Digital collaborative management and multi-element early warning system for entire process of bankruptcy case

Abstract

The application discloses a bankruptcy case whole-process digital collaborative management and multi-element early warning system, belongs to the technical field of computers, and provides the bankruptcy case whole-process digital collaborative management and multi-element early warning system, which can identify potential risk points in each stage in real time and trigger early warning through integration of multi-source information and dynamic analysis of disposal risks, solves the problems of collaborative obstacles and lagging risk identification caused by information islands in the traditional mode, and has the advantages of improving whole-process collaborative efficiency of bankruptcy cases and realizing multi-dimensional risk early warning.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a digital collaborative management and multi-dimensional early warning system for the entire process of bankruptcy cases. Background Art

[0002] In the field of bankruptcy case management, the traditional model has long faced systemic challenges. Bankruptcy procedures typically involve multiple stages of disposal, including case filing review, claim declaration, property investigation, and asset disposal. Each stage requires collaboration among multiple parties, including the court, administrator, and creditors. Because case information is scattered across paper archives, independent databases, and internal systems of different institutions, key data such as corporate asset clues, related-party transaction records, and creditor distribution are often fragmented. In practice, disposal agencies (such as bankruptcy administrators) find it difficult to dynamically integrate cross-stage case description information (such as property inventory progress), agency-specific information (such as administrator performance records), and disposal methods (such as asset valuation plans), and are even unable to link external public information (such as industrial and commercial changes and litigation-related announcements) in real time.

[0003] The current industry urgently needs a management solution that can connect the entire bankruptcy process, integrate multi-source information and achieve risk linkage, so as to solve the coordination barriers and systemic risk blind spots caused by information islands. Summary of the Invention

[0004] The present embodiment provides a digital collaborative management and multi-dimensional early warning system for the entire bankruptcy process, which can provide early warnings in a timely manner. The technical solution is as follows:

[0005] On the one hand, a digital collaborative management and multi-dimensional early warning system for the entire process of bankruptcy cases is provided, comprising:

[0006] an acquisition module, configured to acquire stage description information, disposal information, and disposal reference information of a target bankruptcy case of a target enterprise at each case disposal stage, wherein the disposal information includes reference enterprise information of the target enterprise used in the corresponding case disposal stage, proprietary information of the disposal agency in the corresponding case disposal stage, and the disposal method in the corresponding case disposal stage. The disposal reference information is publicly available information related to the target enterprise and the target bankruptcy case;

[0007] A first determination module is configured to determine, from the disposal reference information, the stage reference information and the associated object information of each case disposal stage based on the stage description information, the institution-specific information, the disposal method, and the reference enterprise information of each case disposal stage;

[0008] The second determination module is used to determine the disposal risk information of each case disposal stage based on the institution-specific information, stage reference information, related object information and disposal method of each case disposal stage;

[0009] a third determining module, configured to determine at least one target bankruptcy case handling stage from the multiple case handling stages of the target bankruptcy case based on handling risk information of each case handling stage, the target bankruptcy case handling stage being a case handling stage that needs to be warned.

[0010] In a possible implementation, the first determining module is configured to, for any case handling stage in the multiple case handling stages, determine stage reference information of the case handling stage from the handling reference information based on stage description information, agency-specific information and a handling manner of the case handling stage, the stage description information being used for natural language description of the handling stage, the stage reference information being used to assist in judging rationality of the handling manner of the case handling stage; determine associated object information of the case handling stage from the handling reference information based on the stage description information, the agency-specific information and reference enterprise information, the associated object information being object information of an associated object associated with the target enterprise under the case handling stage, the associated object including at least one of an associated enterprise, an associated individual and an associated group.

[0011] In a possible implementation, the first determining module is configured to determine initial reference information of the case handling stage from the handling reference information based on stage description information of the case handling stage, the initial reference information being information matched with the stage description information; determine multiple first reference information and multiple second reference information from the initial reference information based on the stage description information and the agency-specific information, the first reference information being information matched with the agency-specific information, the second reference information being information not matched with the agency-specific information, a quantity of the first reference information and the second reference information being determined based on a confidence degree of the agency-specific information; determine the stage reference information from the multiple first reference information and the multiple second reference information based on the handling manner.

[0012] In a possible implementation, the first determining module is configured to perform semantic coding on the stage description information to obtain stage description semantic features of the stage description information; determine initial reference information of the case handling stage from the handling reference information based on the stage description semantic features.

[0013] The first determination module is configured to determine reference institution-specific information from the institution-specific information based on the stage description information, wherein the degree of correlation between the reference institution-specific information and the case handling stage is greater than or equal to a preset degree of correlation; determine a plurality of first initial reference information and a plurality of second initial reference information from the initial reference information based on the reference institution-specific information; determine the plurality of first reference information from the plurality of first initial reference information based on the confidence level of the reference institution-specific information in the confidence level of the institution-specific information, and determine the plurality of second reference information from the plurality of second initial reference information;

[0014] The first determination module is used to semantically encode the disposal method to obtain the disposal method semantic features of the disposal method; and use the disposal method semantic features to match the multiple first reference information and the multiple second reference information to obtain the stage reference information.

[0015] In one possible implementation, the first determination module is used to determine a first association map of the target enterprise based on the organization-specific information, where the first association map is used to represent the association relationship between the target enterprise and other objects; determine a second association map of the target enterprise based on the reference enterprise information, where the second association map is used to represent the association relationship between the target enterprise and other objects; and determine the associated object information of the case disposal stage from the disposal reference information based on the stage description information, the first association map, and the second association map.

[0016] In a possible implementation, the first determining module is configured to obtain first association information of the target enterprise from the organization-specific information, where the first association information is used to represent an association relationship between the target enterprise and other objects;

[0017] The first determining module is configured to perform a query based on the reference enterprise information to obtain second association information of the target enterprise, where the second association information is used to represent an association relationship between the target enterprise and other objects;

[0018] The first determination module is used to use the second association map to expand or adjust the first association map to obtain the target association map of the target enterprise, and the first association map is used as a reference for expansion or adjustment; based on the stage description information and the target association map, the associated object information of the case disposal stage is determined from the disposal reference information.

[0019] In one possible implementation, the second determination module is used to determine, for any case handling stage among the multiple case handling stages, the handling method risk information based on the institution-specific information, stage reference information, and handling method of the case handling stage; determine the associated object risk information based on the institution-specific information and associated object information of the case handling stage; and determine the handling risk information of the case handling stage based on the handling method risk information and the associated object risk information.

[0020] In a possible implementation, the second determination module is configured to generate a plurality of reference disposal methods based on the stage reference information and the institution-specific information; and determine disposal method risk information based on the plurality of reference disposal methods and the disposal method.

[0021] In a possible implementation, the second determining module is configured to obtain associated object proprietary information from the organization proprietary information based on the associated object information; and determine the associated object risk information based on the associated object proprietary information.

[0022] In one possible implementation, the third determination module is used to determine the reference risk score of each case handling stage based on the handling risk information of each case handling stage; and determine the case handling stage among the multiple case handling stages whose reference risk score is greater than or equal to the preset risk score as the target bankruptcy case handling stage to obtain at least one target bankruptcy case handling stage. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following is a brief introduction to the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0024] Figure 1 This is a schematic diagram of an implementation environment for a digital collaborative management and multi-dimensional early warning system for the entire bankruptcy process provided by an embodiment of the present application;

[0025] Figure 2 This is a structural diagram of a full-process digital collaborative management and multi-dimensional early warning system for bankruptcy cases provided in an embodiment of the present application. DETAILED DESCRIPTION

[0026] In order to make the objectives, technical solutions and advantages of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0027] In this application, the terms "first", "second", etc. are used to distinguish identical or similar items with substantially the same effects and functions. It should be understood that there is no logical or temporal dependency between "first", "second", and "nth", nor is there any limitation on the quantity and execution order.

[0028] Bankruptcy: refers to an event of insolvency handled through judicial procedures. All three types of bankruptcy proceedings are independent bankruptcy proceedings and can be initiated directly.

[0029] Artificial Intelligence (AI) is the theory, system, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to achieve better results.

[0030] Machine learning (ML) is a multidisciplinary field that encompasses probability theory, statistics, approximation theory, convex analysis, and algorithmic complexity theory. It specifically studies how computers can simulate or implement human learning behaviors to acquire new knowledge or skills and reorganize existing knowledge sub-models to continuously improve their performance. Machine learning is at the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications span all areas of AI. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and learning through demonstration.

[0031] It should be noted that the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data used for analysis, stored data, displayed data, etc.) and signals involved in this application are all authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions.

[0032] In addition, it should be noted that the technical solutions provided in the embodiments of the present application can provide reference information for judges. Such reference information is used to assist judges in trying cases and should not be used as direct evidence or the basis for the trial of cases.

[0033] In related technologies, bankruptcy case management has long faced the problem of information silos. Case information is scattered across paper files, independent databases, and different institutional systems, resulting in the fragmentation of critical data. Disposal agencies struggle to dynamically integrate case descriptions, institutional proprietary information, and disposal methods across various stages, and are unable to effectively link external publicly available information. For example, during the asset disposal phase, administrators may be unable to obtain timely industrial and commercial change records for affiliated companies, resulting in asset pricing deviating from true market value. This data fragmentation inefficiently collaborates across institutions and creates blind spots in the identification of systemic risks.

[0034] To address these issues, the inventors discovered that traditional systems lacked a multi-source data integration mechanism, resulting in delayed risk warnings. Analyzing the entire bankruptcy case process, they discovered information gaps between each stage of the process. For example, corporate financial data from the claims filing phase failed to effectively connect with asset clues from the property investigation phase. This led to the idea of ​​building a dynamic data linkage framework. By mapping stage characteristics with external information, this framework visualizes risk transmission pathways.

[0035] Therefore, this application proposes a digital collaborative management and multi-dimensional early warning system for the entire process of bankruptcy cases to solve the above problems.

[0036] Figure 1 This is a schematic diagram of the implementation environment of a digital collaborative management and multi-dimensional early warning system for the entire process of bankruptcy cases provided by the embodiment of this application. Figure 1 , the implementation environment may include a terminal 101 and a system 102.

[0037] Terminal 101 is connected to system 102 via a wireless or wired network. Optionally, terminal 101 is a smartphone, tablet computer, laptop computer, desktop computer, etc., but is not limited thereto. Terminal 101 is installed and runs an application that supports digital collaborative management of the entire bankruptcy process and multiple early warning systems.

[0038] System 102 is an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. System 102 can provide background services for applications running on the first terminal 101.

[0039] The following is an explanation of the digital collaborative management and multi-dimensional early warning system for the entire bankruptcy process proposed in this application. Figure 2The system includes an acquisition module 201 , a first determination module 202 , a second determination module 203 and a third determination module 204 .

[0040] Acquisition module 201 is used to obtain stage description information, disposal information, and disposal reference information for each disposal stage of the target bankruptcy case. Disposal information includes reference enterprise information, institution-specific information, and disposal methods. First determination module 202 is used to determine stage reference information and associated object information based on relevant information about the disposal stage. Second determination module 203 is used to calculate disposal risk information. Third determination module 204 is used to select disposal stages that require early warning.

[0041] Among them, the acquisition module refers to a processing unit for collecting multi-source data, which can be implemented specifically by a distributed data interface, and can be connected to heterogeneous data sources such as court systems and industrial and commercial databases through standardized protocols to solve the problem of inconsistent data formats. Stage description information refers to the characteristics of the disposal stage described in natural language, which can be implemented specifically by a semantic parsing engine to convert unstructured text into a computable stage identifier. Institutional proprietary information refers to the internal business data of the disposal agency, which can be implemented specifically by an encrypted transmission channel, and data security is guaranteed by an authority classification mechanism. Disposal reference information refers to a collection of external public data, which can be implemented specifically by web crawler technology to regularly capture public information sources such as corporate credit information disclosures and judicial judgments. Of course, in the embodiments of the present application, the acquisition of relevant information and data needs to strictly comply with the requirements of relevant laws and regulations.

[0042] Specifically, the acquisition module 201 first establishes a data collection channel, for example, it obtains the stage progress data of the court case management system through the API interface, and at the same time connects to the enterprise credit information disclosure system to obtain the related party change records. The first determination module 202 performs semantic analysis on the stage description information, for example, matching the keywords of the "asset valuation" stage with the auction announcement in the disposal reference information, and screening out the transaction records of similar assets as stage reference information. The second determination module 203 identifies abnormal disposal behaviors that deviate from routine operations by constructing a risk assessment matrix, for example, comparing the disposal method with the historical case library. The third determination module 204 sets a dynamic threshold. For example, when the risk score of a certain stage exceeds the average value of similar cases, an early warning is triggered, prompting the judges to focus on reviewing the materials of that stage.

[0043] Compared to related technologies, traditional systems can only process structured data from a single source, while this solution enables the fusion of multimodal data. While related technologies rely on manual comparison of publicly available information, this solution automatically links phase characteristics with external data through semantic matching. Existing risk assessment models are limited to single-phase analysis, while this solution establishes a cross-phase risk transmission analysis mechanism. For example, if an abnormal deregistration of an affiliated enterprise is discovered during the debt confirmation phase, ownership verification records can be automatically traced back to the asset inventory phase.

[0044] Through the above-mentioned technical solution, this application achieves dynamic integration of data from the entire bankruptcy process, improving cross-institutional collaboration efficiency. By establishing a mechanism for automatically linking stage characteristics with external information, it effectively identifies deviations from the rationality of disposal methods and promptly identifies potential risks that are transmitted across stages. For example, during the property distribution stage, the system can automatically match shareholder change records in the industrial and commercial system, providing early warning of potential asset transfer risks and assisting judges in taking preemptive preservation measures.

[0045] The present application further proposes a first determination module 202, which is used to determine the stage reference information of the case handling stage from the handling reference information for any case handling stage among multiple case handling stages based on the stage description information, organization-specific information and handling method of the case handling stage, the stage description information is used to describe the handling stage in natural language, and the stage reference information is used to assist in judging the rationality of the handling method of the case handling stage; based on the stage description information, organization-specific information and reference enterprise information, the associated object information of the case handling stage is determined from the handling reference information, the associated object information is the object information of the associated object associated with the target enterprise under the case handling stage, and the associated object includes at least one of an associated enterprise, an associated individual and an associated group.

[0046] Phase description information refers to the business scenario characteristics of the current disposal phase, described in natural language. This can be achieved by converting unstructured text into structured semantic features using text encoding technology, providing a semantic benchmark for information screening. Institutional proprietary information refers to the business rule data generated by the disposal agency during a specific disposal phase. This can be achieved by accessing structured field data through the agency's internal database interface, constraining the business logic boundaries of information screening. Disposal method refers to the business operation plan adopted during the current disposal phase. This can be achieved by converting it into a quantifiable and evaluable operational feature vector using process modeling technology, used to match adaptive reference information. Phase reference information refers to decision-making support data that matches the business scenario and characteristics of the disposal agency during the current disposal phase. This can be achieved through a dual filtering mechanism of semantic similarity calculation and business rules, used to verify the compliance and feasibility of the disposal method. Related object information refers to entity data that has a potential relationship with the target enterprise during a specific disposal phase. This can be achieved through enterprise graph construction and association path analysis technology, used to identify stakeholders and potential risk sources. Reference enterprise information refers to the target enterprise's operating-related data in a non-bankruptcy state. This can be achieved by collecting structured data through the enterprise credit information disclosure system or commercial database, and is used to supplement the missing related relationships in the bankruptcy case disposal stage.

[0047] Specifically, during the bankruptcy case disposal phase, natural language processing technology is first used to semantically encode the phase description information and extract the core business elements of the disposal phase. Institutional proprietary information is then feature-extracted using a pre-set business rule library to generate the institutional business logic constraints. The disposal method is converted into a feature vector containing dimensions such as operational steps, execution entities, and resource requirements. The screening process for phase reference information utilizes a dual mechanism of semantic similarity calculation and rule matching: the semantic model calculates the correlation between the disposal reference information and the phase description information, the business rule engine verifies its compatibility with the institutional proprietary information, and ultimately determines the valid reference information through weighted scoring. The identification of associated object information is achieved by constructing a dynamic association graph: a basic enterprise relationship network is generated based on reference enterprise information, and transaction records, equity relationships, and other data in the institutional proprietary information are superimposed. A graph neural network algorithm is used to mine potential association paths across data sources, ultimately outputting a list of associated entities that meet the characteristics of the current disposal phase.

[0048] Compared with related technologies, traditional methods rely on manual experience to filter reference information from isolated data sources, resulting in highly subjective screening criteria and incomplete coverage of related relationships. This solution establishes an information screening mechanism that synergizes semantic features with business rules to achieve automated and precise matching of reference information. By building a relationship graph based on multi-source data fusion, it overcomes the limitations of relationship identification based on a single data source. For example, during the claim declaration stage, it can simultaneously integrate shareholder information from corporate business data, debtor related party information from court execution data, and transaction information from tax data to form a complete relationship object identification network.

[0049] Through the above technical solution, this application effectively solves the problem of reference information screening bias caused by information fragmentation. Through the dual constraint mechanism of semantics and rules, the accuracy of the screening of reference information in the stage is improved to a level that can support automated decision-making; at the same time, it breaks through the data source limitations of traditional related object identification methods, and can discover hidden related transaction parties in the asset inventory stage, and identify falsely declared related creditors in the debt review stage, thereby improving the standardization of bankruptcy case handling and risk prevention and control capabilities.

[0050] The present application further proposes a first determination module 202, which is used to determine the initial reference information of the case handling stage from the handling reference information based on the stage description information of the case handling stage, where the initial reference information is information that matches the stage description information; based on the stage description information and the organization-specific information, determine multiple first reference information and multiple second reference information from the initial reference information, where the first reference information is information that matches the organization-specific information, and the second reference information is information that does not match the organization-specific information, and the number of the first reference information and the second reference information is determined based on the confidence of the organization-specific information; based on the handling method, determine the stage reference information from the multiple first reference information and the multiple second reference information.

[0051] Among them, stage description information refers to the feature set that describes the case disposal stage in natural language. Specifically, semantic coding technology can be used to convert text descriptions into vector features to achieve this, and it is used to establish semantic associations with disposal reference information. Institutional proprietary information refers to the proprietary data set formed by the disposal agency in the process of performing its duties. Specifically, a structured database can be used to store the agency's performance records and historical case experience data to provide a screening basis from the agency's perspective. Confidence refers to a quantitative indicator of the credibility of institutional proprietary information. Specifically, historical data verification accuracy or expert scoring mechanism can be used to achieve this, and it is used to dynamically adjust the quantitative ratio of the first reference information to the second reference information. Disposal method semantic features refer to the semantic representation of disposal methods such as asset valuation plans and debt recognition methods. Specifically, deep learning models can be used to extract feature vectors of text operation points to achieve deep matching between disposal plans and reference cases.

[0052] Specifically, first, the stage description text is converted into a high-dimensional semantic vector through semantic coding technology, and similarity matching is performed in the disposal reference information database to screen out the initial reference information set. Subsequently, the confidence index is calculated based on the historical performance data in the institution's proprietary information, such as the success rate data of the manager in handling similar cases in the past. When the confidence is higher than the preset threshold, for example, it reaches above 85%, 70% of the first reference information is extracted from the initial reference information; when the confidence is lower than the threshold, for example, in the range of 60%-85%, the proportion of the first reference information is linearly reduced to 50% according to the confidence. Finally, the disposal method semantic matching model is used to compare the operational points of the disposal plan with the disposal cases in the first and second reference information, and the top N items with the highest similarity are selected as the stage reference information.

[0053] Compared with related technologies, traditional reference information screening methods rely solely on keyword matching or single-institution experience data, failing to dynamically adjust the credibility weight of institutional information, which can easily lead to information overfitting or omission of valid cases. This solution, however, utilizes a dynamic confidence adjustment mechanism to retain high-credibility institutional experience while introducing cross-institutional reference information as a supplement, effectively avoiding information screening bias. Furthermore, related technologies often use rule engines to determine disposal method matching, making it difficult to handle complex semantic associations. This solution, however, employs semantic feature vector matching to capture the implicit operational logic within disposal plans.

[0054] Through the above technical solution, this application solves the problem of imbalanced reference information credibility caused by differences in institutional information quality. It balances the use of institutional proprietary information and external reference information through a dynamic confidence matching mechanism. It also breaks through the limitations of traditional keyword matching and leverages semantic features to achieve deep adaptation between disposal methods and reference cases, ensuring that the selected reference information is both consistent with the characteristics of the current disposal stage and can effectively support the implementation of specific disposal plans.

[0055] The present application further proposes a first determination module 202, which is used to semantically encode the stage description information to obtain the semantic features of the stage description, and determine the initial reference information from the disposal reference information based on the semantic features of the stage description; determine the reference organization-specific information with a satisfactory degree of correlation from the organization-specific information based on the stage description information, and divide the initial reference information into the first initial reference information and the second initial reference information according to the reference organization-specific information; determine the quantity ratio of the first reference information and the second reference information based on the confidence of the reference organization-specific information; semantically encode the disposal method to obtain the semantic features of the disposal method, and filter the stage reference information from the first reference information and the second reference information through semantic feature matching.

[0056] Among them, the semantic features of the stage description refer to the vectorized representation of the stage description text converted through the natural language processing model. Specifically, the BERT model can be used for semantic encoding to capture the business logic semantics of the disposal stage. Reference institution proprietary information refers to institutional data whose business relevance to the current disposal stage exceeds a preset threshold. Specifically, this can be achieved by calculating the semantic similarity between institutional information and stage description information to ensure that institutional data with substantial business relevance is screened out. Confidence grading processing refers to dynamically adjusting the amount of reference information based on the source credibility of institutional data. Specifically, a credibility coefficient weighted algorithm can be used to complement the historical data of high-credibility institutions with the innovative models of low-credibility institutions. The semantic features of the disposal method refer to the vectorized representation of the disposal plan text. Specifically, an LSTM network can be used for sequence modeling to evaluate the operational adaptability of the reference information and the disposal plan.

[0057] Specifically, during the bankruptcy property investigation phase, the phase description information is semantically encoded and then matched against disposal reference information such as industrial and commercial registration and litigation announcements to identify initial reference information related to asset inventory. By calculating the correlation between court execution records and the property investigation phase, proprietary information from reference institutions with asset freezing experience is identified. Based on the credibility coefficient of court data, the initial reference information is divided into a first initial reference information containing historical freezing cases and a second initial reference information involving new asset concealment models. Semantic features of the disposal methods are then vector-space matched against these two types of reference information, ultimately identifying phase reference information that complies with both conventional freezing procedures and incorporates new asset tracking methods.

[0058] Compared with related technologies, traditional approaches rely on keyword matching to filter reference information, failing to identify the underlying business logic of the disposal phase. Furthermore, their use of static organizational data results in insufficient information relevance. This solution constructs a multidimensional feature space through semantic encoding, enabling dynamic mapping of business scenarios and reference information during the disposal phase. This approach also leverages confidence grading to preserve the dual value of high-credibility disposal experience and innovative models. While related technologies use rule-based matching between disposal methods and reference information, this solution achieves operational-level solution adaptation through semantic feature matching.

[0059] Through the above technical solution, this application effectively solves the problem of inaccurate reference information screening caused by semantic matching deviation, avoids interference from irrelevant data through the correlation screening of institutional proprietary information, balances the reference value of historical experience and innovative models through confidence grading processing, and semantic matching of disposal methods ensures that the screened information is operational, thereby improving the decision-making support accuracy in the bankruptcy case disposal stage.

[0060] This application further proposes a first determination module 202, which is used to determine the first association map of the target enterprise based on the organization's proprietary information, determine the second association map of the target enterprise based on the reference enterprise information, and determine the associated object information of the case disposal stage from the disposal reference information in combination with the stage description information.

[0061] Specifically, proprietary information refers to the relationship data stored within the disposal agency. This can be achieved through proprietary data sources such as corporate shareholder change records and administrator performance records in the court case management system, and is used to construct a relationship network reflecting the agency's internal perspective. Reference enterprise information refers to the relationship data publicly available from enterprises. This can be achieved through public data sources such as industrial and commercial registration information and listed company announcements, and is used to construct a relationship network reflecting an external perspective. The first relationship graph refers to the relationship network structure generated based on proprietary information. This can be achieved through graph database technology, connecting enterprise nodes and related object nodes through edges such as equity relationships and transaction relationships, and is used to represent the relationships understood by the disposal agency. The second relationship graph refers to the relationship network structure generated based on reference enterprise information. This can be achieved through data crawling technology to obtain public data and construct multi-layer relationships. This is used to supplement relationships not covered by proprietary information. Stage description information refers to the characteristics of the disposal stage described in natural language. This can be achieved through the stage name and task description text recorded in the case management system, and is used to provide semantic constraints for the selection of related objects.

[0062] Specifically, during the claim declaration stage, the undisclosed related-party transaction data between the target enterprise and its suppliers is first extracted from the administrator's performance records to construct a first association graph containing implicit associations. At the same time, the target enterprise's publicly disclosed foreign investment information is obtained through the enterprise credit information disclosure system to construct a second association graph. The two graphs are imported into the graph computing engine for node alignment and edge fusion to form a comprehensive network covering explicit and implicit associations. Based on the "claim verification" keyword in the current stage description information, the supplier nodes and guarantor enterprise nodes related to accounts payable are screened out from the comprehensive network as related object information. Through the dual data source construction and dynamic screening mechanism, the information loss of a single data source is avoided, and it is ensured that the identification of related objects meets the core needs of the current disposal stage.

[0063] Compared with related technologies, traditional methods rely solely on the disposal agency's internal data or a single public data source to construct an association map, resulting in the omission of implicit associations or delayed updates to public information. This solution achieves complementary verification of internal proprietary data and external public data through the collaborative processing of institutional proprietary information and reference enterprise information. For example, it cross-validates the undisclosed related-party transaction data held by the administrator with the outbound investment data registered with the Industrial and Commercial Bureau, effectively identifying implicit related parties formed through equity holding. At the same time, the dynamic screening mechanism of stage description information overcomes the disconnection between static association maps and business scenarios, enabling the identification of related objects to accurately match the core needs of different disposal stages.

[0064] Through the above technical solution, this application solves the problem of incomplete relationship identification caused by the dispersion of related object information and realizes the dynamic integration of related relationships across data sources. During the asset disposal stage, it can simultaneously capture the bidder related information and public equity structure information held by the auction agency, and identify the bidding entities that circumvent related relationships through multi-layer shareholding; during the creditor's rights review stage, it can integrate the transaction flow data and litigation announcement information in the administrator's performance record to discover undeclared hidden related creditors. This dual data fusion mechanism increases the completeness of related object identification from 62% for a single data source to 89% for a dual data source, and the matching accuracy of the screened related object information with the current disposal stage reaches 93%.

[0065] The present application further proposes a first determination module 202, which is used to obtain first association information of the target enterprise from the organization's proprietary information, obtain second association information based on the reference enterprise information, use the second association map to expand or adjust the first association map to form a target association map, and determine the associated object information based on the stage description information and the target association map. Technical solution.

[0066] Among them, the first related information refers to the related relationship data confirmed by legal procedures and stored within the disposal agency. Specifically, it can be achieved by parsing structured documents such as court rulings and administrator performance reports, and is used to construct a core related graph with legal effect. The second related information refers to the related relationship data obtained through public channels. Specifically, it can be achieved by using web crawler technology to capture equity relationship data from corporate disclosure systems and credit information platforms, and is used to supplement potential related clues that are not covered within the agency. The construction of the target related graph is based on the nodes and edges of the first related graph. The newly added nodes in the second related information are associated with the existing nodes through the entity alignment algorithm. For example, when a guarantee relationship that has not been recorded by the agency appears in the second related information, a new guarantee relationship edge is added while ensuring the integrity of the original graph structure.

[0067] Specifically, during the asset disposal phase, a court-confirmed list of affiliated companies is first extracted from the administrator's working papers to form a first association graph. Simultaneously, the target company's outbound investment records are obtained through the industrial and commercial information query interface to form a second association graph. Using an entity name fuzzy matching algorithm, newly discovered shareholding companies in the second association graph are linked to existing nodes in the first association graph, forming a fused graph containing implicit associations. Based on the semantics of the "asset valuation" phase description, the fused graph screens for affiliated parties holding mortgages on the target company's fixed assets, automatically generating a list of related parties requiring focused review.

[0068] Compared to related technologies, traditional methods rely solely on the internal records of the disposal agency and are unable to identify hidden related parties not reflected in judicial documents. For example, during the property investigation phase, a situation where an affiliated company indirectly controls the target company's assets through a proxy holding agreement may be overlooked because it is not disclosed in the bankruptcy documents. This solution integrates external public data to build an extended map, which can effectively identify such hidden relationships formed through nested equity and contractual control.

[0069] The present application further proposes a second determination module 203, which is used to determine the disposal method risk information for any case disposal stage among multiple case disposal stages based on the institution-specific information, stage reference information and disposal method of the case disposal stage; determine the associated object risk information based on the institution-specific information and associated object information of the case disposal stage; and determine the disposal risk information of the case disposal stage based on the disposal method risk information and the associated object risk information.

[0070] Institutional proprietary information refers to the internal specifications or historical operational data used by the disposal agency during the case disposal phase. This information can be obtained through the administrator's performance records or court review standards, and is used to assess the alignment of the current disposal method with the agency's operational specifications. Phase reference information refers to historical cases or industry standards related to the current disposal phase. This information can be obtained by matching the descriptive information of similar case disposal phases with public databases, and is used to generate a basis for judging the rationality of the disposal method. Disposal method risk information refers to the degree of deviation between the current disposal method and the reference disposal method. This information can be obtained by calculating the similarity between the disposal method and the reference disposal method using a semantic feature matching algorithm, and is used to identify procedural violations or decision-making biases. Related party information refers to information on entities with vested interests in the target enterprise. This information can be obtained through enterprise equity relationship maps or transaction record analysis, and is used to identify potential related party risks. Related party risk information refers to the risk of interest transfer or asset transfer that may arise from related parties during the disposal phase. This information can be obtained by cross-validating the related party's credit rating with the agency's review standards, and is used to quantify related party moral hazard.

[0071] Specifically, during the asset valuation and disposal phase, proprietary information is extracted and compared with historical asset disposal cases in the reference information, generating multiple reference disposal methods that comply with the institution's operational specifications. The actual asset valuation plan and these reference disposal methods are semantically encoded for similarity. Disposal method risk information is generated when the similarity falls below a threshold. Simultaneously, the transaction data of related enterprises in the related party information is matched with the related party review standards in the proprietary information. If abnormal high-frequency transactions are detected in related enterprises and fail to pass the review standards, related party risk information is generated. These two types of risk information are combined using a weighted fusion algorithm to form a comprehensive disposal risk value. When this value exceeds a preset threshold, an alert is triggered.

[0072] Compared to related technologies, traditional approaches only assess risk based on a single dimension of data. For example, they may only verify the superficial compliance of disposal procedures while ignoring the risks of related-party networks, or rely solely on mechanical comparisons of historical case libraries. This solution establishes a dual-path risk analysis model for disposal methods and related objects. This model not only verifies the alignment of disposal actions with institutional regulations and historical experience, but also dynamically monitors abnormal behavior patterns within the related-party network, achieving a three-dimensional identification of complex risks during the disposal phase.

[0073] Through the aforementioned technical solution, this application accurately identifies procedural risks arising from disposal methods that deviate from institutional norms or historical best practices, while also effectively capturing potential hidden asset transfers by related parties during the asset disposal process. Through independent analysis and comprehensive assessment of these two risk sources, the application avoids the misjudgment of complex inter-interest scenarios by a single risk assessment model, significantly improving the coverage and accuracy of risk warnings during the bankruptcy case disposal phase.

[0074] The present application further proposes a second determination module 203 for generating multiple reference disposal methods based on stage reference information and institution-specific information, and determining disposal method risk information based on multiple reference disposal methods and disposal methods.

[0075] The stage reference information refers to historical treatment mode data related to the current case treatment stage selected from the treatment reference information. Specifically, a semantic matching algorithm can be used to extract treatment records of similar cases at similar stages from the public information library to provide industry benchmark data. The institution-specific information refers to the operation experience parameters of the treatment institution at a specific case treatment stage. Specifically, feature extraction can be performed through the job records in the historical case database of the institution to reflect the professional characteristics of the institution. The reference treatment method refers to the standardized treatment scheme generated by combining the stage reference information and the institution-specific information. Specifically, a machine learning model can be used to fuse and calculate historical treatment modes and institutional operation characteristics to form a treatment scheme set that takes into account industry commonality and institutional characteristics. The treatment method risk information refers to the deviation degree quantitative index of the actual treatment scheme from the reference treatment scheme. Specifically, it can be achieved by calculating the similarity difference of the treatment method and the reference treatment method at key operation nodes to identify abnormal operation risks.

[0076] Specifically, in the asset disposal stage, the system first extracts similar enterprise asset valuation schemes from public judicial documents as stage reference information, and obtains the asset disposal strategies adopted by the institution in the past from the administrator's job database as institution-specific information. Through a neural network model, the two types of data are fused to generate a reference set containing treatment methods such as auction, agreement transfer, and debt-to-equity. Then, the actual asset split auction scheme is compared with the schemes in the reference set in multiple dimensions, such as target valuation method, bidder screening rules, and delivery time nodes. When there is a significant difference between the actual scheme and the reference scheme in key dimensions, the system automatically marks that the treatment method has an operation risk. For example, if the target valuation method does not use the industry-standard income approach but uses the cost approach, and there are no similar cases in the institution's historical operations, a risk warning is triggered.

[0077] Compared with related technologies, the traditional method only relies on single experience to judge the reasonableness of the treatment method, such as evaluating the asset disposal scheme only according to the subjective experience of the administrator, lacking objective data support. However, the present scheme builds a dynamically updated reference treatment method set, which not only avoids the subjective bias of pure artificial experience, but also overcomes the defect that single historical data cannot adapt to the characteristics of the institution. For example, in the debt claim stage, related technologies cannot automatically identify whether the unconventional debt registration method is compliant, while the present scheme can effectively find the debt registration operation that does not comply with the regulations by fusing the reference method generated by the court announcement data and the institution's operation records.

[0078] Through the above technical solution, this application can accurately identify potential risks arising from disposal methods that deviate from industry standards or institutional operating practices, such as discovering that the asset penetration verification of related companies has not been carried out according to standard procedures during the property investigation phase. This solution, through a dual data verification mechanism, addresses the misjudgment problem caused by traditional methods due to the lack of multi-dimensional reference standards, thereby improving the accuracy and interpretability of risk identification.

[0079] The present application further proposes a second determining module 203 for acquiring associated object proprietary information from the organization proprietary information based on the associated object information, and determining associated object risk information based on the associated object proprietary information.

[0080] Among them, related object information refers to the entity information that has an association relationship with the target enterprise during the case handling stage. Specifically, association graph technology or data matching algorithms can be used to extract the identification information and transaction records of related enterprises and related individuals from public information to establish an association network between the target enterprise and external entities. Among them, institution-proprietary information refers to the proprietary data accumulated by the disposal agency in the course of performing its duties. Specifically, the administrator's performance records, internal risk assessment indicators, and historical case processing data can be obtained through the authority control mechanism to provide internal risk assessment dimensions. Among them, related object proprietary information refers to the data fragments in the institution's proprietary information that are directly related to the related object. Specifically, entity recognition technology can be used to structure the institution's proprietary information, and through name matching and unified social credit code comparison, accurate mapping of related objects and proprietary information can be achieved.

[0081] Specifically, during the bankruptcy case disposal phase, a multi-dimensional search is first performed within the institution's proprietary information database, using information such as the names of related companies and legal representatives contained in the related party information. For example, if the related party is a supplier, the system automatically retrieves proprietary data such as the supplier's performance evaluation records and asset freeze status from the institution's historical cases. Subsequently, based on a pre-set risk assessment model, key indicators in the related party's proprietary information are compared with risk thresholds. For example, if a related party has a history of multiple debt defaults, the system automatically generates risk information for the related party, indicating that the related party may affect the fairness of asset disposal.

[0082] Compared to related technologies, traditional methods rely solely on publicly available business information or judicial documents for risk assessment, failing to access proprietary data accumulated within the disposal agency. For example, while an affiliated company may have no public record of default, the agency may have engaged in malicious bidding collusion in past cases. This information is only stored in the agency's proprietary database. This solution effectively identifies these hidden risks by breaking down the data barriers between associated entity information and the agency's proprietary information, avoiding risk assessment biases caused by information fragmentation.

[0083] By the technical solution, the application realizes accurate identification of the risk of the associated object, and solves the problem of one-sidedness of the traditional method due to the inability to obtain special data of the disposal mechanism. By extracting the historical behavior data directly related to the associated object in the special information of the mechanism, potential moral risk, operational risk and other non-public risk factors of the associated party can be found, and more comprehensive risk warning basis can be provided for the formulation of the asset disposal plan.

[0084] The application further proposes a third determination module 203 for determining a reference risk score based on the disposal risk information of each case disposal stage, and determining the disposal stage of the target bankruptcy case requiring early warning as the disposal stage of the target bankruptcy case requiring early warning when the reference risk score is greater than or equal to a preset risk score.

[0085] The reference risk score refers to a quantitative evaluation index of the risk degree of the case disposal stage, which can be calculated by using a risk score model in combination with multi-dimensional risk factors in the disposal risk information, for example, the disposal mode risk information and the associated object risk information are superimposed according to the weight to generate a comprehensive score. The index realizes horizontal comparison of cross-stage risks through a unified quantitative standard, and solves the problem of non-uniform risk dimensions in the traditional method.

[0086] The preset risk score refers to a critical threshold for triggering an early warning, which can be set to different values according to the case type and the risk preference of the disposal mechanism by using a dynamic threshold configuration module, for example, a threshold higher than that of the creditor's claim stage is set for the asset disposal stage. The threshold serves as an objective judgment criterion, and through the configurability, the differentiated early warning demand is met, and the subjective bias of manual experience judgment is eliminated.

[0087] Specifically, for the case disposal stages such as case filing review and property investigation in the whole process of the bankruptcy case, first, the disposal risk information such as the disposal mode compliance risk and the associated party conflict of interest risk of each stage is extracted. Then, the multi-dimensional risk information is converted into a numerical reference risk score by using a risk quantification model, for example, the asset concealment risk in the property investigation stage is given a score of 0-100. The system automatically compares the scores of each stage with the preset threshold, and selects the stages with scores exceeding the threshold as the high-risk target disposal stages. For example, in the asset valuation stage, when the associated transaction risk score reaches the preset threshold of 70 points, an early warning is triggered to prompt the administrator to check the transaction records of the associated party.

[0088] Compared with the related art, the traditional method relies on manual experience to qualitatively judge the risk of the disposal stage, and cannot dynamically quantify the risk degree of different stages, resulting in lagging identification of high-risk stages. The present scheme realizes real-time processing and objective judgment of risk data by establishing a risk score quantification mechanism and an automatic threshold screening mechanism.

[0089] By the technical solution, the application solves the problem of insufficient early warning accuracy caused by the lack of risk dynamic assessment in the disposal stage of the bankruptcy case. Through automatic comparison of the quantitative risk score and the preset threshold, real-time identification of the high-risk disposal stage is realized. For example, in the experimental process, when the associated creditor's declaration is found to be abnormal in the creditor verification stage, the system completes the risk scoring and triggers the early warning within 10 seconds.

[0090] The above is only an optional embodiment of the application and is not used to limit the application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the application shall be included in the protection scope of the application.

Claims

1. A digital collaborative management and multi-dimensional early warning system for the entire bankruptcy process, characterized by: The system comprises: an acquisition module, configured to acquire stage description information, disposal information, and disposal reference information of a target bankruptcy case of a target enterprise at each case disposal stage, wherein the disposal information includes reference enterprise information of the target enterprise used in the corresponding case disposal stage, proprietary information of the disposal agency in the corresponding case disposal stage, and the disposal method in the corresponding case disposal stage. The disposal reference information is publicly available information related to the target enterprise and the target bankruptcy case; a first determination module configured to determine, for any one of the multiple case handling stages, stage reference information of the case handling stage from the handling reference information based on the stage description information, institution-specific information, and handling method of the case handling stage, wherein the stage description information is used to describe the case handling stage in natural language, and the stage reference information is used to assist in determining the rationality of the handling method of the case handling stage; and to determine, from the handling reference information, associated object information of the case handling stage based on the stage description information, the institution-specific information, and reference enterprise information, wherein the associated object information is object information of associated objects associated with the target enterprise at the case handling stage, wherein the associated objects include at least one of associated enterprises, associated individuals, and associated groups; a second determination module configured to determine, for any case handling stage among the multiple case handling stages, handling method risk information based on the institution-specific information, stage reference information, and handling method of the case handling stage; determine associated object risk information based on the institution-specific information and associated object information of the case handling stage; and determine handling risk information of the case handling stage based on the handling method risk information and the associated object risk information; The third determination module is used to determine at least one target bankruptcy case disposal stage from multiple case disposal stages of the target bankruptcy case based on the disposal risk information of each case disposal stage, and the target bankruptcy case disposal stage is the case disposal stage that requires early warning.

2. The system according to claim 1, wherein: The first determining module is configured to determine, based on the stage description information of the case handling stage, initial reference information of the case handling stage from the handling reference information, wherein the initial reference information is information matching the stage description information; Based on the stage description information and the institution-specific information, determining a plurality of first reference information and a plurality of second reference information from the initial reference information, wherein the first reference information is information that matches the institution-specific information, and the second reference information is information that does not match the institution-specific information, and the number of the first reference information and the second reference information is determined based on the confidence level of the institution-specific information; Based on the handling manner, the phase reference information is determined from the plurality of first reference information and the plurality of second reference information.

3. The system according to claim 2, characterized in that The first determining module is configured to perform semantic encoding on the stage description information to obtain a stage description semantic feature of the stage description information; determining, based on the semantic features of the stage description, initial reference information of the case handling stage from the handling reference information; The first determining module is configured to determine reference institution-specific information from the institution-specific information based on the stage description information, wherein a correlation degree between the reference institution-specific information and the case handling stage is greater than or equal to a preset correlation degree; determining a plurality of first initial reference information and a plurality of second initial reference information from the initial reference information based on the reference organization-specific information; determining the plurality of first reference information from the plurality of first initial reference information and determining the plurality of second reference information from the plurality of second initial reference information based on the confidence level of the reference institution-specific information in the confidence level of the institution-specific information; The first determining module is configured to perform semantic coding on the disposal method to obtain a disposal method semantic feature of the disposal method; The disposal method semantic feature is used to match the plurality of first reference information and the plurality of second reference information to obtain the stage reference information.

4. The system according to claim 1, wherein: The first determining module is configured to determine a first association graph of the target enterprise based on the institution-specific information, wherein the first association graph is configured to represent associations between the target enterprise and other objects; Determining a second association graph of the target enterprise based on the reference enterprise information, where the second association graph is used to represent association relationships between the target enterprise and other objects; Based on the stage description information, the first association map, and the second association map, the associated object information of the case handling stage is determined from the handling reference information.

5. The system according to claim 4, characterized in that The first determining module is configured to obtain a first association graph of the target enterprise from the institution-specific information, wherein the first association graph is configured to represent an association relationship between the target enterprise and other objects; The first determining module is configured to perform a query based on the reference enterprise information to obtain a second association graph of the target enterprise, wherein the second association graph is configured to represent an association relationship between the target enterprise and other objects; The first determination module is used to use the second association map to expand or adjust the first association map to obtain the target association map of the target enterprise, and the first association map is used as a reference for expansion or adjustment; based on the stage description information and the target association map, the associated object information of the case disposal stage is determined from the disposal reference information.

6. The system according to claim 1, wherein: The second determination module is configured to generate a plurality of reference disposal methods based on the stage reference information and the institution-specific information; and determine disposal method risk information based on the plurality of reference disposal methods and the disposal method.

7. The system according to claim 1, wherein: The second determining module is configured to obtain the associated object proprietary information from the organization proprietary information based on the associated object information; Based on the associated object specific information, the associated object risk information is determined.

8. The system according to claim 1, wherein: The third determination module is configured to determine a reference risk score for each case handling stage based on the handling risk information of each case handling stage; The case handling stage with a reference risk score greater than or equal to a preset risk score among the multiple case handling stages is determined as the target bankruptcy case handling stage to obtain the at least one target bankruptcy case handling stage.

Images