A method for identifying a dispute focus, an electronic device, and a storage medium

Abstract

This application discloses a method, electronic device, and storage medium for identifying points of contention. The method includes: acquiring factual information of a target case; based on the factual information, calling a large-scale point-of-contention identification model to identify points of contention according to identification logic, thereby obtaining relevant information about the points of contention in the target case; wherein, the relevant information about the points of contention includes the points of contention, the identification logic includes rules for identifying points of contention, and the rules for identifying points of contention are used to constrain the output of the large-scale point-of-contention identification model to conform to the rules for identifying points of contention. This solution can improve the accuracy of point-of-contention identification.

Description

Technical Field

[0001] This application relates to the field of legal data analysis, and in particular to a method for identifying points of contention, an electronic device, and a storage medium. Background Technology

[0002] In legal practice and case analysis, accurately identifying the points of contention in a case is the foundation for effective legal reasoning and strategy development.

[0003] Traditional methods rely on lawyers reading and analyzing case materials, which is inefficient. With the development of artificial intelligence technology, automated legal analysis based on large language models has become possible. However, existing methods using general large language models for identifying points of contention typically treat this as a holistic, general task, often resulting in low accuracy and impacting subsequent legal analysis. Summary of the Invention

[0004] This application provides at least one method, electronic device, and storage medium for identifying points of contention, in order to address the problem of inaccurate automated identification of points of contention in the prior art.

[0005] This application provides a method for identifying points of contention, comprising: acquiring factual information of a target case; based on the factual information, calling a large-scale point-of-contention identification model to identify points of contention according to identification logic, thereby obtaining relevant information on points of contention of the target case; wherein, the relevant information on points of contention includes points of contention, the identification logic includes rules for identifying points of contention, the rules for identifying points of contention are used to constrain the output of the large-scale point-of-contention identification model to points of contention that conform to the rules for identifying points of contention, and the rules for identifying points of contention include at least one of preset category rules, quantity rules, and content restriction rules.

[0006] This application provides an electronic device, including a memory and a processor, wherein the processor is configured to execute program instructions stored in the memory to implement any of the methods described above.

[0007] This application provides a computer-readable storage medium having program instructions stored thereon, which, when executed by a processor, implement any of the methods described above.

[0008] In the above scheme, by setting up identification logic that includes rules for identifying the focus of the dispute, the identification behavior of the large model for identifying the focus of the dispute is precisely constrained, so that the large model strictly follows the preset logic and performs identification in accordance with the identification rules, thereby guiding the identification results to a track that conforms to legal practice norms and improving the accuracy of automated identification of the focus of the dispute.

[0009] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this application. Attached Figure Description

[0010] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the specification, serve to explain the technical solutions of this application.

[0011] Figure 1 This is a flowchart illustrating an embodiment of the disputed focus identification method of this application; Figure 2 This is a flowchart illustrating another embodiment of the disputed focus identification method of this application; Figure 3 This is a flowchart illustrating yet another embodiment of the disputed focus identification method of this application; Figure 4 This is a flowchart illustrating yet another embodiment of the disputed focus identification method of this application; Figure 5 This is a schematic diagram of the framework of an embodiment of the electronic device of this application; Figure 6 This is a schematic diagram of a framework of an embodiment of the computer-readable storage medium of this application. Detailed Implementation

[0012] The embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0013] In the following description, specific details such as particular subsystem structures, interfaces, and technologies are presented for illustrative purposes rather than for limiting purposes, in order to provide a thorough understanding of this application.

[0014] In this document, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship. Furthermore, "many" in this document means two or more. Moreover, the term "at least one" in this document means any combination of at least two of any one or more of a plurality of objects. For example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C.

[0015] The embodiments of this application can be applied to scenarios such as legal case management systems and enterprise internal legal systems, and the provided methods can run on devices such as servers.

[0016] Users can run a front-end program on their own terminal device to interact with the aforementioned device: they can send text, files, and other content to the device, or view the results sent back by the device after processing on their terminal device.

[0017] See Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the disputed focus identification method of this application. Specifically, the method may include: Step S110: Obtain the factual information of the target case.

[0018] The target case is the case in which the focus of the dispute needs to be identified, and the case fact information represents information related to the facts of the target case. The case fact information can be obtained based on the case materials provided by the user regarding the target case.

[0019] In some embodiments, case materials can be obtained by acquiring user input text, acquiring and parsing user-uploaded files, etc. Case fact information can be obtained by analyzing case materials. Case fact information can reflect the facts of the case in a concentrated, concise and accurate manner, and can be used as material for subsequent analysis to accurately identify the focus of the dispute.

[0020] Step S120: Based on the case fact information, call the focus identification model to identify the disputed focus according to the identification logic, and obtain the relevant information of the disputed focus of the target case.

[0021] The identification logic is used to instruct the large model to identify the points of contention according to the identification logic. The identification logic can be in the form of text, and the meaning of its text content is used to represent the constraints on the large model. It can cover at least one of the following: the thinking logic to be followed, the rules to be followed, or descriptive text to reflect the meaning of the task or a certain noun.

[0022] The focus recognition model can be a Large Language Model (LLM), a deep learning model trained on massive amounts of text data, capable of understanding natural language and generating text. The specific model used can be adjusted according to the actual application needs, such as selecting a model based on reasoning difficulty and resource consumption. For example, models like the Tongyi Qianwen Large Model, GPT, and LLaMA can be selected.

[0023] Step S120 does not simply input case factual information into the large model for analysis; instead, it specifies the identification logic to the large model. The identification logic is a set of rules, constraints, and descriptive texts that instruct the large model on how to perform the task of identifying disputed issues. The identification logic contains explicit rules for identifying disputed issues, specifying the concrete thought process the large model follows when performing this task. This fully leverages the large model's powerful understanding and reasoning abilities while accurately limiting the logic and rules governing its identification. This solves the problem of poor accuracy in identifying disputed issues in open environments using general large models. Through the constraints of the identification logic, the large model's ability to generate disputed issues is guided towards meeting the requirements of legal practice, ensuring that the final disputed issue-related information is accurate, standardized, and practical.

[0024] The information related to the points of contention can include the points of contention themselves, listing each point of contention involved in the target case to help users understand the potential content of the dispute and facilitate further analysis using the identified points of contention. The identification logic includes point-of-contention identification rules that instruct the large-scale model how to identify points of contention, thereby constraining the model's reasoning.

[0025] Taking a specific application scenario as an example, when a user inputs a description of a contract dispute, this method is initiated on the backend. Step S110 obtains the case fact information, which includes the signing, performance, breach of contract, and damages. In step S120, the device constructs an identification logic that includes rules for identifying disputed issues, such as content restriction rules that "prohibit jurisdiction and statute of limitations from being listed as disputed issues," and preset category rules that "identify from the dimensions of rights and obligations, fact-finding, and applicable law." Subsequently, this identification logic loaded with rules and the case fact information are submitted to the disputed issue identification model for processing. During the processing, the model, according to the identification logic, excludes procedural issues such as "which court has jurisdiction" and focuses on analyzing substantive disputes such as "whether the breach of contract is established" and "whether the defendant has a liability for compensation," ultimately outputting a structured list of disputed issues focused on substantive rights and obligations.

[0026] In some embodiments, the identification logic may include dispute point identification rules, which instruct the large model how to identify dispute points and output dispute points that meet the rule requirements. Dispute point identification rules may include one or more rules, and different rules can define the content that the large model needs to follow from different perspectives.

[0027] Furthermore, the rules for identifying points of contention may include at least one of the following: preset category rules, quantity rules, and content restriction rules.

[0028] In some implementation scenarios, the rules for identifying points of contention may include preset category rules, quantity rules, and content restriction rules. Preset category rules can be used to instruct the model from which to identify points of contention, while quantity and content restriction rules can be used to instruct the model to output points of contention according to quantity and content restrictions. This constitutes the detailed scheme of this application for precisely constraining the model output from multiple dimensions.

[0029] In some embodiments, the identification logic may also include a rule for identifying related information, which instructs the large model to identify related information as relevant to the focus of the dispute. Related information represents information that is related to the focus of the dispute, facilitating the user's understanding of the target case, particularly regarding the focus of the dispute, such as the source of the dispute and its possible future handling.

[0030] In some implementation scenarios, the associated information includes at least one of the following: target case type, legal relationship, cause of action, and optional claims.

[0031] In some specific application scenarios, the execution of step S120 also includes at least one additional action: first, identifying at least one of the target case type, legal relationship, and cause of action from the case fact information; second, generating optional litigation claims based on the case fact information. The former helps users understand the situation of the target case, identify the category of the target case, and help users understand the typological characteristics of the target case, while the latter can provide users with expectations about the development of the case based on the focus of the dispute, so that users can grasp the development of the target case throughout the entire time process.

[0032] In some embodiments, the associated information includes case types, the associated information identification rules include preset case type rules, and the preset case type rules include multiple preset case types; identifying the case type from the case fact information includes: selecting one of the multiple preset case types that matches the case fact information as the target case type.

[0033] The above scheme combines the focus of the dispute with related information to build a more three-dimensional and richer analysis framework, which enhances the completeness of the output information and the contextual richness of subsequent analysis. A single call can output the key profile of the case in the dispute identification stage.

[0034] In some implementation scenarios, the rules for identifying points of contention may include preset category rules, quantity rules, and content restriction rules.

[0035] Among them, the preset category rules include multiple preset dispute focus categories, which means that multiple categories are pre-configured to instruct the large model to identify whether the case contains specific dispute focus under each preset dispute focus category.

[0036] Specifically, the process of calling the model based on the case's factual information and identification logic essentially instructs the focus identification model to identify, from the case's factual information, multiple pre-defined categories of disputed issues, such as rights, obligations, and factual determinations. Through these pre-defined category rules, the model no longer extracts information aimlessly; it checks whether one or more of the aforementioned dimensions of dispute exist in the case, thereby refining multi-dimensional focus points. This effectively prevents the omission of certain key disputed dimensions, ensuring coverage of multiple legal aspects of the case and improving the comprehensiveness and accuracy of disputed issue identification.

[0037] Furthermore, the predefined category rules may also include category descriptions for each predefined category of dispute, instructing the large model to understand the meaning of the predefined dispute category based on the category description and thereby identify the specific dispute that matches the meaning of the category description. These descriptive texts provide specific judgment criteria for the model to identify various types of disputes, further improving the accuracy of dispute classification and identification.

[0038] It should be noted that a single point of contention can correspond to one preset point of contention category, or it can correspond to multiple preset point of contention categories.

[0039] Among them, the quantity rule can represent the quantity limit on the points of contention, which is used to instruct the large model to identify and output the points of contention according to the limit of this quantity rule.

[0040] In some embodiments, the number of points of contention rule can be used to indicate that the number of points of contention is positively correlated with the complexity of the case.

[0041] Among them, the content restriction rule indicates the content restriction on the focus of the dispute, which is used to instruct the large model to control the content of the focus of the dispute to comply with the set content restriction.

[0042] In some embodiments, the dispute focus identification rule also includes an output priority rule, which instructs the large model to sort the obtained dispute focuses according to the priority rule and output them in descending order of priority.

[0043] In some application scenarios, based on the factual information of the case, a large-scale focus identification model is invoked to identify the points of contention according to the identification logic. The resulting information on the points of contention in the target case includes: evaluating the priority of each point of contention according to output priority rules, sorting the points of contention by priority, and outputting the identified points of contention sequentially. This means that the list of points of contention obtained by the user or downstream nodes is not randomly arranged, but rather evaluated for importance. This approach allows legal professionals to focus on the most important points of contention first, thereby quickly grasping the core of the case and optimizing decision-making efficiency. In the output, the highest priority points are most likely to determine the course of the case, reflecting professional legal judgment.

[0044] In some embodiments, the dispute point identification rule also includes at least one of the task description text, role description text, and output format.

[0045] In some embodiments, the rules for identifying points of contention also include task description text, role description text, and output format.

[0046] The task description text can be used to describe the task content and define the task boundaries so that the large model can execute the task described in the text within the task boundaries.

[0047] Among them, the character description text can be used to restrict the character perspective during large model inference, so that the large model can reason according to the specified character perspective, making the reasoning results more consistent with the specified character properties and improving accuracy.

[0048] The output format can be used to restrict the format of the data output by large models, so as to obtain output data that meets user needs and is easy to view and use directly later.

[0049] By using task description text and role description text, the focus recognition model is given a specific "identity" (such as "a Chinese practicing lawyer with many years of litigation experience") and clear task boundaries. This is the core application of cue word engineering in this method, which can greatly enhance the model's professional capabilities in the legal field. Constraints on the output format (such as requiring a specific JSON structure) ensure the machine readability of the results.

[0050] See Figure 2 , Figure 2 This is a flowchart illustrating another embodiment of the disputed focus identification method of this application. The method includes: Step S210: Obtain the factual information of the target case.

[0051] Step S220: Call the focus identification big model to identify the disputed focus that matches each preset disputed focus category from the case fact information, and take the identified disputed focus and its preset focus category as the disputed focus.

[0052] In some implementation scenarios, multiple pre-set categories of disputed issues include at least one of the following: rights and obligations, fact-finding, application of law, evidence, common sense and reason, and remedies.

[0053] In some implementation scenarios, multiple pre-defined categories of disputed issues include rights and obligations, fact-finding, application of law, evidence, common sense and reason, and remedies.

[0054] In some implementation scenarios, the preset category rules also include category descriptions for each preset category of dispute focus.

[0055] In some application scenarios, the preset category rules also include category descriptions for each preset dispute focus category. The category descriptions of rights and obligations include at least one of the following: whether the rights and obligations relationship exists or whether the rights and obligations relationship has been violated.

[0056] In some application scenarios, the categories of fact-finding include at least one of the following: whether there is a dispute over the key facts, whether the act occurred, whether a causal relationship is established, and whether damage actually occurred.

[0057] In some application scenarios, the category description of applicable law includes whether the legal norms are clear, whether there are competing or conflicting legal norms, or at least one of these.

[0058] In some application scenarios, the description of evidence categories includes at least one of the following: whether the existing evidence is sufficient to support the claims of all parties, whether the evidence of all parties is sufficient, and whether the probative value of the evidence is sufficient.

[0059] In some application scenarios, the category description of common sense and logic includes whether there are any obvious illogical aspects when considering transaction habits, industry practices, and rules of thumb in daily life.

[0060] In some application scenarios, the description of the relief category includes whether the relief claimed by the party is appropriate and whether there is at least one of the alternatives.

[0061] See Figure 3 , Figure 3 This is a flowchart illustrating another embodiment of the disputed focus identification method of this application. The method includes: Step S310: Obtain the factual information of the target case.

[0062] Step S320: Call the focus identification model to identify disputed focuses according to the focus quantity rules, and obtain the number of disputed focuses that meet the quantity limit.

[0063] Among them, the focus number rule can link the complexity of the case, which is represented by the meaning of the factual information, with the final focus number control.

[0064] In some application scenarios, the focus quantity rule is used to instruct the focus identification model to determine the complexity of the target case from simple, moderately complex, or very complex based on the case facts. In response to a simple case complexity, the number of disputed focuses output is 1 or 2; in response to a moderate case complexity, the number of disputed focuses output is 3 or 4; and in response to a very complex case complexity, the number of disputed focuses output is 5 or 6. This dynamic quantity matching mechanism avoids artificially creating focuses for simple cases while ensuring that focuses in complex cases are not overlooked, thus guaranteeing the conciseness and completeness of the output results.

[0065] See Figure 4 , Figure 4 This is a flowchart illustrating another embodiment of the disputed focus identification method of this application. The method includes: Step S410: Obtain the factual information of the target case.

[0066] Step S420: Call the focus identification model to identify the disputed focus according to the content restriction rules, and obtain the disputed focus that meets the content restriction.

[0067] In some embodiments, content restriction rules can be used to indicate that the focus of the dispute does not include jurisdiction and statute of limitations. This hard constraint ensures that the output focuses on substantive legal issues, conforms to the definition of "focus of the dispute" in legal practice, and improves the professionalism and accuracy of focus identification.

[0068] In some embodiments, content restriction rules are used to instruct the focus identification model to determine whether the factual information of the case contains an objection to the statute of limitations. In response to the determination that the factual information does not contain an objection to the statute of limitations, the focus of the dispute is determined to exclude jurisdiction and the statute of limitations. In response to the determination that the factual information contains an objection to the statute of limitations, the meaning of the focus of the dispute is determined to exclude jurisdiction but may include the statute of limitations. This approach allows for flexible control of the content restrictions on the focus of the dispute based on the content of the factual information of the case. This enables the determination of the focus of the dispute in a scenario where there are objections to the statute of limitations by the parties, rather than simply restricting the focus from containing the statute of limitations entirely. This refined judgment logic gives the rules both rigidity and flexibility.

[0069] In some implementation scenarios, content restriction rules can include two sub-rules. The first sub-rule is used to instruct the model to control the generated content to conform to the content restriction rules during generation. The second sub-rule is used to instruct the model to check whether the preliminary results conform to the content restriction rules after generation, and to remove the parts of the preliminary results that do not conform to the content restrictions so as to use them as the final output.

[0070] It should be noted that the identification logic can include multiple rules. Furthermore, the rules for identifying the focus of the dispute can also include preset category rules, quantity rules, content restriction rules, and other rules. The large model for identifying the focus of the dispute needs to follow all the rules in the identification logic to identify the focus.

[0071] In some embodiments, the method may further include: analyzing the applicable legal rules for each point of contention, retrieving and analyzing related case information for each point of contention, and analyzing strategies for dealing with each point of contention. The points of contention can serve as the basis for subsequent legal analysis. Analyzing each point of contention separately refines the legal analysis, accurately identifying the points of contention makes the basis for subsequent analysis more accurate and clear, thus improving the precision and accuracy of the analysis. Reasonably breaking down cases according to the points of contention controls the complexity of the content involved in a single analysis, avoiding interference from irrelevant content and also improving the accuracy of subsequent analysis.

[0072] The recognition logic can include a textual description of the rules and logic for focus recognition. Given the understanding capabilities of large models, the recognition logic is configured to constrain model reasoning and output, thus constraining automated focus recognition to predetermined logic. Through the layered refinement of the logic, the automated focus recognition is refined, thereby improving the accuracy of focus recognition.

[0073] In some embodiments, the focus identification model constraints include corresponding requirements, namely, identifying the core legal disputes of a case from six dimensions: rights and obligations, fact-finding, applicable law, evidence, common sense and reason, and remedies. Furthermore, it prohibits listing procedural issues such as jurisdiction and statute of limitations as points of contention; the focus must be on the substantive rights and obligations between the parties. Each point of contention must be concise and impactful, limited to 30 words, and ordered by importance. The type of document to be judged must also be limited, and must be selected from 10 standard categories. In addition, the number of points of contention should be flexibly determined according to the substantive complexity of the case: 1-2 for simple cases, 3-4 for moderately complex cases, and up to 5-6 for extremely complex cases; fabricating disputes simply to reach the required number is prohibited.

[0074] Traditional methods of using models to analyze the focus of disputes typically treat the identification of the focus of disputes as a general task, which often results in low accuracy. For example, if a case description of a dispute between Wang and Li over a dog injuring someone is used as input, the result will be that the focus of the dispute is whether Wang's kicking of the dog constitutes an infringement.

[0075] Using the same case description as in the previous example as input, when a user inputs a description of a dispute between Wang and Li over a dog injury, the method provided in this application is initiated on the backend. By calling the focus identification model output in the above manner, the case fact information is obtained, and an identification logic containing rules for identifying disputed points of contention is constructed, such as content restriction rules and preset category rules. Subsequently, this rule-loaded identification logic and the case fact information are submitted to the focus identification model for processing. During processing, the model, according to the identification logic, excludes procedural issues such as "which court has jurisdiction" based on the rules, focusing instead on analyzing the substantive dispute. Ultimately, it outputs a structured list of disputed points of contention focusing on substantive rights and obligations. Specifically, disputed point 1 is whether Wang's kicking the dog and causing injury constitutes a tort, corresponding to the categories of fact-finding and rights and obligations; disputed point 2 is whether Li himself was negligent in managing the dog and the proportion of his liability, corresponding to the categories of evidence and common sense, and the legal relationship is a tort relationship in a dispute over liability for damage caused by animals.

[0076] Compared with existing technologies, the method of this application, by identifying the constraints of specific rules in the logic on the model, enables the model to be more detailed when identifying the points of contention, resulting in more accurate points of contention and thus improving the accuracy of subsequent analyses based on the points of contention.

[0077] In some embodiments, the points of contention can be applied to the rule retrieval and mapping process, and the large model can be invoked to perform the retrieval and mapping of applicable rules for each point of contention.

[0078] In some embodiments, the points of contention can be applied to the legal reasoning process, and a large model can be invoked to conduct reasoning for each point of contention separately, deriving the possible legal conclusions corresponding to that point of contention.

[0079] Specifically, by using the legal reasoning logic to constrain the legal reasoning model, legal reasoning can be conducted based on the facts of the case and the points of contention, and the reasoning analysis results of the current case can be obtained.

[0080] Furthermore, the focus analysis rules in legal reasoning logic are used to instruct the large-scale legal reasoning model to analyze the applicable law, reference cases, factual application, and reasoning conclusions of the focus of the dispute, while the analysis and summary rules are used to instruct the large-scale legal reasoning model to sort out the overall conclusions and risk levels of the current case.

[0081] In one embodiment, the process includes: selecting the applicable law corresponding to the focus of the dispute from several relevant laws, selecting referable cases corresponding to the focus of the dispute from several related cases, reasoning the application of facts by combining the facts of the case and the main points of the judgments of the referable cases, and then drawing a reasoning conclusion on the focus of the dispute based on the analysis of the application of facts.

[0082] In one embodiment, the process of analyzing the overall conclusion and risk level of the current case includes: based on the risk level descriptions corresponding to all risk levels, filtering out the risk level descriptions that semantically match the current case, and determining the corresponding risk level as the risk level of the current case.

[0083] In some embodiments, the points of contention can be applied to the adversarial simulation process, and a large model can be invoked to conduct a round of multi-role adversarial simulation for each point of contention.

[0084] Specifically, the adversarial deduction model, which utilizes multi-role adversarial deduction logic constraints, conducts adversarial deduction based on the points of contention, case facts, and reasoning analysis results to obtain the deduction results.

[0085] Furthermore, the focus deduction rule in the multi-role adversarial deduction logic is used to instruct the adversarial deduction model to generate legal opinions from multiple preset roles regarding the focus of the dispute according to the rules, while the deduction summary rule is used to instruct the adversarial deduction model to summarize and analyze the current case from a neutral perspective.

[0086] The focus deduction rules include sub-deduction rules corresponding to each preset role: the sub-deduction rules for the opposing role are used to instruct the adversarial deduction model to analyze the argument doubts of our role in the current case and generate legal opinions unfavorable to our role based on the argument doubts; the sub-deduction rules for our role are used to instruct the adversarial deduction model to generate legal opinions favorable to our role using information from related cases; the sub-deduction rules for the neutral role are used to instruct the adversarial deduction model to issue legal opinions against the claims of non-neutral roles using information from related cases.

[0087] See Figure 5 , Figure 5 This is a schematic diagram of the framework of an embodiment of the electronic device of this application.

[0088] The electronic device 50 includes a memory 51 and a processor 52. The processor 52 is used to execute program instructions stored in the memory 51 to implement the steps in any of the above embodiments of the dispute focus identification method. In a specific implementation scenario, the electronic device 50 may include, but is not limited to, computer equipment, electrical equipment, microcomputers, desktop computers, and servers. In addition, the electronic device 50 may also include mobile devices such as laptops and tablets, which are not limited here.

[0089] Specifically, processor 52 controls itself and memory 51 to implement the steps in any of the above-described embodiments of the dispute focus identification method. Processor 52 may also be referred to as a CPU (Central Processing Unit). Processor 52 may be an integrated circuit chip with signal processing capabilities. Processor 52 may also be a general-purpose processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component. A general-purpose processor may be a microprocessor or any conventional processor. Furthermore, processor 52 may be implemented using integrated circuit chips.

[0090] See Figure 6 , Figure 6 This is a schematic diagram of a framework of an embodiment of the computer-readable storage medium of this application.

[0091] The computer-readable storage medium 60 provided in this embodiment stores program instructions 61 that can be executed by a processor. When the program instructions 61 are executed by the processor, they are used to implement the steps in any of the above embodiments of the dispute focus identification method.

[0092] The description of the various embodiments above tends to emphasize the differences between the various embodiments. The similarities or similarities between them can be referred to, and for the sake of brevity, they will not be repeated here.

[0093] In the several embodiments provided in this application, it should be understood that the disclosed methods and apparatus can be implemented in other ways. For example, the apparatus implementations described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, units or components may be combined or integrated into another subsystem, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection of devices or units may be electrical, mechanical, or other forms.

[0094] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a server, network device, etc.) or processor to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for identifying points of contention, characterized in that, The method includes: Obtain factual information about the target case; Based on the factual information of the case, the focus identification model is invoked to identify the focus of the dispute according to the identification logic, thereby obtaining relevant information on the focus of the dispute in the target case; The relevant information of the dispute focus includes the dispute focus itself, and the identification logic includes dispute focus identification rules. The dispute focus identification rules are used to constrain the output of the large focus identification model to the dispute focus that conforms to the dispute identification rules. The dispute focus identification rules include at least one of preset category rules, quantity rules, and content restriction rules.

2. The method according to claim 1, characterized in that, The preset category rules include multiple preset dispute focus categories. Based on the case fact information, the process of calling the focus identification model to identify dispute focuses according to identification logic, and obtaining relevant information about the dispute focus of the target case, includes: The focus identification model is invoked to identify the points of contention that match each of the preset focus categories from the case fact information, and the identified points of contention and their corresponding preset focus categories are taken as the points of contention. The focus quantity rule represents a limit on the number of disputed focuses output by the focus identification model; the step of calling the focus identification model based on the case fact information to identify disputed focuses according to the identification logic, and obtaining relevant information about the disputed focuses of the target case, includes: The focus identification model is invoked to identify disputed focuses according to the focus quantity rules, and the number of disputed focuses that meets the quantity limit is obtained. The content restriction rules represent content restrictions on the points of contention. The step of using the point-of-contention identification model based on the case facts to identify the points of contention according to identification logic, and obtaining relevant information about the points of contention in the target case, includes: The focus identification model is invoked to identify the controversial focus according to the content restriction rules, thereby obtaining the controversial focus that conforms to the content restrictions.

3. The method according to claim 2, characterized in that, The multiple pre-defined categories of dispute points include at least one of the following: rights and obligations, fact-finding, application of law, evidence, common sense and reason, and remedies; The rule on the number of points of contention indicates that the number of points of contention is positively correlated with the complexity of the case. The content restriction rule indicates that the issues in dispute do not include jurisdiction and statute of limitations.

4. The method according to claim 3, characterized in that, The preset category rules also include category descriptions for each of the preset dispute focus categories. The category descriptions for rights and obligations include at least one of the following: whether a rights and obligations relationship exists, and whether a rights and obligations relationship has been violated. The category descriptions for fact-finding include at least one of the following: whether a key fact is in dispute, whether an act has occurred, whether a causal relationship has been established, and whether damage has actually occurred. The category descriptions for applicable law include at least one of the following: whether legal norms are clear, and whether there are competing or conflicting legal norms. The category descriptions for evidence include at least one of the following: whether existing evidence is sufficient to support the claims of all parties, whether the evidence of all parties is sufficient, and whether the probative value of the evidence is sufficient. The category descriptions for common sense and reason include whether there are any obvious illogicalities when considering transaction habits, industry practices, and rules of daily life experience. The category descriptions for remedies include at least one of the following: whether the remedy claimed by the parties is appropriate, and whether there are alternative solutions. The focus number rule is used to instruct the focus identification big model to determine the case complexity of the target case from simple, medium, or very complex based on the case fact information. In response to the case complexity being simple, the number of disputed focuses output is determined to be 1 or 2; in response to the case complexity being medium, the number of disputed focuses output is determined to be 3 or 4; and in response to the case complexity being very complex, the number of disputed focuses output is determined to be 5 or 6. The content restriction rule is used to instruct the focus identification big model to determine whether the case fact information contains an objection to the statute of limitations. In response to determining that it does not contain such an objection, the focus of the dispute is determined to not include the jurisdiction and the statute of limitations. In response to determining that it does contain such an objection, the meaning of the focus of the dispute is determined to not include the jurisdiction.

5. The method according to claim 1, characterized in that, The information related to the point of contention also includes associated information, and the identification logic also includes rules for identifying associated information; The associated information includes at least one of the following: target case type, legal relationship, cause of action, and optional claims; the process of using the focus identification model based on the case fact information to identify the focus of the dispute according to the identification logic, and obtaining the relevant information of the focus of the dispute in the target case, also includes at least one of the following: Identify at least one of the target case type, the legal relationship, and the cause of action from the case fact information; Based on the aforementioned case facts, the optional litigation claims are generated.

6. The method according to claim 5, characterized in that, The associated information includes case types, and the associated information identification rules include preset case type rules, which include multiple preset case types. The process of identifying the case type from the case fact information includes: Select one of the multiple preset case types that matches the case fact information as the target case type.

7. The method according to claim 1, characterized in that, The dispute focus identification rule also includes an output priority rule. The step of calling the focus identification model based on the case fact information to identify the dispute focus according to the identification logic, and obtaining relevant information about the dispute focus of the target case, further includes: The priority of each point of contention is evaluated according to the output priority rules, and the points of contention are sorted according to the priority.

8. The method according to claim 1, characterized in that, The rules for identifying points of contention also include at least one of the following: task description text, role description text, and output format. The method further includes: For each of the aforementioned points of contention, the applicable legal rules are analyzed, related case information is retrieved and analyzed, and strategies for dealing with the points of contention are analyzed.

9. An electronic device, characterized in that, The device includes a memory and a processor, wherein the memory stores program instructions that, when executed by the processor, implement the method of any one of claims 1 to 8.

10. A computer-readable storage medium having program instructions stored thereon, characterized in that, When the program instructions are executed by the processor, they implement the method of any one of claims 1 to 8.

Images