CN117391466A - Novel early warning method and system for contradictory dispute cases - Google Patents

Novel early warning method and system for contradictory dispute cases Download PDF

Info

Publication number
CN117391466A
CN117391466A CN202311318564.5A CN202311318564A CN117391466A CN 117391466 A CN117391466 A CN 117391466A CN 202311318564 A CN202311318564 A CN 202311318564A CN 117391466 A CN117391466 A CN 117391466A
Authority
CN
China
Prior art keywords
vector
content
case
current time
time
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
Application number
CN202311318564.5A
Other languages
Chinese (zh)
Inventor
王佐成
刘跃
马韵洁
王晓思
罗艳丽
王飞
邱俊云
彭萌
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Institute of Advanced Technology University of Science and Technology of China
Original Assignee
Institute of Advanced Technology University of Science and Technology of China
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Institute of Advanced Technology University of Science and Technology of China filed Critical Institute of Advanced Technology University of Science and Technology of China
Priority to CN202311318564.5A priority Critical patent/CN117391466A/en
Publication of CN117391466A publication Critical patent/CN117391466A/en
Pending legal-status Critical Current

Links

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
    • G06Q10/0637—Strategic management or analysis, e.g. setting a goal or target of an organisation; Planning actions based on goals; Analysis or evaluation of effectiveness of goals
    • G06Q10/06375—Prediction of business process outcome or impact based on a proposed change
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/253—Fusion techniques of extracted features
    • 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
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
    • 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
    • G06N3/0455—Auto-encoder networks; Encoder-decoder 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0464—Convolutional networks [CNN, ConvNet]
    • 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
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • 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
    • G06Q10/0635—Risk analysis of enterprise or organisation activities
    • 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)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Economics (AREA)
  • General Health & Medical Sciences (AREA)
  • Strategic Management (AREA)
  • Biomedical Technology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Molecular Biology (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Tourism & Hospitality (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Educational Administration (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Game Theory and Decision Science (AREA)
  • Development Economics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Primary Health Care (AREA)
  • Technology Law (AREA)

Abstract

The application discloses a novel early warning method and system for contradictory disputes, wherein the method comprises the following steps: acquiring a first content vector of a case to be processed and all related incident objects; determining historical cases of each related object in a target historical period, and acquiring a second content vector of each historical case; inputting the first content vector and the second content vector into a trained comprehensive risk recognition model to obtain a target risk level of a case to be processed, wherein the comprehensive risk recognition model comprises a first feature extraction model and a second feature extraction model, the first feature extraction model is used for extracting features based on spatial relevance, and the second feature extraction model is used for extracting features based on temporal relevance; and carrying out corresponding early warning operation on the cases to be processed according to the target risk level, so that the early warning operation can be flexibly carried out on legal cases of various contradictory disputes, the phenomenon that accumulated cases are difficult to process due to unclear responsibility is reduced, and the case processing efficiency is improved.

Description

Novel early warning method and system for contradictory dispute cases
Technical Field
The application belongs to the technical field of data processing, and particularly relates to a novel early warning method and system for contradictory dispute cases.
Background
Along with the large transformation of social and economic aspects, various legal cases of contradiction and dispute are frequently sent out, so that a rapid processing method for the legal cases of the contradiction and dispute is urgently needed, and the case processing efficiency is improved.
Disclosure of Invention
The present application aims to solve at least one of the technical problems existing in the prior art. Therefore, the novel early warning method and system for the contradictory disputes can effectively and flexibly execute early warning operation for various legal cases of the contradictory disputes, and is beneficial to reducing the phenomenon of accumulation and difficult cases caused by unclear responsibility.
In a first aspect, the present application provides a method for early warning of a novel contradictory dispute case, including:
acquiring a first content vector of a case to be processed and all related incident objects;
determining historical cases of each related object in a target historical period, and acquiring a second content vector of each historical case;
inputting the first content vector and the second content vector into a trained comprehensive risk recognition model to obtain a target risk level of the to-be-processed case, wherein the comprehensive risk recognition model comprises a first feature extraction model and a second feature extraction model, the first feature extraction model is used for extracting features based on spatial relevance, and the second feature extraction model is used for extracting features based on time relevance;
And carrying out corresponding early warning operation on the to-be-processed case according to the target risk level.
In some embodiments, the integrated risk recognition model further includes a classification model connected to the first feature extraction model and the second feature extraction model, the classification model including an attention module and an output layer, the inputting the first content vector and the second content vector into the trained integrated risk recognition model, to obtain the target risk level of the case to be processed, including:
inputting the first content vector into the first feature extraction model to obtain a first feature vector corresponding to the to-be-processed case, and inputting the second content vector into the second feature extraction model to obtain a second feature vector corresponding to the historical case;
inputting the first feature vector and the second feature vector into the attention module to obtain a fusion feature vector;
and inputting the fusion feature vector into the output layer to obtain the target risk level of the case to be processed.
In some embodiments, the second feature extraction model includes a gating loop unit and a variation self-encoder that are connected, and the inputting the second content vector into the second feature extraction model, to obtain a second feature vector corresponding to the historical case, includes:
Organizing all the second content vectors corresponding to each related object into a time sequence, wherein each time in the time sequence corresponds to a group of the second content vectors;
inputting each time sequence into the gating circulation unit for processing;
in the processing process of the gating circulation unit, a group of second content vectors corresponding to the current time are obtained from the time sequence, and the coding vectors output by the encoder at the last time of the variation and the hiding state vectors output by the gating circulation unit at the last time are obtained;
determining an input vector of the current time of the gating circulation unit according to the coding vector output at the previous time, the hidden state vector output at the previous time and the group of second content vectors of the current time, and inputting the input vector into the gating circulation unit to obtain the hidden state vector output at the current time of the gating circulation unit; and then, inputting the hidden state vector output by the current time of the gating circulating unit into the variation self-encoder to obtain the encoded vector output by the current time of the variation self-encoder, updating the next time to the current time, and returning to execute the step of acquiring a group of second content vectors corresponding to the current time from the time sequence.
In some embodiments, the determining the input vector for the current time of the gating loop unit according to the encoded vector output at the last time, the hidden state vector output at the last time, and the set of second content vectors for the current time comprises:
splicing the coding vector output at the previous time and the hidden state vector output at the previous time to obtain a spliced vector at the current time;
and taking the spliced vector of the current time and the group of second content vectors of the current time as input vectors of the current time of the gating circulating unit.
In some embodiments, the inputting the first feature vector and the second feature vector into the attention module, to obtain a fused feature vector, includes:
mapping, by the attention module, the first feature vector and the second feature vector into a same vector space;
and carrying out weighted summation on the mapped first feature vector and the mapped second feature vector to obtain a fusion feature vector.
In some embodiments, the method for early warning of a novel contradictory dispute case further includes:
acquiring a first sample content vector of each case sample in the case sample set, a related sample related object and a risk level label;
Acquiring a sample history case of each sample related object in a sample history period, and acquiring a second sample content vector of each sample history case;
training the created comprehensive risk identification model by using the first sample content vector, the second sample content vector and the risk level label.
In a second aspect, the present application provides a novel early warning system for contradictory disputes, including:
the first acquisition module is used for acquiring a first content vector of a case to be processed and all related incident objects;
the second acquisition module is used for determining historical cases of each related object in a target historical period and acquiring a second content vector of each historical case;
the recognition module is used for inputting the first content vector and the second content vector into a trained comprehensive risk recognition model to obtain a target risk level of the case to be processed, the comprehensive risk recognition model comprises a first feature extraction model and a second feature extraction model, the first feature extraction model is used for carrying out feature extraction based on spatial relevance, and the second feature extraction model is used for carrying out feature extraction based on temporal relevance;
And the early warning module is used for carrying out corresponding early warning operation on the to-be-processed case according to the target risk level.
In some embodiments, the integrated risk recognition model further includes a classification model connected to the first feature extraction model and the second feature extraction model, the classification model including an attention module and an output layer, the recognition module being specifically configured to:
inputting the first content vector into the first feature extraction model to obtain a first feature vector corresponding to the to-be-processed case, and inputting the second content vector into the second feature extraction model to obtain a second feature vector corresponding to the historical case;
inputting the first feature vector and the second feature vector into the attention module to obtain a fusion feature vector;
and inputting the fusion feature vector into the output layer to obtain the target risk level of the case to be processed.
In some embodiments, the second feature extraction model includes a connected gating loop unit and a variation self-encoder, and the identification module is specifically configured to:
organizing all the second content vectors corresponding to each related object into a time sequence, wherein each time in the time sequence corresponds to a group of the second content vectors;
Inputting each time sequence into the gating circulation unit for processing;
in the processing process of the gating circulation unit, a group of second content vectors corresponding to the current time are obtained from the time sequence, and the coding vectors output by the encoder at the last time of the variation and the hiding state vectors output by the gating circulation unit at the last time are obtained;
determining an input vector of the current time of the gating circulation unit according to the coding vector output at the previous time, the hidden state vector output at the previous time and the group of second content vectors of the current time, and inputting the input vector into the gating circulation unit to obtain the hidden state vector output at the current time of the gating circulation unit; and then, inputting the hidden state vector output by the current time of the gating circulating unit into the variation self-encoder to obtain the encoded vector output by the current time of the variation self-encoder, updating the next time to the current time, and returning to execute the step of acquiring a group of second content vectors corresponding to the current time from the time sequence.
In some embodiments, the identification module is specifically configured to:
The determining the input vector of the current time of the gating loop unit according to the coding vector output at the previous time, the hidden state vector output at the previous time and the set of second content vectors of the current time comprises:
splicing the coding vector output at the previous time and the hidden state vector output at the previous time to obtain a spliced vector at the current time;
and taking the spliced vector of the current time and the group of second content vectors of the current time as input vectors of the current time of the gating circulating unit.
In some embodiments, the identification module is specifically configured to:
mapping, by the attention module, the first feature vector and the second feature vector into a same vector space;
and carrying out weighted summation on the mapped first feature vector and the mapped second feature vector to obtain a fusion feature vector.
In some embodiments, the early warning system for a new contradictory dispute case further includes a training module configured to:
acquiring a first sample content vector of each case sample in the case sample set, a related sample related object and a risk level label;
Acquiring a sample history case of each sample related object in a sample history period, and acquiring a second sample content vector of each sample history case;
training the created comprehensive risk identification model by using the first sample content vector, the second sample content vector and the risk level label.
In a third aspect, the present application provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method for pre-warning of a novel contradictory dispute case of any one of the above.
In a fourth aspect, the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor, where the processor implements the method for early warning of a novel contradictory dispute case according to any one of the above when executing the program.
The early warning method and the early warning system for the novel contradictory dispute cases are provided by the embodiment of the application, and the first content vector of the case to be processed and all related incident objects are obtained; determining historical cases of each related object in a target historical period, and acquiring a second content vector of each historical case; inputting the first content vector and the second content vector into a trained comprehensive risk recognition model to obtain a target risk level of a case to be processed, wherein the comprehensive risk recognition model comprises a first feature extraction model and a second feature extraction model, the first feature extraction model is used for extracting features based on spatial relevance, and the second feature extraction model is used for extracting features based on temporal relevance; and carrying out corresponding early warning operation on the cases to be processed according to the target risk level, so that the early warning operation can be flexibly carried out on legal cases of various contradictory disputes, the phenomenon that accumulated cases are difficult to process due to unclear responsibility is reduced, and the case processing efficiency is improved.
Drawings
The foregoing and/or additional aspects and advantages of the present application will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings, wherein:
fig. 1 is a flow chart of a method for early warning of a novel contradictory dispute provided in an embodiment of the present application;
FIG. 2 is another flow chart of a method for early warning of a new contradictory dispute provided by an embodiment of the present application;
FIG. 3 is a schematic diagram of a data processing process of a second feature extraction model according to an embodiment of the present application;
FIG. 4 is a schematic structural diagram of a warning system for a novel contradictory dispute provided in an embodiment of the present application;
fig. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present application;
fig. 6 is a schematic hardware structure of an electronic device according to an embodiment of the present application.
Detailed Description
Embodiments of the present application are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals refer to the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary only for the purpose of explaining the present application and are not to be construed as limiting the present application.
The embodiment of the application provides a novel early warning method, a novel early warning system, a novel storage medium and novel electronic equipment for contradictory disputes.
Referring to fig. 1, fig. 1 is a flowchart of a method for early warning of a novel contradictory dispute according to an embodiment of the present application. The novel early warning method for the contradictory disputes is applied to electronic equipment, wherein the electronic equipment comprises a mobile phone, a tablet personal computer, a personal computer (personal computer, PC), wearable electronic equipment (such as a smart watch), augmented reality (augmented reality, AR) equipment or Virtual Reality (VR) equipment and the like. Specifically, the early warning method of the novel contradictory dispute comprises the following steps 101-104, wherein:
101. a first content vector of the case to be processed and all related incident objects are obtained.
The cases to be processed mainly refer to legal cases of contradictory disputes, which can be manually input by a user, or can be automatically collected from a plurality of legal channels of contradictory disputes through interfaces, crawlers or other automatic modes, wherein the legal channels are official legal websites such as Chinese judge paperwork networks, chinese judge flow information disclosure networks and the like. The first content vector is obtained by preprocessing data of case contents of the case to be processed, and can be expressed as a word vector or a word vector. The data preprocessing is mainly divided into three steps, namely data cleaning, data conversion and data standardization. The data cleaning comprises the steps of removing null values, abnormal values, repeated values and the like so as to ensure the accuracy and the integrity of the data; the data conversion comprises the steps of numerical value, coding and feature selection, and the data standardization comprises the steps of normalization, standardization, dimension reduction and the like so as to improve the efficiency and the accuracy of model processing.
The case content mainly comprises four dimensions of content: event information, personnel information, evidence materials and social influence, wherein the event information comprises dispute problem categories, time spans, places, levels and the like of the cases, is mainly used for analyzing the theme and the characteristics of the contradictory disputes, and the levels can comprise county level, city level, province level, country level and the like. The personnel information comprises the job, unit, age and the like of the related object, and is mainly used for analyzing the identity and the background of the related object. The related objects mainly comprise parties subject to contradiction disputes, such as individuals and individuals, individuals and units, or units and units, etc. For legal cases of different legal types, the related objects may have different designations, such as for civil litigation cases, the related objects generally include textual notices and interviews. Legal types may include civil, economic, marital, administrative, and so forth. Evidence materials can be presented in multiple modalities, such as text, images, and audio, and are primarily used to fully understand contradictory disputes. The social influence comprises media attention, public opinion popularity and the like, and is mainly used for measuring the social influence possibly caused by contradiction disputes.
For example, the content vector x (i.e., the first content vector described above) of legal cases for a single contradictory dispute class may be represented by the following notation: x= [ X ] 1 ,X 2 ,X 3 ,X 4 ]Wherein X is 1 To X 4 Four vectors corresponding to the four dimensions representing the case.
102. And determining historical cases of each related object in the target historical period, and acquiring a second content vector of each historical case.
Wherein the target history period may be set manually, such as the last five years or ten years, etc. The history cases mainly refer to cases which have occurred and have been finalized in the past. The second content vector is similar to the first content vector and is obtained by preprocessing data of case contents of historical cases.
103. Inputting the first content vector and the second content vector into a trained comprehensive risk recognition model to obtain a target risk level of the to-be-processed case, wherein the comprehensive risk recognition model comprises a first feature extraction model and a second feature extraction model, the first feature extraction model is used for extracting features based on spatial relevance, and the second feature extraction model is used for extracting features based on time relevance.
The first feature extraction model is mainly used for extracting features of the first content vector, the second feature extraction model is mainly used for extracting features of the second content vector, and the target risk level of the case to be processed is determined based on the features extracted by the two models. The target risk level is mainly used for indicating the risk level of the case to be processed, such as primary risk, secondary risk, tertiary risk and the like, and the smaller the value is, the higher the risk is.
In some embodiments, please refer to fig. 2, fig. 2 is another flow chart of a novel method for early warning of contradictory disputes provided in the embodiments of the present application, where the integrated risk recognition model further includes a classification model connected to the first feature extraction model and the second feature extraction model, the classification model includes an attention module and an output layer, and step 103 may specifically include 1031-1033, where:
1031. inputting the first content vector into the first feature extraction model to obtain a first feature vector corresponding to the to-be-processed case, and inputting the second content vector into the second feature extraction model to obtain a second feature vector corresponding to the historical case.
Wherein the first feature extraction model may comprise a CNN neural network model, the second feature extraction model may comprise a GRU gating loop unit and a VAE variation self-encoder, the VAE being a generation model that can learn potential distributions of data and can learn feature representations of the data through the encoder and decoder, the GRU being a recurrent neural network that can capture time-dependent relationships and patterns in time-series data. In legal cases of contradictory disputes, the GRU can be utilized to analyze historical case data of the case related object, and the trend and change of the related object in different time periods can be identified.
Specifically, the CNN neural network model firstly adopts convolution operation to extract the space feature of the first content vector (namely x) to obtain a convolution feature map, then carries out pooling operation on the convolution feature map, and then carries out dimension reduction treatment on the pooled convolution feature map through a flattening layer to obtain the first feature vector. For example, the convolution kernel size of the CNN neural network model may be set to be k×m, and the first content vector may be subjected to feature extraction by the convolution kernel to obtain a convolution feature map. Then, the convolved feature map is pooled through a pooling layer, the largest pooling or average pooling operation can be used at this time, the dimension is reduced and the important features are reserved, the pooling layer performs dimension reduction and downsampling through a 2×2 or 3×3 filter and a sliding window step (stride=2), after the features with the large weight are activated to the maximum extent, the pooled feature map is input into a flattening layer, and flattened into a one-dimensional vector serving as a first feature vector v cnn 。
In some embodiments, the second feature extraction model includes a GRU gating cyclic unit and a VAE variation self-encoder that are connected, and in this case, the step of inputting the second content vector into the second feature extraction model to obtain a second feature vector corresponding to the historical case may specifically include:
Organizing all the second content vectors corresponding to each related object into a time sequence, wherein each time in the time sequence corresponds to a group of the second content vectors;
inputting each time sequence into the gating circulation unit for processing;
in the processing process of the gating circulation unit, a group of second content vectors corresponding to the current time are obtained from the time sequence, and the coding vectors output by the encoder at the last time of the variation and the hiding state vectors output by the gating circulation unit at the last time are obtained;
determining an input vector of the current time of the gating circulation unit according to the coding vector output at the previous time, the hidden state vector output at the previous time and the group of second content vectors of the current time, and inputting the input vector into the gating circulation unit to obtain the hidden state vector output at the current time of the gating circulation unit; and then, inputting the hidden state vector output by the current time of the gating circulating unit into the variation self-encoder to obtain the encoded vector output by the current time of the variation self-encoder, updating the next time to the current time, and returning to execute the step of acquiring a group of second content vectors corresponding to the current time from the time sequence.
If the to-be-processed case includes n related objects, the second content vectors of all the history cases of each related object in the target history period are organized into a time sequence, for example, the time sequence corresponding to the ith related object may be y i ＝[y 1i ,y 2i ,...,y Ti ]，i∈{1,2,3,...,n}，y ti For time series y i A second content vector corresponding to a time T of T, T e {1,2,3, T, T depends on the target history period described above. The time units of the time series may be measured in terms of year, quarter, month, etc., e.g., if the time of the time series is in units of 2 months, then t=2 represents 3 rd to 4 th months, y 2i Representing the time series y i The second content vector corresponding to the historical case in the 3 rd month to the 4 th month.
Let us assume that the overall content vector of the historical case of all the incident-related objects of the case to be processed is represented by Y: y= [ Y ] 1 ,y 2 ,…,y n ]For each time series in Y, a second can be inputAnd processing in the feature extraction model, and extracting a corresponding second feature vector. For example, referring to fig. 3, fig. 3 is a schematic diagram illustrating a data processing procedure of the second feature extraction model according to the embodiment of the present application, where the GRU calculates a hidden state vector h at each time t of the time sequence t And uses the VAE to further encode this hidden state vector h t Obtaining a coding vector S corresponding to the time t t Then the vector S is encoded t And hidden state vector h t Splicing to obtain a spliced vector H corresponding to the time t t And splice vector H t A second content vector y corresponding to the next time t+1 (t+1)i Processing as input at next time t+1 of GRU, and looping until processing of all second content vectors in time series is completed, and finally splicing vector H T Then as a second feature vector.
It should be noted that, the second feature extraction model in this embodiment combines the GRU and the VAE model, and further uses the VAE to encode the hidden state of the time sequence to generate the latent variable on the basis that the GRU can capture the time dependency in the time sequence data, so as to enrich the expression of each time sequence feature, capture more complexity and change modes, and at the same time, compress the data to the low-dimensional latent variable, and reduce the calculation and memory requirements. Finally, the potential variables of the VAE code enhance the interpretability, clearly explain the contribution of each time series, and improve the interpretability and understandability of the model.
In some embodiments, the determining the input vector of the current time of the gating loop unit according to the encoded vector output at the previous time, the hidden state vector output at the previous time, and the set of second content vectors of the current time may specifically include:
splicing the coding vector output at the previous time and the hidden state vector output at the previous time to obtain a spliced vector at the current time;
and taking the spliced vector of the current time and the group of second content vectors of the current time as input vectors of the current time of the gating circulating unit.
For example, if the time series y is to be calculated i Splicing vector H of next time t+1 t The updated formulas for the GRU and VAE models can be expressed as:
r t+1 ＝σ(W r ·[H t ,y (t+1)i ])，
z t+1 ＝σ(W z ·[H t ,y (t+1)i ])，
H t ＝[h y ,S y ]，H y+1 ＝[h t+1 ,S y+1 ]，
wherein y is (t+1)i Time series y, which is the ith concern object i A second content vector h corresponding to the next time t+1 t Represents hidden state vector, h, generated by GRU current time t t After VAE coding, a coding vector S is generated t ，h t And S is equal to t After splicing, a spliced vector H is generated t ，r t+1 Is the output of the reset gate in the GRU model at time t+1, z t+1 Is the output of the update gate in the GRU model at time t +1,is a candidate hidden state at time t+1, +. r 、W 2 、W h Is a learnable weight matrix.
1032. And inputting the first feature vector and the second feature vector into the attention module to obtain a fusion feature vector.
Wherein the CNN model can be extracted from the first eigenvector v of the case to be processed cnn And the GRU model calendarSecond feature vector H extracted from history case T Fusion and splicing are carried out through an attention mechanism, so that the attention degree of the model to different features is enhanced. This helps to better integrate the features such as time series and space, and promote the performance of the model.
In some embodiments, the step 1032 may specifically include:
mapping, by the attention module, the first feature vector and the second feature vector into a same vector space;
and carrying out weighted summation on the mapped first feature vector and the mapped second feature vector to obtain a fusion feature vector.
The processing procedure of the attention module can be expressed as the following formula:
H map ＝W g ·H T ，V map ＝W c ·v cnn ，
V fusion ＝α·H map +(1-α)V map ，
wherein H is map 、V map Respectively H T And v cnn Mapping vectors with the same latitude after being mapped to the same vector space, wherein alpha is the weight in the attention mechanism, and the calculation formula of alpha is as follows:
e＝u·tanh(W gru H map +W cnn v cnn +b)，
wherein the similarity score α is obtained by applying a second feature vector H t And a first eigenvector v cnn And carrying out weighted summation and activating by using a tanh activation function. The normalized attention weight α is then obtained by softmax operation. u is a learnable weight vector, W gru And W is cnn Is a matrix of parameters that can be learned, and b is a bias term. D represents a set of times in the time series {1,2,3,..t }, T being an element in set D.
1033. And inputting the fusion feature vector into the output layer to obtain the target risk level of the case to be processed.
For example, the mathematical formula for a particular output layer can be expressed as:
V final ＝W out ·V fusion +b out ，
P＝softmax(V final )，
wherein W is out Is the weight matrix of the output layer, b out Is a bias term. P is a predictive probability distribution vector containing a plurality of risk levels. softmax () is the operation of softmax that is mapped to a probability distribution.
It should be noted that, because the classification model in the comprehensive risk recognition model processes the feature vectors extracted by the GRU model and the CNN model, the GRU model and the CNN model have strong automatic feature learning capability, can learn higher-level feature representations from original data, and are beneficial to extracting potential key information, and meanwhile, the classification model adopts an attention mechanism to fuse the feature vectors, so that the comprehensive risk recognition model can provide more accurate prediction results of risk grades, and is beneficial to carrying out early warning operations of different degrees for different risk grade cases.
104. And carrying out corresponding early warning operation on the to-be-processed case according to the target risk level.
Wherein, different risk levels represent different case characteristics and risk degrees, for example, legal cases of contradictory dispute types can be classified into four different risk levels: primary risk, secondary risk, tertiary risk, and quaternary risk, the following is a description of the various risk levels:
first-order risk (red warning): such cases present a real risk and may cause a significant social impact, or involve a large number of parties, or involve special situations such as urgent visits, out-of-order visits, etc. Cases are highly sensitive in social public opinion, requiring special attention and timely intervention.
Secondary risk (orange warning): the potential risk of the cases is high, and a certain contradiction and dispute risk can be caused in the future. Cases may involve general problems or procedural problems during the case handling process, requiring moderate attention and handling.
Three-level risk (yellow warning): the risk of the cases is small, and serious contradiction and dispute problems are unlikely to be caused under the general circumstances. Cases may involve general warnings, quarantines, complaints, etc., that require routine processing and tracking.
Four-level risk (blue pre-warning): the risk of the cases is extremely low, and the contradiction and dispute problems are seldom caused. No particular attention is generally required.
Specifically, different early warning operations can be set for different risk levels in advance, and then early warning is performed by directly determining the early warning operation corresponding to the target risk level in a matching mode. By the mode, the decision maker and the law enforcement department can better allocate resources, optimize case processing flow and provide better decision support.
For example, for a case with lower risk (such as a four-level risk), only the law enforcement staff currently and directly responsible for the case can be used as an early warning object, and early warning is performed by a way of marking the case with a blue label on a system, for a case with higher risk (such as a two-level risk), besides the law enforcement staff currently and directly responsible for the case as an early warning object, the lead of the law enforcement staff and even the staff of the upper-level law enforcement department can be used as the early warning object together, meanwhile, besides the orange label marking of the case on the system, an early warning report can be generated and actively pushed to the early warning objects or other related systems, so that the case is particularly valued and timely intervened, the phenomenon of forming a difficult-to-integrate case due to unclear responsibility is reduced, and the case processing efficiency and effect are effectively improved.
It is easy to understand that the comprehensive risk recognition model is obtained by collecting a large number of historical cases in advance, namely, the early warning method of the novel contradictory dispute case can further comprise the following steps:
acquiring a first sample content vector of each case sample in the case sample set, a related sample related object and a risk level label;
acquiring a sample history case of each sample related object in a sample history period, and acquiring a second sample content vector of each sample history case;
training the created comprehensive risk identification model by using the first sample content vector, the second sample content vector and the risk level label.
The case samples in the case sample set can be automatically collected from a plurality of contradictory dispute legal channels through interfaces, crawlers or other automatic modes. The risk level labels are manually labeled, including the four risk levels described above. The first sample content vector and the second sample content vector are also obtained after data preprocessing, and the training process of the comprehensive risk recognition model is similar to the application process, in the training process, the first feature extraction model is used for extracting the feature vector of the first sample content vector, the second feature extraction model is used for extracting the feature vector of the second sample content vector, when the predicted risk level of a single case sample is predicted through the comprehensive risk recognition model, an error value between the predicted risk level and a corresponding risk level label is required to be calculated through a loss function, and model parameters of each model in the comprehensive risk recognition model are reversely adjusted in an iterative mode through the error value.
As can be seen from the above, the method for early warning of a novel contradictory dispute case provided in the embodiments of the present application obtains the first content vector of the case to be processed and all related incident objects; determining historical cases of each related object in a target historical period, and acquiring a second content vector of each historical case; inputting the first content vector and the second content vector into a trained comprehensive risk recognition model to obtain a target risk level of a case to be processed, wherein the comprehensive risk recognition model comprises a first feature extraction model and a second feature extraction model, the first feature extraction model is used for extracting features based on spatial relevance, and the second feature extraction model is used for extracting features based on temporal relevance; and carrying out corresponding early warning operation on the cases to be processed according to the target risk level, so that the early warning operation can be flexibly carried out on legal cases of various contradictory disputes, the phenomenon that accumulated cases are difficult to process due to unclear responsibility is reduced, and the case processing efficiency is improved.
According to the method described in the above embodiment, the embodiment of the present application further provides a novel early warning system for a contradictory dispute case, which is configured to execute the steps in the novel early warning method for the contradictory dispute case. Referring to fig. 4, fig. 4 is a schematic structural diagram of an early warning system for a novel contradictory dispute according to an embodiment of the present application. The novel early warning system 200 for contradictory disputes is applied to electronic equipment, and comprises a first acquisition module 201, a second acquisition module 202, an identification module 203 and an early warning module 204, wherein:
A first obtaining module 201, configured to obtain a first content vector of a case to be processed and all related incident objects;
a second obtaining module 202, configured to determine a history case of each of the incident objects in a target history period, and obtain a second content vector of each of the history cases;
the recognition module 203 is configured to input the first content vector and the second content vector into a trained comprehensive risk recognition model, to obtain a target risk level of the case to be processed, where the comprehensive risk recognition model includes a first feature extraction model and a second feature extraction model, the first feature extraction model performs feature extraction based on spatial correlation, and the second feature extraction model performs feature extraction based on temporal correlation;
and the early warning module 204 is configured to perform corresponding early warning operation on the to-be-processed case according to the target risk level.
In some embodiments, the integrated risk recognition model further comprises a classification model connected to the first feature extraction model and the second feature extraction model, the classification model comprising an attention module and an output layer, the recognition module 203 being specifically configured to:
Inputting the first content vector into the first feature extraction model to obtain a first feature vector corresponding to the to-be-processed case, and inputting the second content vector into the second feature extraction model to obtain a second feature vector corresponding to the historical case;
inputting the first feature vector and the second feature vector into the attention module to obtain a fusion feature vector;
and inputting the fusion feature vector into the output layer to obtain the target risk level of the case to be processed.
In some embodiments, the second feature extraction model includes a connected gating loop unit and a variation self-encoder, and the identification module 203 is specifically configured to:
organizing all the second content vectors corresponding to each related object into a time sequence, wherein each time in the time sequence corresponds to a group of the second content vectors;
inputting each time sequence into the gating circulation unit for processing;
in the processing process of the gating circulation unit, a group of second content vectors corresponding to the current time are obtained from the time sequence, and the coding vectors output by the encoder at the last time of the variation and the hiding state vectors output by the gating circulation unit at the last time are obtained;
Determining an input vector of the current time of the gating circulation unit according to the coding vector output at the previous time, the hidden state vector output at the previous time and the group of second content vectors of the current time, and inputting the input vector into the gating circulation unit to obtain the hidden state vector output at the current time of the gating circulation unit; and then, inputting the hidden state vector output by the current time of the gating circulating unit into the variation self-encoder to obtain the encoded vector output by the current time of the variation self-encoder, updating the next time to the current time, and returning to execute the step of acquiring a group of second content vectors corresponding to the current time from the time sequence.
In some embodiments, the identification module 203 is specifically configured to:
the determining the input vector of the current time of the gating loop unit according to the coding vector output at the previous time, the hidden state vector output at the previous time and the set of second content vectors of the current time comprises:
splicing the coding vector output at the previous time and the hidden state vector output at the previous time to obtain a spliced vector at the current time;
And taking the spliced vector of the current time and the group of second content vectors of the current time as input vectors of the current time of the gating circulating unit.
In some embodiments, the identification module 203 is specifically configured to:
mapping, by the attention module, the first feature vector and the second feature vector into a same vector space;
and carrying out weighted summation on the mapped first feature vector and the mapped second feature vector to obtain a fusion feature vector.
In some embodiments, the early warning system for a new contradictory dispute case further includes a training module configured to:
acquiring a first sample content vector of each case sample in the case sample set, a related sample related object and a risk level label;
acquiring a sample history case of each sample related object in a sample history period, and acquiring a second sample content vector of each sample history case;
training the created comprehensive risk identification model by using the first sample content vector, the second sample content vector and the risk level label.
It should be noted that, the specific details of each module unit in the early warning system 200 of the novel contradictory dispute are described in detail in the embodiment of the early warning method of the novel contradictory dispute, and are not described herein again.
In some embodiments, the early warning system of the novel contradictory dispute in the embodiments of the present application may be an electronic device, or may be a component in the electronic device, such as an integrated circuit or a chip. The electronic device may be a terminal device. By way of example, the electronic device may be a mobile phone, tablet computer, notebook computer, palm computer, vehicle-mounted electronic device, mobile internet appliance (Mobile Internet Device, MID), augmented reality (augmented reality, AR)/Virtual Reality (VR) device, robot, wearable device, ultra-mobile personal computer, UMPC, netbook or personal digital assistant (personal digital assistant, PDA), etc., but may also be a server, network attached storage (Network Attached Storage, NAS), personal computer (personal computer, PC), television (TV), teller machine or self-service machine, etc., and the embodiments of the present application are not limited in particular.
In some embodiments, as shown in fig. 5, the embodiment of the present application further provides an electronic device 300, including a processor 301, a memory 302, and a computer program stored in the memory 302 and capable of running on the processor 301, where the program, when executed by the processor 301, implements each process of the foregoing novel early warning method embodiment of the contradictory dispute case, and the same technical effect can be achieved, so that repetition is avoided and redundant description is omitted herein.
The electronic device in the embodiment of the application includes the mobile electronic device and the non-mobile electronic device described above.
Fig. 6 is a schematic hardware structure of an electronic device implementing an embodiment of the present application.
The electronic device 400 includes, but is not limited to: radio frequency unit 401, network module 402, audio output unit 403, input unit 404, sensor 405, display unit 406, user input unit 407, interface unit 408, memory 409, and processor 410.
Those skilled in the art will appreciate that the electronic device 400 may also include a power source (e.g., a battery) for powering the various components, which may be logically connected to the processor 410 by a power management system to perform functions such as managing charge, discharge, and power consumption by the power management system. The electronic device structure shown in fig. 6 does not constitute a limitation of the electronic device, and the electronic device may include more or less components than shown, or may combine certain components, or may be arranged in different components, which are not described in detail herein.
It should be appreciated that in embodiments of the present application, the input unit 404 may include a graphics processor (Graphics Processing Unit, GPU) 4041 and a microphone 4042, with the graphics processor 4041 processing image data of still pictures or video obtained by an image capture device (e.g., a camera) in a video capture mode or an image capture mode. The display unit 406 may include a display panel 4061, and the display panel 4061 may be configured in the form of a liquid crystal display, an organic light emitting diode, or the like. The user input unit 407 includes at least one of a touch panel 4071 and other input devices 4072. The touch panel 4071 is also referred to as a touch screen. The touch panel 4071 may include two parts, a touch detection device and a touch controller. Other input devices 4072 may include, but are not limited to, a physical keyboard, function keys (e.g., volume control keys, switch keys, etc.), a trackball, a mouse, a joystick, and so forth, which are not described in detail herein.
Memory 409 may be used to store software programs as well as various data. The memory 409 may mainly include a first memory area storing programs or instructions and a second memory area storing data, wherein the first memory area may store an operating system, application programs or instructions (such as a sound playing function, an image playing function, etc.) required for at least one function, and the like. Further, the memory 409 may include volatile memory or nonvolatile memory, or the memory 409 may include both volatile and nonvolatile memory. The nonvolatile Memory may be a Read-Only Memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically Erasable EPROM (EEPROM), or a flash Memory. The volatile memory may be random access memory (Random Access Memory, RAM), static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (ddr SDRAM), enhanced SDRAM (Enhanced SDRAM), synchronous DRAM (SLDRAM), and Direct RAM (DRRAM). Memory 409 in embodiments of the present application includes, but is not limited to, these and any other suitable types of memory.
Processor 410 may include one or more processing units; the processor 410 integrates an application processor that primarily processes operations involving an operating system, user interface, application programs, etc., and a modem processor that primarily processes wireless communication signals, such as a baseband processor. It will be appreciated that the modem processor described above may not be integrated into the processor 410.
The embodiment of the application also provides a non-transitory computer readable storage medium, on which a computer program is stored, which when executed by a processor, implements each process of the above-mentioned novel early warning method embodiment of contradictory disputes, and can achieve the same technical effect, and in order to avoid repetition, the description is omitted here.
Wherein the processor is a processor in the electronic device described in the above embodiment. The readable storage medium includes computer readable storage medium such as computer readable memory ROM, random access memory RAM, magnetic or optical disk, etc.
The embodiment of the application also provides a computer program product, which comprises a computer program, and the computer program realizes the early warning method of the novel contradiction dispute case when being executed by a processor.
Wherein the processor is a processor in the electronic device described in the above embodiment. The readable storage medium includes computer readable storage medium such as computer readable memory ROM, random access memory RAM, magnetic or optical disk, etc.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element. Furthermore, it should be noted that the scope of the methods and apparatus in the embodiments of the present application is not limited to performing the functions in the order shown or discussed, but may also include performing the functions in a substantially simultaneous manner or in an opposite order depending on the functions involved, e.g., the described methods may be performed in an order different from that described, and various steps may also be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solutions of the present application may be embodied essentially or in a part contributing to the prior art in the form of a computer software product stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk), comprising several instructions for causing a terminal (which may be a mobile phone, a computer, a server, or a network device, etc.) to perform the methods described in the embodiments of the present application.
The embodiments of the present application have been described above with reference to the accompanying drawings, but the present application is not limited to the above-described embodiments, which are merely illustrative and not restrictive, and many forms may be made by those of ordinary skill in the art without departing from the spirit of the present application and the scope of the claims, which are also within the protection of the present application.
The terms first, second and the like in the description and in the claims, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged, as appropriate, such that embodiments of the present application may be implemented in sequences other than those illustrated or described herein, and that the objects identified by "first," "second," etc. are generally of a type and not limited to the number of objects, e.g., the first object may be one or more. Furthermore, in the description and claims, "and/or" means at least one of the connected objects, and the character "/", generally means that the associated object is an "or" relationship.
In the description of the present application, the meaning of "plurality" is two or more.
In the description of the present specification, reference to the terms "one embodiment," "some embodiments," "illustrative embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present application. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
While embodiments of the present application have been shown and described, it will be understood by those of ordinary skill in the art that: many changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the application, the scope of which is defined by the claims and their equivalents.

Claims (14)

1. The utility model provides a novel early warning method of contradiction dispute cases which is characterized in that the method comprises the following steps:
acquiring a first content vector of a case to be processed and all related incident objects;
determining historical cases of each related object in a target historical period, and acquiring a second content vector of each historical case;
inputting the first content vector and the second content vector into a trained comprehensive risk recognition model to obtain a target risk level of the to-be-processed case, wherein the comprehensive risk recognition model comprises a first feature extraction model and a second feature extraction model, the first feature extraction model is used for extracting features based on spatial relevance, and the second feature extraction model is used for extracting features based on time relevance;
and carrying out corresponding early warning operation on the to-be-processed case according to the target risk level.
2. The method for early warning of a new type of contradictory dispute according to claim 1, wherein the comprehensive risk recognition model further comprises a classification model connected to the first feature extraction model and the second feature extraction model, the classification model comprises an attention module and an output layer, the first content vector and the second content vector are input into the trained comprehensive risk recognition model to obtain a target risk level of the case to be processed, and the method comprises:
inputting the first content vector into the first feature extraction model to obtain a first feature vector corresponding to the to-be-processed case, and inputting the second content vector into the second feature extraction model to obtain a second feature vector corresponding to the historical case;
inputting the first feature vector and the second feature vector into the attention module to obtain a fusion feature vector;
and inputting the fusion feature vector into the output layer to obtain the target risk level of the case to be processed.
3. The method for early warning of a new type of contradictory dispute according to claim 2, wherein the second feature extraction model comprises a gating circulation unit and a variation self-encoder connected to each other, the inputting the second content vector into the second feature extraction model to obtain a second feature vector corresponding to the historical case, comprises:
Organizing all the second content vectors corresponding to each related object into a time sequence, wherein each time in the time sequence corresponds to a group of the second content vectors;
inputting each time sequence into the gating circulation unit for processing;
in the processing process of the gating circulation unit, a group of second content vectors corresponding to the current time are obtained from the time sequence, and the coding vectors output by the encoder at the last time of the variation and the hiding state vectors output by the gating circulation unit at the last time are obtained;
determining an input vector of the current time of the gating circulation unit according to the coding vector output at the previous time, the hidden state vector output at the previous time and the group of second content vectors of the current time, and inputting the input vector into the gating circulation unit to obtain the hidden state vector output at the current time of the gating circulation unit; and then, inputting the hidden state vector output by the current time of the gating circulating unit into the variation self-encoder to obtain the encoded vector output by the current time of the variation self-encoder, updating the next time to the current time, and returning to execute the step of acquiring a group of second content vectors corresponding to the current time from the time sequence.
4. The method for pre-warning of a new type of contradictory dispute according to claim 3, wherein determining the input vector of the current time of the gating loop unit from the encoded vector output at the previous time, the hidden state vector output at the previous time, and the set of second content vectors of the current time comprises:
splicing the coding vector output at the previous time and the hidden state vector output at the previous time to obtain a spliced vector at the current time;
and taking the spliced vector of the current time and the group of second content vectors of the current time as input vectors of the current time of the gating circulating unit.
5. The method for pre-warning a new type of contradictory dispute according to claim 2, wherein said inputting the first feature vector and the second feature vector into the attention module to obtain a fused feature vector comprises:
mapping, by the attention module, the first feature vector and the second feature vector into a same vector space;
and carrying out weighted summation on the mapped first feature vector and the mapped second feature vector to obtain a fusion feature vector.
6. The method for pre-warning of a new contradictory dispute in any one of claims 1-5 further comprising:
acquiring a first sample content vector of each case sample in the case sample set, a related sample related object and a risk level label;
acquiring a sample history case of each sample related object in a sample history period, and acquiring a second sample content vector of each sample history case;
training the created comprehensive risk identification model by using the first sample content vector, the second sample content vector and the risk level label.
7. The utility model provides a novel early warning system of contradiction dispute case which characterized in that includes:
the first acquisition module is used for acquiring a first content vector of a case to be processed and all related incident objects;
the second acquisition module is used for determining historical cases of each related object in a target historical period and acquiring a second content vector of each historical case;
the recognition module is used for inputting the first content vector and the second content vector into a trained comprehensive risk recognition model to obtain a target risk level of the case to be processed, the comprehensive risk recognition model comprises a first feature extraction model and a second feature extraction model, the first feature extraction model is used for carrying out feature extraction based on spatial relevance, and the second feature extraction model is used for carrying out feature extraction based on temporal relevance;
And the early warning module is used for carrying out corresponding early warning operation on the to-be-processed case according to the target risk level.
8. The novel conflict and dispute case alert system of claim 7, wherein said integrated risk recognition model further comprises a classification model coupled to said first feature extraction model and said second feature extraction model, said classification model comprising an attention module and an output layer, said recognition module being specifically configured to:
inputting the first content vector into the first feature extraction model to obtain a first feature vector corresponding to the to-be-processed case, and inputting the second content vector into the second feature extraction model to obtain a second feature vector corresponding to the historical case;
inputting the first feature vector and the second feature vector into the attention module to obtain a fusion feature vector;
and inputting the fusion feature vector into the output layer to obtain the target risk level of the case to be processed.
9. The novel conflict and dispute early warning system of claim 8, wherein the second feature extraction model comprises a connected gating circulation unit and a variation self-encoder, and wherein the identification module is specifically configured to:
Organizing all the second content vectors corresponding to each related object into a time sequence, wherein each time in the time sequence corresponds to a group of the second content vectors;
inputting each time sequence into the gating circulation unit for processing;
in the processing process of the gating circulation unit, a group of second content vectors corresponding to the current time are obtained from the time sequence, and the coding vectors output by the encoder at the last time of the variation and the hiding state vectors output by the gating circulation unit at the last time are obtained;
determining an input vector of the current time of the gating circulation unit according to the coding vector output at the previous time, the hidden state vector output at the previous time and the group of second content vectors of the current time, and inputting the input vector into the gating circulation unit to obtain the hidden state vector output at the current time of the gating circulation unit; and then, inputting the hidden state vector output by the current time of the gating circulating unit into the variation self-encoder to obtain the encoded vector output by the current time of the variation self-encoder, updating the next time to the current time, and returning to execute the step of acquiring a group of second content vectors corresponding to the current time from the time sequence.
10. The novel conflict and dispute pre-warning system of claim 9, wherein said identification module is specifically configured to:
the determining the input vector of the current time of the gating loop unit according to the coding vector output at the previous time, the hidden state vector output at the previous time and the set of second content vectors of the current time comprises:
splicing the coding vector output at the previous time and the hidden state vector output at the previous time to obtain a spliced vector at the current time;
and taking the spliced vector of the current time and the group of second content vectors of the current time as input vectors of the current time of the gating circulating unit.
11. The novel conflict and dispute pre-warning system of claim 8, wherein the identification module is specifically configured to:
mapping, by the attention module, the first feature vector and the second feature vector into a same vector space;
and carrying out weighted summation on the mapped first feature vector and the mapped second feature vector to obtain a fusion feature vector.
12. The novel pre-warning system for contradictory disputes cases according to any one of claims 7-11, further comprising a training module for:
Acquiring a first sample content vector of each case sample in the case sample set, a related sample related object and a risk level label;
acquiring a sample history case of each sample related object in a sample history period, and acquiring a second sample content vector of each sample history case;
training the created comprehensive risk identification model by using the first sample content vector, the second sample content vector and the risk level label.
13. A non-transitory computer readable storage medium having stored thereon a computer program, which when executed by a processor, implements the method of pre-warning of novel contradictory disputes cases according to any one of claims 1-6.
14. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of pre-warning of a novel contradictory dispute case as claimed in any one of claims 1 to 6 when the program is executed by the processor.
CN202311318564.5A 2023-10-11 2023-10-11 Novel early warning method and system for contradictory dispute cases Pending CN117391466A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311318564.5A CN117391466A (en) 2023-10-11 2023-10-11 Novel early warning method and system for contradictory dispute cases

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311318564.5A CN117391466A (en) 2023-10-11 2023-10-11 Novel early warning method and system for contradictory dispute cases

Publications (1)

Publication Number Publication Date
CN117391466A true CN117391466A (en) 2024-01-12

Family

ID=89469418

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311318564.5A Pending CN117391466A (en) 2023-10-11 2023-10-11 Novel early warning method and system for contradictory dispute cases

Country Status (1)

Country Link
CN (1) CN117391466A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117648980A (en) * 2024-01-29 2024-03-05 数据空间研究院 Novel entity relationship joint extraction algorithm based on contradiction dispute data

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117648980A (en) * 2024-01-29 2024-03-05 数据空间研究院 Novel entity relationship joint extraction algorithm based on contradiction dispute data
CN117648980B (en) * 2024-01-29 2024-04-12 数据空间研究院 Novel entity relationship joint extraction method based on contradiction dispute data

Similar Documents

Publication Publication Date Title
CN109104620B (en) Short video recommendation method and device and readable medium
Alam et al. MEDIC: a multi-task learning dataset for disaster image classification
WO2022105117A1 (en) Method and device for image quality assessment, computer device, and storage medium
CN112016467B (en) Traffic sign recognition model training method, recognition method, system, device and medium
CN111898675B (en) Credit wind control model generation method and device, scoring card generation method, machine readable medium and equipment
CN117391466A (en) Novel early warning method and system for contradictory dispute cases
CN114445143A (en) Service data prediction method, device, equipment and medium
US20220309597A1 (en) Computer vision framework for real estate
CN114241411B (en) Counting model processing method and device based on target detection and computer equipment
Boppana et al. Comparative analysis of single-stage yolo algorithms for vehicle detection under extreme weather conditions
CN117496269A (en) Composite electric energy quality disturbance identification method
CN113239883A (en) Method and device for training classification model, electronic equipment and storage medium
Li et al. SCD-YOLO: a lightweight vehicle target detection method based on improved YOLOv5n
Sun et al. Vehicle classification approach based on the combined texture and shape features with a compressive DL
CN116310382A (en) Public opinion prediction method, public opinion prediction device, electronic equipment and storage medium
CN115757844A (en) Medical image retrieval network training method, application method and electronic equipment
CN115359468A (en) Target website identification method, device, equipment and medium
Sharma et al. Real-time attention-based embedded LSTM for dynamic sign language recognition on edge devices
CN113920720A (en) Highway tunnel equipment fault processing method and device and electronic equipment
CN113239215A (en) Multimedia resource classification method and device, electronic equipment and storage medium
CN112801960A (en) Image processing method and device, storage medium and electronic equipment
Yang et al. Abnormal Interference Recognition Based on Rolling Prediction Average Algorithm
Xin et al. DELFormer: detail-enhanced lightweight transformer for road segmentation
Chen et al. LPR‐MLP: A Novel Health Prediction Model for Transmission Lines in Grid Sensor Networks
CN111178630A (en) Load prediction method and device

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