An intelligent case source allocation method and system based on online judicial mediation

Abstract

The application provides an intelligent case source distribution method and system based on online judicial mediation, relates to the field of intelligent legal technology, and comprises the following steps: extracting case characteristic labels and party background labels respectively; constructing an evidence graph by analyzing the relationship between evidences, extracting evidence breakpoint labels; predicting the probability of new disputes and the field of new disputes; obtaining mediator information; calculating the matching degree of each mediator based on the case characteristic labels, the probability of disputes, the field of new disputes and the mediator information, sorting the matching degree to generate a candidate list; and generating push distribution information according to the probability of disputes. The application integrates three types of labels, i.e. case characteristics, party backgrounds and evidence breakpoints, predicts the probability and field of new disputes, realizes deep adaptation of case demand and mediator capacity, and avoids resource mismatch. By introducing a workload simulation mechanism, the application avoids excessive concentration of high-quality resources or workload imbalance, ensures mediation efficiency and quality, and improves the overall case source turnover speed.

Description

Technical Field

[0001] This invention relates to the field of intelligent legal technology, specifically to an intelligent case source allocation method and system based on online judicial mediation. Background Technology

[0002] Online judicial mediation, centered on digital transformation and improved judicial efficiency, integrates the needs of judicial practice with technologies from multiple fields, forming a technical support system covering data processing, intelligent decision-making, and collaborative management. Among these, the intelligent allocation of cases is a key aspect of this transformation. Assigning mediators with extensive relevant experience based on the case details effectively improves case turnover efficiency and increases the likelihood of successful mediation. However, when the parties' demands change during online judicial mediation, it can trigger a chain reaction. If new points of contention involve specialized fields, such as equity division requiring "financial legal expertise," the originally assigned mediator may lack the necessary skills, and their workload may have changed. In such cases, reassigning mediators requires the new mediators to familiarize themselves with the case, resulting in a waste of human resources, and the parties may become dissatisfied with the change. Therefore, how to select suitable mediators for cases from the outset is a problem that needs optimization. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent case allocation method and system based on online judicial mediation to improve the aforementioned problems. To achieve the above objective, the technical solution adopted by this invention is as follows:

[0004] Firstly, this application provides an intelligent case allocation method based on online judicial mediation, including:

[0005] Obtain background information of the parties involved and case text information, and extract case feature tags and party background tags respectively;

[0006] By analyzing the relationships between pieces of evidence, an evidence map is constructed, and evidence breakpoint labels are extracted based on the evidence map.

[0007] Predict the probability of new disputes and the areas of new disputes based on case feature tags, party background tags, and evidence breakpoint tags;

[0008] Obtain mediator information, which includes historical mediation success rate, area of ​​expertise, and number of complex cases handled;

[0009] The matching degree of each mediator is calculated based on case feature tags, dispute probability, newly added dispute areas and mediator information, and a candidate list is generated by sorting the matching degree.

[0010] If the probability of a dispute is less than a preset threshold, push assignment information is generated based on the matching degree; otherwise, a preset number of mediators are selected from the candidate list, and the workload of the selected mediators is simulated for a preset time after matching according to the situation of new disputes; push assignment information is generated based on the matching degree and workload.

[0011] Secondly, this application also provides an intelligent case allocation system based on online judicial mediation, including:

[0012] The first module is used to obtain background information of the parties involved and case text information, and to extract case feature tags and party background tags respectively.

[0013] The second module is used to construct an evidence map by analyzing the relationships between evidence, and to extract evidence breakpoint labels based on the evidence map;

[0014] The third module is used to predict the probability of new disputes and the areas of new disputes based on case feature tags, party background tags, and evidence breakpoint tags.

[0015] The fourth module is used to obtain mediator information, which includes historical mediation success rate, professional field, and number of complex cases handled.

[0016] The fifth module is used to calculate the matching degree of each mediator based on case feature tags, dispute probability, newly added dispute areas and mediator information, and generate a candidate list by sorting the matching degree.

[0017] The sixth module is used to determine whether the probability of a dispute is less than a preset threshold. If so, push allocation information is generated based on the matching degree; if not, a preset number of mediators are selected from the candidate list, and the workload of the selected mediators is simulated for a preset time after matching according to the situation of new disputes. Push allocation information is generated based on the matching degree and workload.

[0018] The beneficial effects of this invention are as follows:

[0019] This invention integrates three types of tags—case characteristics, party background, and evidence breakpoints—to predict the probability and scope of new disputes, identify potential case risks in advance, and achieve a deep match between case needs and mediator capabilities, thus avoiding resource misallocation. For cases with a high probability of dispute, a workload simulation mechanism is introduced to predict the future capacity of mediators, preventing excessive concentration of high-quality resources or workload imbalance, ensuring mediation efficiency and quality, and improving the overall case flow speed.

[0020] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart illustrating the intelligent case allocation method based on online judicial mediation, as described in this application.

[0023] Figure 2 This is a diagram of an intelligent case allocation device based on online judicial mediation, as described in an embodiment of this application.

[0024] Symbol explanation: 800 - Intelligent case source allocation device based on online judicial mediation; 801 - Processor; 802 - Memory; 803 - Multimedia component; 804 - I / O interface; 805 - Communication component. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0026] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0027] When the parties' demands are adjusted during online judicial mediation, it can trigger a chain reaction, for example:

[0028] Family disputes (divorce, inheritance, etc.): The core conflict is prone to spread from a single focus. For example, if the initial claim is "custody", the mediation may add claims such as "adjustment of child support amount", "division of hidden property", and "details of visitation rights". Especially in family cases involving complex property (such as equity, real estate, cross-border assets), the probability of adding new claims is relatively high.

[0029] Commercial disputes (sales, service contracts, etc.): New claims often arise due to supplementary evidence or disagreements on liability determination. For example, in cross-regional commercial cooperation disputes, the initial claim may be "payment of contract price," but during mediation, multiple new claims may be added due to issues such as branch company liability, third-party guarantees, and cargo transportation losses.

[0030] Labor disputes: The initial claims are mostly "wage payment", but during mediation, they often extend to claims such as "overtime pay calculation", "social security arrears", "economic compensation", and "removal of non-compete restrictions".

[0031] It is evident that the emergence of new demands during online judicial mediation is not accidental, especially in complex disputes where it is a "high-probability event," requiring further strengthening of adaptability for high-probability scenarios.

[0032] Example 1:

[0033] See Figure 1 This embodiment provides an intelligent case source allocation method based on online judicial mediation, including steps S100, S200, S300, S400, S500 and S600.

[0034] S100. Obtain background information of the parties involved and case text information, and extract case feature tags and party background tags respectively;

[0035] Case text information includes indictments, evidence materials, and may also include transcripts of voice calls. Using Natural Language Processing (NLP) technology, based on the BERT model, the case text is automatically parsed to generate case feature tags, specifically including:

[0036] Case attribute tags: Dispute type (e.g., "Family dispute - complex property division", "Financial loan contract - overdue repayment"), Dispute amount range (<100,000 / 100,000 - 500,000 / >500,000);

[0037] Processing needs tags: cross-regional mediation required (based on the parties' addresses), professional assessment required (such as "engineering cost assessment"), emotional risk level (detecting anger / depression tendencies in the conversation through voice recognition technology);

[0038] Judicial procedure tags: whether it enters pre-litigation mediation, whether it needs to be synchronized with the court's case filing system, and the method of performance guarantee (guarantee / mortgage / smart contract).

[0039] Background information of the parties involved includes the age, gender, occupation, education level, geographical distribution, family structure, asset status (real estate, vehicles, stocks, etc.), debt situation, credit record, history of involvement in previous disputes, relationship with the other party, and clarity of expression of demands of both parties;

[0040] Special background: ethnic minority status (language adaptation requirements), identification of vulnerable groups such as disabled persons, whether public officials are involved, etc.

[0041] S200. By analyzing the relationships between pieces of evidence, an evidence map is constructed, and evidence breakpoint labels are extracted based on the evidence map; specifically including:

[0042] S210. Obtain textual and non-textual evidence, and convert non-textual evidence into text format to form a text summary;

[0043] This includes contracts, IOUs, transfer records, chat logs, witness testimonies, expert reports, administrative documents, etc.; non-text evidence such as chat log screenshots and audio recordings need to be converted into text format for preservation.

[0044] S220. Generate a unique identifier for each piece of evidence and extract the evidence type, creation time, purpose of proof, and key information;

[0045] Evidence types include six major categories: text, physical evidence, audio-visual materials, electronic evidence, witness testimony, and expert / professional opinions;

[0046] S230. Analyze the evidence text and establish citation relationships by extracting evidence-citation evidence association information;

[0047] For example, if evidence A directly mentions evidence B, such as the loan amount stipulated in the contract referencing the actual amount in the transfer record, or a witness testimony directly mentioning a certain piece of physical evidence, it is marked as "reference (A→B)", thus establishing the reference relationship between evidence A and evidence B;

[0048] S240. Establish supporting relationships by calculating the similarity between the purposes of proof;

[0049] If the similarity between the purposes of proof is higher than the preset evaluation threshold, it means that the two pieces of evidence jointly prove the same fact (such as both the IOU and the chat records proving the agreement to lend money), thus establishing a supporting relationship between the two pieces of evidence.

[0050] S250. By comparing the formation time and key information, establish a temporal relationship for evidence related to the same fact;

[0051] Evidence A was created earlier than Evidence B, and key information was used to identify that both pieces of evidence are related to the same fact (e.g., the contract was signed earlier than the transfer time, and both are related to the same contract), so they are marked as having a chronological relationship.

[0052] S260. Generate an evidence map based on the citation relationships, supporting relationships, and temporal relationships among the evidence;

[0053] Each piece of evidence is treated as a node, and related nodes are connected. Edge attributes are generated based on the relationship type. There may be multiple edges with different attributes between two nodes.

[0054] The generated evidence map is cleaned and optimized, and duplicate relationships are merged (such as multiple "support" relationships between a pair of nodes, only one is retained).

[0055] Obtain a standard chain of evidence based on the case type, search for evidence based on the standard chain of evidence, and record logical breakpoints based on the search results.

[0056] The standard chain of evidence is formulated in conjunction with the requirements for the integrity of the chain of evidence in judicial mediation. For example, for the core facts of a loan dispute, the standard chain of evidence is: loan agreement + loan delivery + repayment commitment.

[0057] Traverse the evidence graph according to the standard chain of evidence to determine whether key evidence is missing. If any is missing, record the logical breakpoint.

[0058] Key information in evidence with citation relationships is compared, and semantic recognition algorithms are used to identify content contradictions or inconsistencies, and citation breakpoints are recorded.

[0059] For example, if evidence A explicitly cites evidence B, such as evidence A citing "loan amount of 1 million", while evidence B actually states the amount as 500,000, this indicates a contradiction in the content.

[0060] Query the supporting relationships of evidence nodes. If there are no related supporting nodes, record the supporting breakpoint. Or, for a certain piece of evidence, if it has related supporting nodes, but its integrity is questionable, such as if it is a single witness testimony and the key information of the testimony is vague, it can also be identified as a supporting breakpoint.

[0061] The system counts the number of breakpoints for different types, calculates the breakpoint density, and generates breakpoint type and severity labels.

[0062] Based on the breakpoint type, generate a single tag or a combination of tags (such as reference breakpoint + time sequence breakpoint); breakpoint density = total number of breakpoints / total number of evidence nodes, reflecting the severity of the broken evidence chain.

[0063] S300: Predict the probability of new disputes and the areas of new disputes based on case feature tags, party background tags, and evidence breakpoint tags;

[0064] S310. Perform feature-weighted fusion on case feature labels, party background labels, and evidence breakpoint labels to generate a global fusion vector;

[0065] Discrete classification labels are encoded using one-hot encoding, while ordered labels are mapped using numerical values ​​(e.g., low emotional risk = 1, medium = 2, high = 3, extremely high = 4). For continuous data such as dispute amount and breakpoint density, Min-Max standardization is used to map them to the [0,1] interval to avoid the impact of dimensional differences on model performance. Correlation analysis is performed based on historical data, and core labels are given higher weights (e.g., dispute type weight 0.25, breakpoint density weight 0.2, asset status weight 0.15), while the weights of other auxiliary labels are set to 0.05-0.1.

[0066] S320. Input the global fusion vector into the pre-trained classification model to obtain the predicted new dispute domain, wherein there is at least one predicted new dispute domain;

[0067] Because disputes occur in different fields in different ways, it is necessary to predict the probability of disputes based on newly added disputed fields. This method uses cascaded prediction, first predicting newly added disputed fields, and then using another model to predict the probability of disputes.

[0068] The classification model employs XGBoost, using a large amount of historical case data (all cases with newly emerging disputes) and their labeled "newly emerging dispute domains" (such as labor disputes, equity distribution, etc.) for supervised training. This results in a multi-classification model that outputs the most probable categories of newly emerging dispute domains, as well as a probability distribution showing the likelihood of the case belonging to each preset domain. However, this probability represents the probability of the dispute belonging to a certain domain given that a new dispute is certain to occur, not the probability of the new dispute actually occurring.

[0069] S330. Combine the global fusion vector and each predicted new dispute domain separately, and input the combination results into the pre-trained probability prediction model to obtain the new dispute probability corresponding to each new dispute domain.

[0070] The newly added disputed areas are used as new features and fused with the global vector of the training set in step S320 to construct an enhanced feature vector. In addition, cases without newly added disputes are added to the training set. A regression model (outputting continuous probability values ​​of 0-1) or a binary classification model (outputting the probability of a dispute) is trained. A logistic regression model or a neural network model can be selected to obtain a probability prediction model.

[0071] Specifically, a general probability prediction model can be trained using data from the entire domain; for core areas with a sufficient number of cases (such as labor disputes and traffic accidents), specialized dispute probability prediction models can be trained using data from that area, thereby more accurately capturing the risk signals unique to that area.

[0072] S340. Perform threshold filtering based on the probability of disputes to obtain at least one target area of ​​newly added disputes and the corresponding probability of newly added disputes.

[0073] This step sets the probability of dispute to be between 0 and 1; sets 0.7 as the first threshold, selects areas with a probability of dispute greater than or equal to the first threshold as the predicted new dispute areas; if the probability of dispute is less than the first threshold, selects the area with the highest probability of dispute as the new dispute area.

[0074] S400. Obtain mediator information, which includes historical mediation success rate, professional field, and number of complex cases handled.

[0075] S500: Calculate the matching degree of each mediator based on case feature tags, dispute probability, newly added dispute areas, and mediator information, and generate a candidate list by sorting the matching degree.

[0076] S510. Based on case feature tags and newly added disputed areas, calculate the similarity between the case area and the mediator's professional area to obtain the area matching degree. The weights of case feature tags and newly added disputed areas are adjusted according to the probability of the dispute; the lower the probability of the newly added dispute, the lower its weight.

[0077] S520. Calculate the tendency weights of historical mediation success rate and complex case handling volume based on the probability of disputes. Calculate the capability matching degree by weighting the historical mediation success rate, complex case handling volume and tendency weights.

[0078] A higher probability of dispute reflects a more complex case, such as involving more people and other departments and institutions; therefore, the weight given to handling complex cases is increased, and mediators with more experience in handling complex cases are preferred; while a lower probability of dispute tends to be chosen, and mediators with a higher historical success rate in mediation are preferred.

[0079] S530. Calculate the similarity between the case feature tags and the mediator's historical successful cases to obtain the experience matching degree;

[0080] The similarity between the case feature tags and the case feature tags of each mediator's historical successful cases is calculated. Historical successful cases with a similarity greater than a preset threshold (such as 0.75) are selected. The total number of selected cases and the average similarity are then counted. The experience matching degree is calculated by weighting the total number of cases and the average similarity.

[0081] S540. The matching degree of the mediator is obtained by weighted summation of the domain matching degree, ability matching degree and experience matching degree.

[0082] S550. Select mediators whose matching degree is greater than the minimum requirement value, sort them from largest to smallest matching degree, and generate a candidate list.

[0083] This step requires selecting at least one mediator; if no mediator meets the matching requirements, the one with the highest match score will be selected.

[0084] In other implementations, all mediators can be sorted by matching degree first, and then the top few can be selected by a preset number to generate a candidate list.

[0085] S600. Determine if the probability of a dispute is less than a preset threshold. If yes, generate push allocation information based on the matching degree. If no, select a preset number of mediators from the candidate list and simulate the workload of the selected mediators within a preset time after matching according to the situation of new disputes. Generate push allocation information based on the matching degree and workload.

[0086] For example, when the probability of dispute is less than or equal to 0.35, there is no need to consider the occurrence of new disputes; the case is directly sent to several mediators with high matching rates, and the mediators then choose whether to accept the case. When the probability of dispute is greater than 0.35, it is necessary to predict the mediators' workload based on the occurrence of new disputes, including:

[0087] S610. Obtain the mediator's task list, find historical similar cases based on the case feature tags of each case, and calculate the average processing time and effort required for historical similar cases.

[0088] Calculate the similarity between the feature tags of the current case and the feature tags of historical cases, select historical similar cases with a similarity greater than 0.7, and calculate the average processing time of these cases; the effort requirement value is calculated by weighting key factors such as the amount of evidence, the level of emotional risk, the number of other departments involved, and the number of parties involved.

[0089] S620. Based on the processing status of the task list, calculate the remaining estimated processing time for each case;

[0090] The remaining estimated processing time is obtained by subtracting the processing time already taken from the historical average processing time of each case in the task list.

[0091] S630. Insert the cases to be assigned into the task list according to the case priority to generate a simulated pending queue; if there is no priority specification, simply insert the cases at the end of the queue.

[0092] S640. Based on the average processing time and energy demand, simulate time consumption and energy consumption according to the order of cases in the simulated pending queue until the preset time is used up or all cases are processed.

[0093] Time consumption simulation involves deducting the mediator's standard working hours (per day) from the remaining estimated processing time of each case according to the case sequence, and extrapolating the time. For example, if the remaining estimated processing time for a case is 0 after 2.5 days (standard working hours are 8 hours per day), the simulation for the next case begins. The simulation extrapolates the case processing stage based on the current processing time (based on historical cases). For example, if a case enters the professional assessment stage after 10 hours of processing, it is necessary to wait for the assessment results before continuing processing. The simulation determines whether the case should be suspended based on the current stage, moving suspended cases to the end of the queue or inserting them into a suitable position as needed. The simulation then proceeds to the next case, continuing until the preset time (e.g., 15 days) is exhausted or all cases are processed.

[0094] S650. After the simulation ends, calculate the total remaining time of the queue and the total energy consumption, and calculate the workload based on the total remaining time of the queue and the total energy consumption.

[0095] The longer the total remaining time in the queue or the greater the total energy consumption, the greater the workload. When the workload exceeds the preset maximum, the mediator is removed from the candidate list. Cases are then pushed to the mediator according to the candidate list after the removal process.

[0096] Example 2:

[0097] This embodiment provides an intelligent case allocation system based on online judicial mediation, including:

[0098] The first module is used to obtain background information of the parties involved and case text information, and to extract case feature tags and party background tags respectively.

[0099] The second module is used to construct an evidence map by analyzing the relationships between evidence, and to extract evidence breakpoint labels based on the evidence map;

[0100] The third module is used to predict the probability of new disputes and the areas of new disputes based on case feature tags, party background tags, and evidence breakpoint tags.

[0101] The fourth module is used to obtain mediator information, which includes historical mediation success rate, professional field, and number of complex cases handled.

[0102] The fifth module is used to calculate the matching degree of each mediator based on case feature tags, dispute probability, newly added dispute areas and mediator information, and generate a candidate list by sorting the matching degree.

[0103] The sixth module is used to determine whether the probability of a dispute is less than a preset threshold. If so, push allocation information is generated based on the matching degree; if not, a preset number of mediators are selected from the candidate list, and the workload of the selected mediators is simulated for a preset time after matching according to the situation of new disputes. Push allocation information is generated based on the matching degree and workload.

[0104] As an optional implementation, the second module includes:

[0105] The first unit is used to obtain textual and non-textual evidence, and to convert non-textual evidence into text format to form a text summary;

[0106] The second unit is used to generate a unique identifier for each piece of evidence and extract the evidence type, time of its formation, purpose of proof, and key information.

[0107] The third unit is used to parse the evidence text and establish citation relationships by extracting information related to the evidence and the cited evidence.

[0108] The fourth unit is used to establish supporting relationships by calculating the similarity between the purposes of proof;

[0109] The fifth unit is used to establish a chronological relationship between evidence related to the same fact by comparing the time of formation and key information;

[0110] The sixth unit is used to generate an evidence map based on the citation relationships, supporting relationships, and chronological relationships between evidence.

[0111] As an optional implementation, the second module further includes:

[0112] The seventh unit is used to obtain a standard chain of evidence based on the case type, search for evidence based on the standard chain of evidence, and record logical breakpoints based on the search results.

[0113] The eighth unit is used to compare key information of evidence with citation relationships, identify content contradictions or inconsistencies through semantic recognition algorithms, and record citation breakpoints.

[0114] Unit 9 is used to query the supporting relationships of evidence nodes. If there are no related supporting nodes, the supporting breakpoint is recorded.

[0115] Unit 10 is used to count the number of breakpoints of different types, calculate the breakpoint density, and generate breakpoint type labels and severity labels.

[0116] As an optional implementation, the third module includes:

[0117] Unit 11 is used to perform feature weighted fusion of case feature tags, party background tags, and evidence breakpoint tags to generate a global fusion vector;

[0118] The twelfth unit is used to input the global fusion vector into a pre-trained classification model to obtain a predicted new dispute domain, wherein there is at least one predicted new dispute domain;

[0119] The thirteenth unit is used to combine the global fusion vector and each predicted new dispute domain separately, and input the combination results into the pre-trained probability prediction model to obtain the new dispute probability corresponding to each new dispute domain.

[0120] The fourteenth unit is used to perform threshold filtering based on the probability of disputes, to obtain at least one target area of ​​newly added disputes and the corresponding probability of newly added disputes.

[0121] Example 3:

[0122] Corresponding to the above method embodiments, this embodiment also provides an intelligent case source allocation device based on online judicial mediation. The intelligent case source allocation device based on online judicial mediation described below can be referred to in correspondence with the intelligent case source allocation method based on online judicial mediation described above.

[0123] Figure 2 This is a block diagram illustrating an intelligent case allocation device 800 based on online judicial mediation, according to an exemplary embodiment. Figure 2As shown, the intelligent case source allocation device 800 based on online judicial mediation includes a processor 801 and a memory 802. The intelligent case source allocation device 800 may also include one or more of the following: a multimedia component 803, an input / output (I / O) interface 804, and a communication component 805. The processor 801 controls the overall operation of the intelligent case source allocation device 800 to complete all or part of the steps in the aforementioned intelligent case source allocation method based on online judicial mediation. The memory 802 stores various types of data to support the operation of the intelligent case source allocation device 800. This data may include, for example, commands for any application or method operating on the intelligent case source allocation device 800, and application-related data such as contact data, sent and received messages, images, audio, video, etc. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0124] Multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touchscreen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals.

[0125] The received audio signals can be further stored in memory 802 or transmitted via communication component 805. The audio component also includes at least one speaker for outputting audio signals. I / O interface 804 provides an interface between processor 801 and other interface modules, such as a keyboard, mouse, buttons, etc. These buttons can be virtual or physical. Communication component 805 is used for wired or wireless communication between the online judicial mediation-based intelligent case allocation device 800 and other devices. Wireless communication includes, for example, Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or a combination thereof; therefore, the corresponding communication component 805 may include a Wi-Fi module, a Bluetooth module, or an NFC module.

[0126] Example 4:

[0127] Corresponding to the above embodiment of the intelligent case source allocation method based on online judicial mediation, this embodiment also provides a readable storage medium. The readable storage medium described below can be referred to in correspondence with the intelligent case source allocation method based on online judicial mediation described above.

[0128] A readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described embodiment of the intelligent case source allocation method based on online judicial mediation.

[0129] The readable storage medium can specifically be a USB flash drive, external hard drive, read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk, or any other readable storage medium capable of storing program code.

[0130] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0131] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for intelligent case allocation based on online judicial mediation, characterized in that, include: Obtain background information of the parties involved and case text information, and extract case feature tags and party background tags respectively; By analyzing the relationships between pieces of evidence, an evidence map is constructed, and evidence breakpoint labels are extracted based on the evidence map. Predict the probability of new disputes and the areas of new disputes based on case feature tags, party background tags, and evidence breakpoint tags; Obtain mediator information, which includes historical mediation success rate, area of ​​expertise, and number of complex cases handled; The matching degree of each mediator is calculated based on case feature tags, dispute probability, newly added dispute areas and mediator information, and a candidate list is generated by sorting the matching degree. Determine if the probability of dispute is less than a preset threshold; if so, generate push allocation information based on the matching degree. If not, select a preset number of mediators from the candidate list, simulate the workload of the selected mediators within a preset time after matching according to the new disputes, and generate push allocation information based on the matching degree and workload.

2. The intelligent case allocation method based on online judicial mediation according to claim 1, characterized in that, By analyzing the relationships between pieces of evidence, an evidence map is constructed, including: Obtain both textual and non-textual evidence, and convert the non-textual evidence into text format to form a text summary; Generate a unique identifier for each piece of evidence and extract the evidence type, creation time, purpose of proof, and key information; The evidence text is parsed, and citation relationships are established by extracting information linking the evidence and the cited evidence. By calculating the similarity between the purposes of proof, a supporting relationship is established; By comparing the time of formation and key information, a temporal relationship is established for evidence related to the same fact; An evidence map is generated based on the citation relationships, supporting relationships, and chronological relationships among the evidence.

3. The intelligent case source allocation method based on online judicial mediation according to claim 2, characterized in that, Extract evidence breakpoint labels from the evidence map, including: Obtain a standard chain of evidence based on the case type, search for evidence based on the standard chain of evidence, and record logical breakpoints based on the search results. Key information in evidence with citation relationships is compared, and semantic recognition algorithms are used to identify content contradictions or inconsistencies, and citation breakpoints are recorded. Query the supporting relationships of evidence nodes. If there are no related supporting nodes, record the supporting breakpoint. The system counts the number of breakpoints for different types, calculates the breakpoint density, and generates breakpoint type and severity labels.

4. The intelligent case source allocation method based on online judicial mediation according to claim 1, characterized in that, Based on case feature tags, party background tags, and evidence breakpoint tags, we predict the probability of new disputes and the areas of new disputes, including: The case feature tags, party background tags, and evidence breakpoint tags are fused using feature weighting to generate a global fusion vector. The global fusion vector is input into a pre-trained classification model to obtain a predicted new disputed domain, wherein there is at least one predicted new disputed domain; The global fusion vector and each predicted new dispute domain are combined separately, and the combined results are input into the pre-trained probability prediction model to obtain the new dispute probability corresponding to each new dispute domain. Threshold filtering is performed based on the probability of disputes to obtain at least one target area of ​​newly added disputes and the corresponding probability of newly added disputes.

5. The intelligent case source allocation method based on online judicial mediation according to claim 1, characterized in that, Based on case feature tags, dispute probability, newly added dispute areas, and mediator information, the matching degree of each mediator is calculated, and a candidate list is generated by sorting the matching degrees, including: The similarity between the case domain and the mediator's professional domain is calculated based on the case feature tags and the newly added disputed domains to obtain the domain matching degree. The weights of the case feature tags and the newly added disputed domains are adjusted according to the probability of the dispute. The propensity weights of historical mediation success rate and complex case handling volume are calculated based on the probability of disputes. The capability matching degree is obtained by weighting the historical mediation success rate, complex case handling volume and propensity weights. The similarity between case feature tags and mediators' historical successful cases is calculated to obtain the experience matching degree; The matching degree of the mediator is obtained by weighting and summing the domain matching degree, ability matching degree and experience matching degree. Mediators with a matching degree greater than the minimum requirement are selected and sorted from highest to lowest matching degree to generate a candidate list.

6. The intelligent case allocation method based on online judicial mediation according to claim 1, characterized in that, Based on the newly added disputes, the selected mediators will be assigned a simulated workload within a preset timeframe after matching, including: Obtain the mediator's task list, find similar historical cases based on the case feature tags of each case, and calculate the average processing time and effort required for similar historical cases; Calculate the remaining estimated processing time for each case based on the processing status of the task list; Cases to be assigned are inserted into the task list according to case priority, generating a simulated to-do queue; Based on the average processing time and energy demand, time consumption simulation and energy consumption simulation are performed according to the order of cases in the simulated pending queue until the preset time is exhausted or all cases are processed. After the simulation ends, calculate the total remaining time in the queue and the total energy consumption, and then calculate the workload based on the total remaining time in the queue and the total energy consumption.

7. An intelligent case allocation system based on online judicial mediation, characterized in that, include: The first module is used to obtain background information of the parties involved and case text information, and to extract case feature tags and party background tags respectively. The second module is used to construct an evidence map by analyzing the relationships between evidence, and to extract evidence breakpoint labels based on the evidence map; The third module is used to predict the probability of new disputes and the areas of new disputes based on case feature tags, party background tags, and evidence breakpoint tags. The fourth module is used to obtain mediator information, which includes historical mediation success rate, professional field, and number of complex cases handled. The fifth module is used to calculate the matching degree of each mediator based on case feature tags, dispute probability, newly added dispute areas and mediator information, and generate a candidate list by sorting the matching degree. The sixth module is used to determine whether the probability of dispute is less than a preset threshold. If so, push allocation information is generated based on the matching degree. If not, select a preset number of mediators from the candidate list, simulate the workload of the selected mediators within a preset time after matching according to the new disputes, and generate push allocation information based on the matching degree and workload.

8. The intelligent case allocation system based on online judicial mediation according to claim 7, characterized in that, The second module includes: The first unit is used to obtain textual and non-textual evidence, and to convert non-textual evidence into text format to form a text summary; The second unit is used to generate a unique identifier for each piece of evidence and extract the evidence type, time of its formation, purpose of proof, and key information. The third unit is used to parse the evidence text and establish citation relationships by extracting information related to the evidence and the cited evidence. The fourth unit is used to establish supporting relationships by calculating the similarity between the purposes of proof; The fifth unit is used to establish a chronological relationship between evidence related to the same fact by comparing the time of formation and key information; The sixth unit is used to generate an evidence map based on the citation relationships, supporting relationships, and chronological relationships between evidence.

9. The intelligent case allocation system based on online judicial mediation according to claim 8, characterized in that, The second module also includes: The seventh unit is used to obtain a standard chain of evidence based on the case type, search for evidence based on the standard chain of evidence, and record logical breakpoints based on the search results. The eighth unit is used to compare key information of evidence with citation relationships, identify content contradictions or inconsistencies through semantic recognition algorithms, and record citation breakpoints. Unit 9 is used to query the supporting relationships of evidence nodes. If there are no related supporting nodes, the supporting breakpoint is recorded. Unit 10 is used to count the number of breakpoints of different types, calculate the breakpoint density, and generate breakpoint type labels and severity labels.

10. The intelligent case allocation system based on online judicial mediation according to claim 7, characterized in that, The third module includes: Unit 11 is used to perform feature weighted fusion of case feature tags, party background tags, and evidence breakpoint tags to generate a global fusion vector; The twelfth unit is used to input the global fusion vector into a pre-trained classification model to obtain a predicted new dispute domain, wherein there is at least one predicted new dispute domain; The thirteenth unit is used to combine the global fusion vector and each predicted new dispute domain separately, and input the combination results into the pre-trained probability prediction model to obtain the new dispute probability corresponding to each new dispute domain. The fourteenth unit is used to perform threshold filtering based on the probability of disputes, to obtain at least one target area of ​​newly added disputes and the corresponding probability of newly added disputes.

Images