics of a plurality of cases. In the specific design, the number of cases constituting the judge is defined as 17. Parameters are set to change the case semantic features that make up the judge representation. The number of cases showing the number of cases constituting the case of the judge belongs to the case category having a high judging quality. When the value is more than 0.5, the case with higher justice quality of the judge is represented by more than 50% of the case semantic features forming the judge representation. We believe that the closer the value is to 1, the more abstract semantics the judge can represent that the judge is adept at.
2. Automatic case and judge matching method based on convolutional neural network
According to the method, a two-channel convolution neural network model is built, case and judge representation texts are respectively processed to obtain abstract semantic representations of the cases and judges, and therefore the matching degree of the judges and the cases is automatically evaluated. The model structure is shown in fig. 2.
The method comprises the steps of utilizing a two-channel convolutional neural network to achieve matching of a case and an officer, firstly mapping expression texts of the officer and the case into low-dimensional real-value vectors through an embedded layer based on a pre-training word vector table, inputting a model, gradually extracting valuable features through a convolutional layer and a pooling layer to obtain abstract semantic feature expression vectors of the officer and the case, splicing the two feature vectors, inputting the two feature vectors into a full-connection layer to achieve matching, finally obtaining probability weights of case distribution and case non-distribution to the officer through Softmax, taking the distributed probability as the matching degree of the case and the officer, and outputting the matching degree.
The representations of judges and cases are both textual constructs. The above process can be formalized as follows:
let L equal to L1,l2,l3… is a description of case facts, where liIs the ith word in the text. At the embedding layer, a word vector table W based on the pre-training wiki, each liIs mapped to a vector. Wherein W ∈ RS×KS denotes a dictionary size and K denotes a vector dimension. Assuming that the text sequence length of the input model is s, the vector representation sequence obtained by mapping L is X ═ X1,x2,…,xs]Wherein x isi∈RK。
The vector representations of the case, the judge and the embedded layer are X respectively1And X2. And (4) respectively extracting the characteristics of the two convolution layers with the same parameter setting. Let convolution kernel Wc∈Rh×KConvolution operation is carried out on the h word vectors to obtain a new characteristic ci. Applying a convolution kernel to each possible vector window { L } in a text sequence1:h,L2:h+1,…,Ln-h+1:nGet a characteristic diagram c ═ c }1,c2,…,cn-h+1]。
And c, processing by a pooling layer to further obtain more valuable characteristics in the text, wherein the method adopts maximum pooling. Is shown as An abstract semantic feature representation vector representing a sentence. The invention applies a plurality of convolution kernels of different sizes to fully extract text features. Multiple different sizes of convolution kernels may be obtainedCombining a plurality of feature vectorsIs spliced intoThenAn abstract semantic feature representation vector representing a sentence. Then corresponds to X1Can obtain the productCorresponds to X2Can obtain the productWill be provided withAndinputting the full connection layer after splicing to adjust the parameters of the whole model,and the probability weight y of allocation and non-allocation is obtained through Softmax.
Matching cases and legal officers in the law academy one by one to obtain y, and extracting the probability distributed in y to obtain the matching degree of the cases and the legal officers. And obtaining a recommended judge through a case division module. And matching M judges on one case to obtain M matching degrees, comparing the M values, and outputting the judges with the top N matching values as N undertaking judges recommended for the case.
The present invention is not described in detail, but is known to those skilled in the art. Finally, the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting, although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions may be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all of them should be covered in the claims of the present invention.
Claims (4)
1. An auxiliary case division method based on a convolutional neural network is characterized in that: the method comprises the following steps: the judge representation method is characterized in that the characteristics of any judge are selected according to the judge quality weight of the judge case, the characteristics of the case with high judge quality of the judge are selected to represent the judge, and the judge quality weight isWherein: w is aijThe judgment quality weight of the judge on any case is represented, i is 1 … n, j is 1 … m, the case number is represented, theta represents the case settlement rate within the legal normal examination limit, alpha represents the rate of sending the first examination and the second examination, and beta represents the trial time of the cases; secondly, a case representation method is provided to obtain a semantic feature representation vector of the case; thirdly, calculating the matching degree of the case and the judge by using a convolutional neural network; and fourthly, outputting the first N judges with high matching values as recommended judges of the case through sorting the matching degrees, and realizing automatic case division.
2. The convolutional neural network-based aided partitioning method of claim 1, wherein: in the second step, parameters are set to change case semantic features forming the expression of the judge, when the value is greater than 0.5, cases which show that more than 50% of case semantic features forming the expression of the judge are cases with high judging quality of the judge, and the more the value is close to 1, the more abstract semantic features expressed by the judge can represent the judging field which is good at the judge.
3. The convolutional neural network-based aided partitioning method of claim 1, wherein: in the third step, firstly, the expression texts of the judge and the case are mapped into low-dimensional real-value vectors through an embedded layer based on a pre-training word vector table, a model is input, valuable features are gradually extracted through a convolutional layer and a pooling layer to obtain abstract semantic feature expression vectors of the judge and the case, the two feature vectors are spliced and input into a full-connection layer to realize matching, finally probability weights of case distribution and case non-distribution to the judge are obtained through Softmax, the distributed probability is taken as the matching degree of the case and the judge, and the matching degree is output.
4. The convolutional neural network-based aided partitioning method of claim 1, wherein: in the fourth step, one y can be obtained by matching the case with the officers in the law and the academy one by one, the matching degree of the case and the officers is obtained by extracting the probability distributed in the y, the recommended officers are obtained through the case division module, one case is matched with M officers to obtain M matching degrees, the M values are compared, and the officers with the top N matching values are output as N undertaking officers recommended for the case.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010995994.0A CN112163752A (en) | 2020-09-21 | 2020-09-21 | Auxiliary case division method based on convolutional neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010995994.0A CN112163752A (en) | 2020-09-21 | 2020-09-21 | Auxiliary case division method based on convolutional neural network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112163752A true CN112163752A (en) | 2021-01-01 |
Family
ID=73864281
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010995994.0A Pending CN112163752A (en) | 2020-09-21 | 2020-09-21 | Auxiliary case division method based on convolutional neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112163752A (en) |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103353888A (en) * | 2013-07-03 | 2013-10-16 | 福建榕基软件股份有限公司 | System and method for automatically allocating court judgment cases |
CN106547886A (en) * | 2016-11-03 | 2017-03-29 | 广东欧珀移动通信有限公司 | Application program recommends method, device, terminal and server |
CN107665212A (en) * | 2016-07-28 | 2018-02-06 | 北京国双科技有限公司 | A kind of method and device for realizing law court's division |
CN109872052A (en) * | 2019-01-28 | 2019-06-11 | 广州大学 | A kind of law court's case intelligence division householder method and system |
CN110009182A (en) * | 2019-01-28 | 2019-07-12 | 广州大学 | A kind of law court's case automation division method and system |
CN110119905A (en) * | 2019-05-23 | 2019-08-13 | 贵州大学 | A kind of accurate division method based on binary channels convolutional neural networks |
CN110765760A (en) * | 2019-09-03 | 2020-02-07 | 平安科技(深圳)有限公司 | Legal case distribution method and device, storage medium and server |
-
2020
- 2020-09-21 CN CN202010995994.0A patent/CN112163752A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103353888A (en) * | 2013-07-03 | 2013-10-16 | 福建榕基软件股份有限公司 | System and method for automatically allocating court judgment cases |
CN107665212A (en) * | 2016-07-28 | 2018-02-06 | 北京国双科技有限公司 | A kind of method and device for realizing law court's division |
CN106547886A (en) * | 2016-11-03 | 2017-03-29 | 广东欧珀移动通信有限公司 | Application program recommends method, device, terminal and server |
CN109872052A (en) * | 2019-01-28 | 2019-06-11 | 广州大学 | A kind of law court's case intelligence division householder method and system |
CN110009182A (en) * | 2019-01-28 | 2019-07-12 | 广州大学 | A kind of law court's case automation division method and system |
CN110119905A (en) * | 2019-05-23 | 2019-08-13 | 贵州大学 | A kind of accurate division method based on binary channels convolutional neural networks |
CN110765760A (en) * | 2019-09-03 | 2020-02-07 | 平安科技(深圳)有限公司 | Legal case distribution method and device, storage medium and server |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111538894B (en) | Query feedback method and device, computer equipment and storage medium | |
CN108509484B (en) | Classifier construction and intelligent question and answer method, device, terminal and readable storage medium | |
CN109460737A (en) | A kind of multi-modal speech-emotion recognition method based on enhanced residual error neural network | |
Jin et al. | Uncertain probabilistic linguistic term sets in group decision making | |
CN110765260A (en) | Information recommendation method based on convolutional neural network and joint attention mechanism | |
CN110287323B (en) | Target-oriented emotion classification method | |
CN110674252A (en) | High-precision semantic search system for judicial domain | |
CN110516057B (en) | Petition question answering method and device | |
CN109753602A (en) | A kind of across social network user personal identification method and system based on machine learning | |
CN110245228A (en) | The method and apparatus for determining text categories | |
CN113326374B (en) | Short text emotion classification method and system based on feature enhancement | |
CN111858940A (en) | Multi-head attention-based legal case similarity calculation method and system | |
CN115455171B (en) | Text video mutual inspection rope and model training method, device, equipment and medium | |
CN108920521A (en) | User's portrait-item recommendation system and method based on pseudo- ontology | |
CN115952292B (en) | Multi-label classification method, apparatus and computer readable medium | |
CN111177386A (en) | Proposal classification method and system | |
CN110222192A (en) | Corpus method for building up and device | |
CN116467443A (en) | Topic identification-based online public opinion text classification method | |
CN112633690A (en) | Service personnel information distribution method, service personnel information distribution device, computer equipment and storage medium | |
CN112988970A (en) | Text matching algorithm serving intelligent question-answering system | |
CN111581364A (en) | Chinese intelligent question-answer short text similarity calculation method oriented to medical field | |
CN110532378B (en) | Short text aspect extraction method based on topic model | |
CN113516094B (en) | System and method for matching and evaluating expert for document | |
CN108509588B (en) | Lawyer evaluation method and recommendation method based on big data | |
CN111859955A (en) | Public opinion data analysis model based on deep learning |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210101 |
|
RJ01 | Rejection of invention patent application after publication |