CN-122019195-A - Legal service flow intelligent management system

CN122019195ACN 122019195 ACN122019195 ACN 122019195ACN-122019195-A

Abstract

The invention relates to the technical field of legal service, in particular to an intelligent management system of legal service flow, which comprises a text analysis module, a text analysis module and a control module, wherein the text analysis module extracts bottom character information to construct a text matrix so as to generate a local abnormal identifier; the circulation judging module calculates delay quantity by combining with the local service tree and constructs a global state vector, the resource scheduling module screens the calculating unit according to the core load characteristics to package the intervention instruction set, and the execution distributing module issues the rechecking task and monitors the bottom layer response flow to construct a closed-loop record. In the invention, a text matrix is accurately constructed by extracting a bottom character rule, data multiple comparison is performed to generate an abnormal intervention mark, delay amount is deeply calculated by combining a service tree and mapped into a state vector, an idle calculation unit is screened according to load characteristics to execute physical sequencing, a command is issued to a terminal, a bottom response stream is monitored to construct a complete closed-loop record, and delay caused by a solidification mechanism is thoroughly eliminated to realize resource self-adaptive scheduling.

Inventors

ZHU ZEQI
YAN BINGRUI
Shao Jinxuan
CHEN FEI

Assignees

南京师范大学

Dates

Publication Date: 20260512
Application Date: 20260414

Claims (10)

1. An intelligent legal service flow management system, comprising: The text analysis module analyzes the bottom character information in the file, constructs a text feature matrix according to the word frequency of the bottom character information, calculates a difference data sequence based on the text feature matrix comparison, and constructs an abnormal intervention mark according to the difference data sequence; The circulation judging module is used for receiving the abnormal intervention identification, reading a service association topology of a memory, analyzing front node parameters bound by the abnormal intervention identification by combining the service association topology, calculating circulation delay quantity by combining the front node parameters, and constructing a global state vector based on the circulation delay quantity; The resource scheduling module extracts core load characteristics in the global state vector, screens a processing computing unit according to the core load characteristics and the idle task queues of the hardware pool, acquires a physical address sequence of the processing computing unit, and encapsulates the physical address sequence to construct a scheduling intervention instruction; And the execution distribution module analyzes the scheduling intervention instruction, issues a business rechecking task to the interaction terminal according to the scheduling intervention instruction, records the distribution time of the business rechecking task, extracts the final task state of the interaction terminal, and aggregates the final task state and the distribution time to construct a business closed-loop record.
2. The legal service flow intelligent management system of claim 1, wherein the text parsing module specifically comprises: The character reading sub-module is used for acquiring the bottom character information in the file, carrying out word segmentation and part-of-speech tagging on the bottom character information in combination with a natural language processing algorithm, eliminating deactivated vocabulary in the bottom character information, reserving service vocabulary, counting the occurrence times of the service vocabulary in the file, and generating word frequency distribution data; The word frequency comparison sub-module is used for obtaining the word frequency distribution data, mapping the word frequency distribution data to a multidimensional vector space, distributing weight coefficients according to coordinate axis attributes in the multidimensional vector space, classifying and arranging the weighted word frequency distribution data according to service attributes, and constructing the text feature matrix; The abnormal boundary sub-module is used for extracting the text feature matrix, calculating Euclidean distance between the text feature matrix and the standard matrix template in corresponding dimension, extracting difference values deviating from a safety interval according to fluctuation amplitude of the Euclidean distance, aggregating the difference values to calculate a difference data sequence, comparing the difference data sequence with an extremum boundary, and constructing the abnormal intervention mark when the values in the difference data sequence break through the extremum boundary; The extremum boundary refers to a numerical threshold for judging whether the characteristic deviation reaches the system abnormal intervention.
3. The legal service flow intelligent management system of claim 1, wherein the circulation decision module specifically comprises: The topology analysis sub-module receives the abnormal intervention identification, reads the service association topology in the memory, carries out reverse addressing in the service association topology according to the abnormal intervention identification, positions a service level triggering abnormality, and extracts the preposed node parameters corresponding to the service level; The delay calculation sub-module is used for receiving the pre-node parameters, extracting standard approval period and time-consuming data in the pre-node parameters, calculating a time difference value between the time-consuming data and the standard approval period, carrying out weighted amplification operation on the time difference value by combining with an emergency degree coefficient, and calculating the circulation delay amount; And the state vector sub-module is used for acquiring the flow delay quantity, performing tensor splicing on the flow delay quantity and the concurrent connection number, generating a numerical array in a multidimensional space, and constructing the global state vector according to the numerical array.
4. The legal service flow intelligent management system of claim 1, wherein the resource scheduling module specifically comprises: The load characteristic submodule extracts the global state vector, performs dimension reduction processing on the global state vector, extracts key values in a dimension reduction result, and extracts the core load characteristic in the global state vector according to the key values; The unit screening submodule receives the core load characteristics, reads the idle task queue in the hardware pool, compares the resource demand corresponding to the core load characteristics with the residual capacity of each node in the idle task queue, locks processing nodes of matching conditions in the hardware pool, and screens the processing calculation unit; And the instruction encapsulation submodule is used for positioning the processing calculation unit, reading the hardware communication port and the network card identifier of the processing calculation unit in the local area network, assembling the hardware communication port and the network card identifier according to a network protocol, acquiring the physical address sequence of the processing calculation unit, binary packaging the physical address sequence and a scheduling control code, and constructing the scheduling intervention instruction.
5. The legal service flow intelligent management system of claim 1, wherein the execution allocation module specifically comprises: The task issuing sub-module receives and analyzes the scheduling intervention instruction, extracts a communication protocol according to the scheduling intervention instruction, establishes an encrypted data transmission channel between the scheduling intervention instruction and the interactive terminal, issues the business review task to the interactive terminal through the encrypted data transmission channel, and reads clock frequency to record the distribution time of the business review task; The state monitoring sub-module is used for keeping the starting state of the encrypted data transmission channel after recording the distribution time, intercepting the digital heartbeat packet returned by the interactive terminal, decoding and mapping binary state codes in the digital heartbeat packet, and extracting the final task state of the interactive terminal; And the closed-loop aggregation sub-module acquires the final task state and the distribution time, converts the distribution time into a standard timestamp format, performs data alignment and binding on the standard timestamp format and the final task state, writes the bound data body into a memory, and constructs the service closed-loop record.
6. The legal service flow intelligent management system of claim 2, wherein the process of calculating the differential data sequence specifically comprises: acquiring the text feature matrix and a standard matrix template, scanning service feature dimensions in the text feature matrix line by line, and extracting real-time word frequency scalar under the service feature dimensions; synchronously extracting historical mean scalar corresponding to the business feature dimension in the standard matrix template, inputting the real-time word frequency scalar and the historical mean scalar into a differential operation logic, and calculating an absolute deviation value of the business feature dimension; Acquiring sensitivity penalty coefficients of service feature dimensions, and multiplying, weighting and amplifying corresponding absolute deviation values by using the sensitivity penalty coefficients to generate a discrete deviation set; And queuing the discrete deviation sets according to the time occurrence sequence, and performing smooth filtering processing on the queued discrete deviation sets by using a sliding window algorithm to calculate the difference data sequence.
7. The legal service flow intelligent management system of claim 3, wherein the process of calculating the flow delay amount specifically comprises: Acquiring the pre-node parameters, extracting time-consuming data, a standard approval period and an emergency degree coefficient from the pre-node parameters, judging whether the time-consuming data is larger than the standard approval period, and triggering a delay penalty mechanism if the time-consuming data is larger than the standard approval period; under a delay punishment mechanism, calculating a logarithmic difference value converted based on a time difference value between time-consuming data and a standard approval period, and carrying out cross multiplication operation on the logarithmic difference value and an emergency degree coefficient to obtain a network congestion index; calculating the circulation delay amount according to a delay quantization model by combining a network congestion index, a logarithmic difference value and an emergency degree coefficient, wherein a specific mathematical expression formula of the delay quantization model is as follows: ; Wherein, the Representing the amount of delay in the flow of the stream, Representing the coefficient of degree of urgency, Representing time-consuming data, Representing a standard period of approval, Representing the reference time constant of the time of day, Representing the ambient compensation constant and, Representing the network congestion index.
8. The legal service flow intelligent management system of claim 4, wherein the process of constructing the scheduling intervention instruction specifically comprises: the physical address sequence of the processing and computing unit is obtained, hash operation is carried out on the physical address sequence, and an addressing head is generated; the method comprises the steps of obtaining a scheduling control code, carrying out syntax tree analysis on the scheduling control code, removing comments and blank lines in the scheduling control code, and converting the simplified scheduling control code into a binary machine instruction stream; Generating a dynamic check code, splicing the addressing head, the machine instruction stream and the dynamic check code according to the data bus bit width standard, adding a network terminator at the tail end of the spliced data packet, and constructing the scheduling intervention instruction.
9. The legal service flow intelligent management system according to claim 5, wherein the process of extracting the final task state of the interactive terminal specifically comprises: after a digital heartbeat packet returned by the interactive terminal is obtained, reading message header information of the digital heartbeat packet, and performing cyclic redundancy check on the message header information; Stripping message header information of the digital heartbeat packet, extracting effective load fields in the digital heartbeat packet, reading a state mapping dictionary, and searching plaintext state interpretation corresponding to the effective load fields in the state mapping dictionary; judging whether the plaintext state interpretation contains abnormal interrupt vocabulary or not, if not, marking the plaintext state interpretation as a normal ending mark, and outputting the normal ending mark as the final task state.
10. The legal service flow intelligent management system of claim 4, wherein the process of screening the processing computing unit specifically comprises: acquiring the resource demand corresponding to the core load characteristics and the residual capacity of each node in the idle task queue, and reading the fault occurrence frequency of each node in the historical operation record to acquire a stability damage variable; Multiplying the residual capacity by the stability impairment variable, calculating the expected value of the available computing force of each node, and calculating the ratio of the expected value of the available computing force to the resource demand to obtain the index of computing force adequacy; according to a screening matching formula, the comprehensive optimization score of each node is calculated by combining the calculation power adequacy index and the physical communication distance of each node, the node with the highest comprehensive optimization score is screened as the processing calculation unit, and the screening matching formula is as follows: ; Wherein, the Representing the overall preference score of the vehicle, Representing the expected value of the available computing force, Representing the amount of resource required, Representing the decay constant of the distance, Representing the physical communication distance.

Description

Legal service flow intelligent management system Technical Field The invention relates to the technical field of legal service, in particular to an intelligent management system for legal service flow. Background The technical field of legal service relates to the overall planning and tracking of massive volume data and manual approval nodes, and the overall planning and tracking comprises authority circulation, state monitoring and a dynamic configuration network of business resources. The traditional legal service flow intelligent management system refers to a record structure for changing states by relying on a solidified relational database form, a unidirectional batch processing program is adopted to transfer text information into a static ledger when a file is recorded, meanwhile, a mail gateway triggered mechanically is used for sending instructions in a service handover link, the node state is checked by relying on manual periodic polling instructions, and each independent station is provided with a local area network storage unit isolated from each other. The traditional business architecture cannot synchronously feed back potential bottom blocking nodes when facing high-frequency concurrent case circulation change requests, and is excessively dependent on manual execution of subjective state verification operation among multi-dimensional isolated accounts, so that data instruction coverage is extremely easy to generate and history handover trace is thoroughly lost when the purely mechanical physical form is overlapped, a bottom isolated storage unit cannot provide panoramic state view when cross-department business handover is carried out, a gateway solidified communication triggering mechanism cannot dynamically interfere hardware pool parameters when facing sudden abnormality, and a lagged node refreshing mechanism severely delays a physical resource allocation period of a cross-node. Disclosure of Invention The invention aims to solve the defects in the prior art, and provides an intelligent legal service flow management system. In order to achieve the above purpose, the present invention adopts the following technical scheme, and an intelligent legal service flow management system comprises: The text analysis module analyzes the bottom character information in the file, constructs a text feature matrix according to the word frequency of the bottom character information, calculates a difference data sequence based on the text feature matrix comparison, and constructs an abnormal intervention mark according to the difference data sequence; The circulation judging module is used for receiving the abnormal intervention identification, reading a service association topology of a memory, analyzing front node parameters bound by the abnormal intervention identification by combining the service association topology, calculating circulation delay quantity by combining the front node parameters, and constructing a global state vector based on the circulation delay quantity; The resource scheduling module extracts core load characteristics in the global state vector, screens a processing computing unit according to the core load characteristics and the idle task queues of the hardware pool, acquires a physical address sequence of the processing computing unit, and encapsulates the physical address sequence to construct a scheduling intervention instruction; And the execution distribution module analyzes the scheduling intervention instruction, issues a business rechecking task to the interaction terminal according to the scheduling intervention instruction, records the distribution time of the business rechecking task, extracts the final task state of the interaction terminal, and aggregates the final task state and the distribution time to construct a business closed-loop record. As a further aspect of the present invention, the text parsing module specifically includes: The character reading sub-module is used for acquiring the bottom character information in the file, carrying out word segmentation and part-of-speech tagging on the bottom character information in combination with a natural language processing algorithm, eliminating deactivated vocabulary in the bottom character information, reserving service vocabulary, counting the occurrence times of the service vocabulary in the file, and generating word frequency distribution data; The word frequency comparison sub-module is used for obtaining the word frequency distribution data, mapping the word frequency distribution data to a multidimensional vector space, distributing weight coefficients according to coordinate axis attributes in the multidimensional vector space, classifying and arranging the weighted word frequency distribution data according to service attributes, and constructing the text feature matrix; The abnormal boundary sub-module is used for extracting the text feature matrix, calculating Euclidean distance between the text feature