A data collaboration method and device based on a unified state diagram

Abstract

The present disclosure relates to the technical field of enterprise collaborative office management, in particular to a data collaboration method and device based on a unified state diagram, which generates a task skeleton corresponding to a document directory structure by analyzing structured directory information of a to-be-processed document, and constructs a unified state diagram containing a content anchor point identifier, a task state identifier, a permission state identifier and a knowledge context field based on the task skeleton, so that the directory position, the task processing state, the permission control state and the related knowledge context in the document can be associated and expressed in the same state node, thereby making the collaborative processing process no longer simply rely on manual synchronization or loose coupling interface association between a document system, a task system or a knowledge base system, and improving the consistency, linkage and traceability between multiple types of collaborative states.

Description

Technical Field

[0001] This disclosure relates to the field of enterprise collaborative office management technology, specifically to a data collaboration method and apparatus based on a unified state diagram. Background Technology

[0002] In the field of enterprise-level digital collaboration and process management, documents such as reports, plans, and review materials are typically not only the carriers of business content but also important bases for task decomposition, collaborative processing, and process advancement. In existing technologies, functions such as document editing, task management, access control, and knowledge management are usually implemented separately by different systems or functional modules, and are linked through manual operation, API calls, or rule synchronization.

[0003] In actual collaboration, due to the fragmented maintenance of document content, task progress, collaboration permissions, and related contextual information, it is difficult to maintain a unified, real-time, and traceable state relationship between different systems or modules. On the one hand, changes in document content are difficult to reflect in the corresponding task processing in a timely and accurate manner, and task status often still relies on manual judgment and manual updates. On the other hand, collaboration information generated around the specific content of a document is difficult to continuously bind with the corresponding tasks and processing procedures, which can easily lead to problems such as state fragmentation, context breakage, and redundant communication.

[0004] Therefore, how to improve the consistency, linkage, and traceability among multiple collaborative states in a document-driven collaborative processing scenario has become an urgent problem to be solved. Summary of the Invention

[0005] This disclosure provides a data collaboration method and apparatus based on a unified state diagram, which at least solves the problem in related technologies that, in document-driven collaborative processing, the lack of a unified state management mechanism among document content changes, task processing progress, collaborative permission control, and related context information leads to lag in cross-object state synchronization and insufficient traceability of the collaborative process.

[0006] Firstly, this disclosure provides a data collaboration method based on a unified state diagram, including: The structured directory information of the document to be processed is parsed, and a task skeleton corresponding to the directory structure of the document to be processed is generated based on the hierarchical relationship in the structured directory information. A unified state graph is constructed based on the task skeleton. The unified state graph includes unified state nodes and connecting edges. The unified state node contains at least a content anchor identifier, a task state identifier, a permission state identifier, and a knowledge context field to form a four-tuple basic data structure. The content anchor identifier is used to associate the unified state node with the corresponding directory location in the document to be processed. The knowledge context field is used to record knowledge entry references or difference description information related to the task state identifier. The connecting edges are used to represent the document hierarchy relationship and / or task dependency relationship between the unified state nodes. Listen for content operation events on the document to be processed, determine the target unified state node based on the correlation between the operation position of the content operation event and the content anchor point identifier, and generate candidate state transitions for the target unified state node. Based on the task state distribution of multiple unified state nodes in the unified state graph, the graph state parameters characterizing the disorder of the state distribution in the unified state graph are calculated. Predict the expected graph state parameters after executing the candidate state transition, and determine whether to trigger the task state identifier flow of the target unified state node based on the amount of change between the expected graph state parameters and the current graph state parameters; When the change quantity meets the preset ordering condition, the task status identifier is triggered to flow, and the permission status identifier and knowledge context field of the target unified status node are updated or maintained according to the flowed task status identifier; when the change quantity does not meet the preset ordering condition, the corresponding state transition is blocked, and the knowledge context field of the target unified status node is updated or maintained according to the blocking result.

[0007] Secondly, this disclosure provides a data collaboration device based on a unified state diagram, comprising: The document parsing and skeleton generation module is used to parse the structured directory information of the document to be processed, and generate a task skeleton corresponding to the directory structure of the document to be processed based on the hierarchical relationship in the structured directory information. A state graph construction module is used to construct a unified state graph based on the task skeleton. The unified state graph includes unified state nodes and connecting edges. The unified state node includes at least a content anchor identifier, a task state identifier, a permission state identifier, and a knowledge context field to form a four-tuple basic data structure. The content anchor identifier is used to associate the unified state node with the corresponding directory position in the document to be processed. The knowledge context field is used to record knowledge entry references or difference description information related to the task state identifier. The connecting edges are used to represent the document hierarchy relationship and / or task dependency relationship between the unified state nodes. The state transition decision module is used to monitor content operation events on the document to be processed, determine the target unified state node based on the association between the operation position of the content operation event and the content anchor point identifier, and generate candidate state transitions for the target unified state node; calculate graph state parameters characterizing the disorder of the state distribution of the unified state graph based on the task state distribution of multiple unified state nodes in the unified state graph; predict the expected graph state parameters after executing the candidate state transitions, and determine whether to trigger the task state identifier transition of the target unified state node based on the change between the expected graph state parameters and the current graph state parameters; when the change meets the preset ordering condition, trigger the transition of the task state identifier, and update or maintain the permission state identifier and knowledge context field of the target unified state node based on the transitioned task state identifier; when the change does not meet the preset ordering condition, block the corresponding state transition, and update or maintain the knowledge context field of the target unified state node based on the blocking result.

[0008] Thirdly, this disclosure provides an electronic device, including: a memory and a processor, which are communicatively connected to each other. The memory stores computer instructions, and the processor executes the computer instructions to perform the aforementioned data coordination method based on a unified state diagram.

[0009] Fourthly, this disclosure provides a computer-readable storage medium storing a computer program for causing a computer to execute the aforementioned data coordination method based on a unified state diagram.

[0010] Fifthly, this disclosure provides a computer program product, including a computer program for causing a computer to execute the aforementioned data collaboration method based on a unified state diagram.

[0011] In the data collaboration method and apparatus based on a unified state diagram in the above embodiments of this disclosure, firstly, a task skeleton corresponding to the document's directory structure is generated by parsing the structured directory information of the document to be processed. Based on the task skeleton, a unified state diagram containing content anchor identifiers, task status identifiers, permission status identifiers, and knowledge context fields is constructed. This allows the directory location, task processing status, permission control status, and related knowledge context in the document to be associated and expressed in the same state node, thereby making the collaborative processing process no longer solely dependent on manual synchronization or loosely coupled interface associations between the document system, task system, or knowledge base system. Secondly, by listening to the content operation events of the document to be processed, and determining the target unified state node based on the association between the operation position of the content operation event and the content anchor identifier, candidate state transitions are generated. Furthermore, the graph state parameters are calculated by combining the task state distribution of multiple unified state nodes in the unified state diagram, and the change in graph state parameters before and after the candidate state transition is executed determines whether to trigger the task state transition. This allows the task state update to be adapted to the actual document operation process and the overall state distribution of the unified state diagram.

[0012] Finally, synchronously updating or maintaining the permission status identifier and knowledge context field when triggering task status transitions can improve the consistency of linkage between document location, task transition, permission control and knowledge context maintenance, thereby reducing the risks of status fragmentation, manual synchronization delay and collaborative context breakage in document-driven collaborative processing, and improving the accuracy and traceability of collaborative status management. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the specific embodiments of this disclosure or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0014] Figure 1 This is a flowchart illustrating a data collaboration method based on a unified state diagram in one embodiment; Figure 2 This is a flowchart illustrating the dynamic task routing based on execution complexity and domain classification in one embodiment. Figure 3 This is a flowchart illustrating the structured accumulation of knowledge item push and collaboration processes in one embodiment; Figure 4 This is a schematic diagram of a process for compliance verification and collaborative task distribution based on an enterprise knowledge base in one embodiment; Figure 5This is a flowchart illustrating the process of generating subtasks based on annotation events and linking them to task states in another embodiment. Figure 6 This is a schematic diagram of a data collaboration device based on a unified state diagram in one embodiment; Figure 7 This is a diagram of the internal structure of an electronic device in one embodiment. Detailed Implementation

[0015] The technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. The described embodiments are only some embodiments of this disclosure, and not all embodiments. Other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without creative effort are all within the protection scope of this disclosure.

[0016] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this disclosure, unless otherwise stated, "a plurality of" means two or more.

[0017] Document-driven collaborative processing technology, as an important technical means for enterprise digital office, project management, and process control, has been widely applied in various business scenarios such as report preparation, solution review, audit material preparation, and project deliverable management. In actual collaboration, documents typically not only carry the business content itself, but also link multiple types of collaborative information such as task assignment, responsible person's processing progress, collaboration permissions, historical data, and business knowledge.

[0018] Typically, collaborative processing methods for document-related tasks fall into the following categories: One method involves using a separate document editing system to write and annotate the document content, followed by project members manually creating, assigning, or updating tasks corresponding to document chapters in the task management system. Another method involves synchronizing some operation records from the document system to the task management system via interface synchronization or rule configuration to assist in generating task reminders or progress records. Additionally, separate permission management and knowledge base systems can be used to manage document access permissions, reference materials, historical cases, or business specifications, allowing collaborators to query and access relevant materials independently when processing tasks.

[0019] However, the document-driven collaborative processing methods in related technologies generally suffer from the following drawbacks: 1. When document systems and task management systems are used separately, there is no direct correlation between the actual modification status of document content and the task progress status. The task status usually relies on manual judgment and updates. When a document chapter has been modified, supplemented, or reviewed, the status in the task management system may still remain at the original stage, resulting in delayed and subjective task progress feedback, making it difficult to reflect the true business processing status in a timely manner.

[0020] 2. When using interface synchronization or rule configuration for association, it is usually only possible to establish relatively coarse-grained associations based on document titles, document links, operation logs, or preset event rules. It is difficult to continuously bind specific chapters, paragraphs, and annotation locations in the document to the corresponding task status and the responsible person's processing process. When document content is repeatedly modified, multiple people process it in parallel, or annotation opinions are circulated multiple times, the collaborative context is easily scattered across document annotations, instant messaging, task notes, and historical versions, resulting in high costs for subsequent tracking and review.

[0021] 3. When using independent permission management and knowledge base management, permission control, knowledge resource retrieval, and task processing progress are usually independent of each other. Collaborators may need different editing, reviewing, or annotation permissions at different task stages, and may also need to access historical data, business specifications, or experience records related to the current document content. However, existing methods mostly rely on manual permission configuration and manual retrieval of knowledge resources, making it difficult to continuously associate and dynamically maintain these permissions as the task progresses. This can easily lead to problems such as delayed permission adjustments, scattered knowledge references, and difficulty in accumulating collaborative experience.

[0022] 4. In complex document collaboration scenarios, a document often includes multiple chapters and tasks, and multiple responsible parties. Different tasks may have sequential dependencies, review dependencies, or content associations. Related technologies typically update the status based on a single task or operation, lacking unified management of the overall relationship between multiple task states. This makes it difficult to determine the impact of a change in a particular state on the overall collaboration process, easily leading to inconsistencies between the progress of local tasks and the overall workflow.

[0023] Therefore, how to manage document content location, task processing status, collaboration permissions, and related context information in a unified, continuous, and traceable manner during document-driven collaborative processing, in order to improve the consistency of linkage among multiple types of collaborative information, has become an urgent technical problem to be solved.

[0024] Based on this, embodiments of this disclosure provide a data collaboration method and apparatus based on a unified state diagram.

[0025] The data collaboration method based on a unified state graph provided in this disclosure can be applied to a document-driven enterprise collaborative processing system. This enterprise collaborative processing system may include a document editing terminal, a collaborative processing terminal, a task processing terminal, a knowledge index, and one or more adapters. The document editing terminal can be a web-based online document editor, a client-side document plugin, or a business system interface with document editing capabilities. The collaborative processing terminal receives content operation events reported by the document editing terminal and performs directory parsing, state mapping, workflow decision-making, permission updates, knowledge retrieval, compliance verification, and result write-back. The task processing terminal can be an independent business processing terminal, a professional application system, or an external task execution service, used to handle complex tasks and return processing results to the collaborative processing terminal. The knowledge index stores enterprise knowledge entries, personal knowledge entries, and graph data instances derived from completed collaborative processes. Adapters may include document adapters, task adapters, permission adapters, and knowledge adapters, used to communicate with different types of document systems, task systems, permission systems, or knowledge bases.

[0026] It is understood that this paragraph is used to describe the applicable operating environment of the method disclosed herein, and does not limit the data structure and state transition rules of the unified state diagram.

[0027] Example 1 This embodiment provides a data collaboration method based on a unified state diagram. This embodiment corresponds to a higher-level overall processing flow, used for unified state management of the content location, task status, permission status, and knowledge context of the document to be processed. For example... Figure 1 As shown, the method can be executed by the collaborative processing end, including steps S101 to S106.

[0028] Step S101: Parse the structured directory information of the document to be processed, and generate a task skeleton corresponding to the directory structure of the document to be processed based on the hierarchical relationship in the structured directory information.

[0029] In this embodiment, the document to be processed can be a report, plan, review material, regulatory submission material, project delivery material, or other document that requires collaborative processing by multiple people.

[0030] Structured directory information can include heading text, heading level, heading number, chapter path, paragraph identifier of the heading, and parent-child hierarchy between headings.

[0031] The collaborative processing end can generate a task skeleton based on the above hierarchical relationship. The skeleton nodes in the task skeleton correspond to the directory positions in the document to be processed, and the skeleton nodes at least record the content anchor point, initial task status and parent-child relationship corresponding to the directory position.

[0032] It should be noted that step S101 in this embodiment only requires generating a task skeleton corresponding to the directory structure based on the hierarchical relationship in the structured directory information. It does not require the use of specific natural language processing tools, nor does it require the determination of complexity, domain classification, or specific responsible person in this step.

[0033] The implementation details of title semantic analysis, task type inference, expected deliverable extraction, and suggested responsible person matching will be further explained in subsequent embodiments.

[0034] Step S102: Construct a unified state diagram based on the task skeleton.

[0035] In this embodiment, the unified state graph can be an attribute graph data structure maintained in memory or persistent storage layer, rather than a simple business process description.

[0036] The collaborative processing end creates a unified state node for multiple skeleton nodes in the task skeleton, and creates connection edges based on the hierarchical relationship and task dependency relationship between the skeleton nodes.

[0037] For example, a unified state diagram can be represented as:

[0038] in, Represents a unified set of state nodes. Represents the set of connecting edges. This represents a node attribute mapping function, used to map nodes of uniform state to corresponding sets of node attributes. This represents the edge attribute mapping function, which is used to map connecting edges to the corresponding set of edge attributes.

[0039] For any unified state node Its basic four-tuple data structure can be represented as:

[0040] in, This serves as a content anchor identifier, used to link unified state nodes. Establish a link with the corresponding directory location in the document to be processed. Content anchor tags can identify content locations such as chapters, paragraphs, tables, images, charts, annotation objects, or character ranges.

[0041] This is a task status identifier used to indicate the status of the task corresponding to this content anchor. This is a permission status identifier used to indicate the permission status of the area corresponding to the content anchor point, such as edit, read-only, locked, or annotation-only.

[0042] This is a knowledge context field used to record information related to... Relevant knowledge entry citations or difference description information may include one or more of the following: knowledge entry ID, knowledge fragment summary, source identifier, search score, difference description information, verification basis, or collaborative accumulation identifier.

[0043] Connecting edges are used to represent document hierarchy relationships and / or task dependencies between nodes in a unified state. As one implementation method, connecting edges... It can be represented as:

[0044] in, As the initial unified state node, To achieve a unified state node for the target, For the connection edge type, This refers to the edge attribute. It can include document hierarchy edges and task dependency edges. Document hierarchy edges are used to represent parent-child or sibling chapter relationships in the directory tree; task dependency edges are used to represent preceding tasks, subsequent tasks, review dependencies, parallel collaboration dependencies, or blocking dependencies. This can include dependency type, dependency strength, dependency source, generation time, and the context fragments in which the dependency was generated.

[0045] Step S103: Listen for content operation events of the document to be processed, determine the target unified state node based on the correlation between the operation position of the content operation event and the content anchor point identifier, and generate candidate state transitions for the target unified state node.

[0046] In this embodiment, content operation events may include edit events, save events, submit events, review events, rollback events, annotation events, or annotation resolution events. After capturing a content operation event, the document editing terminal can report the event type, operation location, operation object, operator identifier, event time, and additional data to the collaborative processing terminal. The collaborative processing terminal determines the target unified status node corresponding to the content operation event based on the association between the operation location and each content anchor point identifier.

[0047] Candidate state transitions are used to represent the task state changes that may be triggered by a content operation event. Candidate state transitions can be represented as:

[0048] in, Represents the target unified state node. This indicates the current task status identifier of the target unified status node. Indicates the status of the candidate task. This indicates the content operation event that triggered the transition of this candidate state. This indicates the logical or system time corresponding to the event.

[0049] Step S104: Based on the task state distribution of multiple unified state nodes in the unified state graph, calculate the graph state parameters that characterize the disorder of the state distribution in the unified state graph.

[0050] In this embodiment, the graph state parameter is used to reflect the overall distribution of multiple task state identifiers in the unified state graph.

[0051] The graph state parameters can be state distribution dispersion, state distribution disorder, state concentration, completion distribution index, or other parameters that can characterize the overall orderliness of the unified state graph. This embodiment does not limit the specific calculation formulas for the graph state parameters.

[0052] Step S105: Predict the expected graph state parameters after the execution of the candidate state transition, and determine whether to trigger the task state identifier flow of the target unified state node based on the amount of change between the expected graph state parameters and the current graph state parameters.

[0053] In this embodiment, the collaborative processing end can temporarily replace the task state identifier of the target unified state node with the candidate task state identifier without actually submitting the candidate state transition, and recalculate the expected graph state parameters of the unified state graph. If the current graph state parameters are... The expected state parameters are: The change can then be expressed as:

[0054] The collaborative processing end is based on the change amount Determine whether the candidate state transition meets the preset ordering conditions, whether it is conducive to the ordering of the unified state diagram, or whether it is within the allowable range.

[0055] Step S106: When the change quantity meets the preset ordering condition, the task status identifier is triggered to flow, and the permission status identifier and knowledge context field of the target unified status node are updated or maintained according to the flowed task status identifier; when the change quantity does not meet the preset ordering condition, the corresponding status transition is blocked, and the knowledge context field of the target unified status node is updated or maintained according to the blocking result.

[0056] In this embodiment, when the task status identifier is triggered to change, the collaborative processing terminal will transfer the target unified status node's... Update the current state to a candidate state, and determine whether an update is needed based on the task state after the transition. .

[0057] For example, when a task enters the review state, the permission status can be updated to "annotate only"; when the task status reverts to the in-progress state, the permission status can be restored to "editable". The knowledge context field can record the knowledge entries, reference information, or reasons for the status change.

[0058] When the change does not meet the preset ordering conditions, the collaborative processing end does not submit the candidate state transition and writes the blocking reason to the output. .

[0059] For example, reasons for blocking may include incomplete prerequisite tasks, dependent nodes still being blocked, unclosed annotations, unprocessed knowledge verification discrepancies, or state distribution disorder exceeding the allowable range.

[0060] In the data collaboration method and apparatus based on a unified state diagram in the above embodiments of this disclosure, firstly, a task skeleton corresponding to the document's directory structure is generated by parsing the structured directory information of the document to be processed. Based on the task skeleton, a unified state diagram containing content anchor identifiers, task status identifiers, permission status identifiers, and knowledge context fields is constructed. This allows the directory location, task processing status, permission control status, and related knowledge context in the document to be associated and expressed in the same state node, thereby making the collaborative processing process no longer solely dependent on manual synchronization or loosely coupled interface associations between the document system, task system, or knowledge base system. Secondly, by listening to the content operation events of the document to be processed, and determining the target unified state node based on the association between the operation position of the content operation event and the content anchor identifier, candidate state transitions are generated. Furthermore, the graph state parameters are calculated by combining the task state distribution of multiple unified state nodes in the unified state diagram, and the change in graph state parameters before and after the candidate state transition is executed determines whether to trigger the task state transition. This allows the task state update to be adapted to the actual document operation process and the overall state distribution of the unified state diagram. Finally, synchronously updating or maintaining the permission status identifier and knowledge context field when triggering task status transitions can improve the consistency of linkage between document location, task transition, permission control and knowledge context maintenance, thereby reducing the risks of status fragmentation, manual synchronization delay and collaborative context breakage in document-driven collaborative processing, and improving the accuracy and traceability of collaborative status management.

[0061] Example 2 This embodiment, based on Embodiment 1, further defines the data structure of the unified state node. The lower-level scheme provided in this embodiment can independently support the implementation of a six-tuple data structure formed on the basis of a four-tuple data structure.

[0062] In this embodiment, on top of the quadruple basic data structure, the unified state node also includes a logical clock identifier and a compliance status field to form a six-tuple data structure.

[0063] For example, for any unified state node Its extended six-tuple data structure can be represented as:

[0064] in, This is a logical clock identifier used to represent a unified state node. The causal order in which state updates occur. Preferably, Lamport clocks, vector clocks, or other logical clocks that can express the causal order of events can be used.

[0065] Here, the Lamport clock increments with each local state update and, upon receiving a remote event, increments based on the larger of the local and remote clock values. The vector clock can then be represented as:

[0066] in, Indicates the number of endpoints or nodes participating in the collaboration. Indicates the first The count of an end or node in the relevant events of the current unified state node.

[0067] By comparing two vector clocks, it can be determined whether there is a causal relationship between two state update events, or whether they are concurrent events.

[0068] This is a compliance status field used to identify unified status nodes. The compliance verification status of associated task data.

[0069] Preferably, It can include , , , , Equal state values. Among them, "Checking" indicates that the verification has not yet been triggered, while "checking" indicates that the verification is in progress. This indicates that the verification has passed. This indicates that the verification failed. This indicates that the session was suspended due to a failed verification.

[0070] In this embodiment, , , and There is a causal relationship between them.

[0071] For example, when Depend on Circulation to hour, It can be constrained as or ; when Depend on Circulation to Before, It can be constrained to pass; when for hour, Constrained to not be transferred to , Information describing compliance differences needs to be recorded.

[0072] Furthermore, changes to identifiers or fields associated with the same state transition event are executed atomically within the same logical unit.

[0073] For example, for a "submit for review" event, the collaborative processing end can update within the same transaction. , , , and If any field fails to update, the status transition will be rolled back to avoid inconsistencies such as the task status having entered the review stage but permissions still being editable, or the verification status having been suspended but the task still showing as completed.

[0074] In the data collaboration method and apparatus based on unified state diagrams in the above embodiments of this disclosure, by adding a logical clock identifier and a compliance state field to the quadruple basic data structure and performing atomic control on changes to multiple fields, the causal consistency of state updates can be maintained in a distributed collaborative environment, and a cross-dimensional state interlocking mechanism can be formed between task flow, access control, knowledge context and compliance verification.

[0075] Example 3 This embodiment further defines the task skeleton generation process in step S101 based on embodiment one. Parsing the structured directory information of the document to be processed and generating the task skeleton may include steps S1011 to S1015.

[0076] Step S1011: Perform semantic analysis on the text of each level heading in the structured directory information to extract task attributes, and generate a task skeleton based on the hierarchical relationship and task attributes in the structured directory information.

[0077] In this embodiment, semantic analysis may include word segmentation, part-of-speech tagging, named entity recognition, dependency parsing, and phrase structure recognition.

[0078] Optionally, semantic analysis can be performed using SpaCy, HanLP, LTP, Stanza, or other natural language processing tools. Task attributes may include task type, expected deliverables, and suggested responsible party.

[0079] Step S1012: Perform dependency parsing on the title text to identify the syntactic structure type of the title text.

[0080] Specifically, syntactic structure types can include verb-object structures, subject-predicate structures, attributive-noun phrases, verbless noun phrases, and compound title structures. The collaborative processing unit can determine the syntactic structure type based on the core verb, object, core noun, modifiers, and parallel components in the title text.

[0081] Step S1013: In response to the syntactic structure type containing a verb-object structure, extract the core verb and object from the title text, match the core verb with the preset task type library to determine the task type in the task attributes, and extract the object as the expected deliverable in the task attributes.

[0082] For example, when the title text is "Verify data source", the core verb is "verify" and the object is "data source". The collaborative processing end can determine the task type as a verification task and set "data source" as the expected deliverable.

[0083] The preset task type library can maintain the mapping relationship between core verbs and task types. For example, "verify" and "check" are mapped to verification tasks, "compile" and "write" are mapped to writing tasks, "review" and "examine" are mapped to review tasks, and "summarize" and "organize" are mapped to summary tasks.

[0084] In step S1014, in response to a noun phrase with no verb as its syntactic structure type, the core noun in the title text is extracted as the expected deliverable, and the task type is inferred based on the semantic attributes of the expected deliverable or the preset default mapping rules.

[0085] For example, if the title text is "Risk Warning," then "Risk Warning" can be taken as the expected deliverable, and the task type can be inferred to be either a writing or review task based on the document content attributes corresponding to "Warning." If the title text is "System Implementation Status," then "System Implementation Status" can be taken as the expected deliverable, and the task type can be inferred to be either a reporting or explanatory task.

[0086] Step S1015: Match the expected deliverables with the preset organizational structure library to determine the suggested responsible person in the task attributes.

[0087] Specifically, the organizational structure database can include mappings between departments, positions, roles, personnel, business areas, keyword sets, and historical task processing records. The collaborative processing end can match keywords in expected deliverables with the role keyword mapping table in the organizational structure database.

[0088] For example, "data source" and "statistical scope" can be mapped to the data management role, "system implementation status" can be mapped to the system management role, and "equipment operation status" can be mapped to the equipment management role. The matching results can be written into the task skeleton node as a suggestion for the responsible person or role.

[0089] In this embodiment, the generated task skeleton can be represented as a tree data structure composed of multiple task skeleton nodes.

[0090] Each task skeleton node corresponds to a title node in the structured directory information, containing the chapter path, content anchor, task type, expected deliverables, suggested responsible person, initial task status, and parent-child relationship. The initial task status of the task skeleton can be... This indicates that the task has been planned according to the document structure, but has not yet been activated by content manipulation events.

[0091] In the data collaboration method and apparatus based on unified state diagrams in the above embodiments of this disclosure, by performing semantic analysis on the title text and using different task attribute extraction methods for verb-object structures and noun phrases, the document directory structure can be converted into a task skeleton with task type, expected deliverables, and suggested responsible persons, reducing the manual decomposition of document tasks and improving the accuracy of the correspondence between the task skeleton and the document directory structure.

[0092] Example 4 Based on Example 1, this embodiment further limits the generation conditions of candidate state transitions in step S103.

[0093] In this embodiment, the collaborative processing end generates candidate state transitions in response to content operation events meeting preset valid operation conditions.

[0094] Among them, the preset valid operation conditions may include: the content operation event is the first editing operation on the area corresponding to the content anchor point identifier of the target unified state node, and the content increment corresponding to the first editing operation exceeds the preset character number threshold.

[0095] Specifically, the collaborative processing end can maintain a content version summary and edit count for each unified state node. When a document editor reports a ContentChanged event, the collaborative processing end determines whether the operation position of the event falls within the target unified state node. The corresponding area. If this area has not been edited previously, and the content is increased in this update. Greater than the preset character count threshold Then the collaborative processing end generates by Circulation to The candidate state transitions. The content increment can be represented as:

[0096] in, This indicates the text length or effective content length of the area corresponding to the target content anchor point at the current moment. This indicates the text length or valid content length of the corresponding area in the previous version. This indicates the preset character count threshold.

[0097] Preferably, The limit can be configured based on the document type, for example, by setting it to 50 characters, or by setting other thresholds suitable for different business scenarios.

[0098] As one implementation, if the content operation event is only cursor movement, format adjustment, blank character input, or undoing an operation that did not produce valid content, then no candidate state transition is generated. If the content operation event is the first valid edit and the content increment reaches the threshold, then a candidate state transition is generated and the graph state parameter judgment process in Embodiment 1 is entered.

[0099] In the data collaboration method and apparatus based on a unified state diagram in the above embodiments of this disclosure, by setting effective operation conditions such as the first edit and the number of characters threshold, the frequent false triggering of the task state caused by cursor movement, input of a small number of invalid characters or format adjustment can be avoided, thereby improving the stability and accuracy of the mapping from content operation events to task state transitions.

[0100] Example 5 This embodiment, based on Embodiment 1, further defines the graph state parameters and preset ordering conditions. In this embodiment, the graph state parameter is the global information entropy, and the preset ordering conditions include entropy reduction conditions or entropy increase allowable conditions.

[0101] In this embodiment, the collaborative processing terminal counts the frequency of each task status identifier appearing in all unified status nodes. Let the total number of unified status nodes be... The task state set is ,state The number of nodes appearing in the unified state graph G is . Then the state Frequency of occurrence for:

[0102] Based on the frequency of occurrence of each task status identifier, the collaborative processing end uses the information entropy algorithm to calculate the global information entropy. :

[0103] Among them, when When that happens, the corresponding item is treated as 0. Used to characterize the disorder of the state distribution in a unified state diagram.

[0104] Generally speaking, when the task states are concentrated in a few states... A lower value indicates a more ordered graph state; when task states are scattered across multiple states and there are many blocked, backtracked, or crossover states, the graph is more ordered. A higher value indicates a more chaotic graph state.

[0105] The collaborative processing unit calculates the expected global information entropy after the candidate state transition. With the current global information entropy The difference between them is taken as entropy change :

[0106] when When less than or equal to zero, it indicates that the disorder of the state distribution in the unified state diagram does not increase after performing candidate state transitions, satisfying the entropy reduction condition. Greater than zero and less than or equal to the preset entropy increase tolerance threshold When this occurs, it indicates that although the transition of the candidate state increases the disorder of the state distribution, the increase is within an acceptable range, satisfying the entropy increase tolerance condition. Greater than If the collaborative processing terminal determines that the preset ordering conditions are not met, it will block the corresponding state transition.

[0107] In a specific example, the unified state graph contains , , , , Task status. When a chapter is... Enter If the transition conforms to the overall task progression pattern and does not introduce new blocking nodes, then It can be less than or equal to The collaborative processing end allows this transition. If a node is submitted without its prerequisites being completed... If this causes multiple subsequent nodes to become blocked or rollback, the expected global information entropy may increase significantly. The collaborative processing end will block the transition and record reasons such as incomplete prerequisite dependencies in the knowledge context field.

[0108] In the data collaboration method and apparatus based on a unified state graph in the above embodiments of this disclosure, by specifically limiting the graph state parameters to global information entropy, and judging whether the candidate state transitions satisfy the entropy reduction condition or the entropy increase allowable condition based on entropy change, the dispersed task state distribution in the document collaboration process can be converted into computable parameters, thereby improving the objectivity and interpretability of state transition decisions.

[0109] Example 6 This embodiment, based on embodiment three, further defines the task execution complexity assessment, domain classification, and dynamic routing.

[0110] In another possible implementation of step S1011 above, semantic analysis is performed on the text of each level of headings in the structured directory information to extract task attributes, and the method further includes: evaluating the execution complexity of the task based on the semantic analysis results and performing domain classification.

[0111] Specifically, task execution complexity It can be calculated using the following formula:

[0112] in, This indicates the length of the document segment corresponding to the target task, the depth of the heading level, or the estimated workload. This represents the dependency of the target task in the unified state graph. Indicates the number of responsible roles or collaborating departments involved. This indicates the knowledge gap or difficulty in knowledge retrieval in the corresponding field for this task. to These are preset weighting coefficients.

[0113] Domain classification can be determined based on title keywords, expected deliverables, historical task tags, or domain tags of knowledge entries, such as technology, auditing, compliance, engineering, production, data management, or general management.

[0114] Furthermore, in one possible implementation of the above embodiments, such as Figure 2 As shown, the method also includes: Step S201 involves constructing the unified state diagram and maintaining the task state identifiers in the collaborative processing terminal.

[0115] In other words, the creation, state transition, permission linkage, knowledge field update, and compliance status maintenance of task nodes are all uniformly maintained by the collaborative processing end, avoiding inconsistencies caused by multiple external task execution ends maintaining task status separately.

[0116] Step S202: Dynamically route tasks based on execution complexity and domain classification.

[0117] Specifically, the collaborative processing end can reduce the complexity of task execution. Compared with the preset complexity threshold Compare and combine domain classifications to determine the task processing method.

[0118] Step S203: When the execution complexity is determined to be lower than the preset complexity threshold, the corresponding collaborative processing operation is directly executed in the collaborative processing terminal.

[0119] For example, tasks such as confirming that the content is correct, confirming that the annotations have been resolved, updating the task status, or generating simple reminders can be handled directly by the collaborative processing terminal.

[0120] Step S204: When the execution complexity is determined to be greater than or equal to a preset complexity threshold, a dispatch request containing execution instructions and context parameters is generated, and the dispatch request is sent to a task processing end independent of the collaborative processing end, and the execution result fed back by the task processing end is asynchronously listened to to trigger the update of the task status identifier.

[0121] Specifically, a dispatch request may include a task ID, content anchor identifier, task type, expected deliverables, domain classification, execution complexity, input document fragment, reference knowledge entry, context parameters, callback address, and logical clock identifier. After the task processing end completes its processing, it returns the execution result, result version, processing log, and exception information to the collaborative processing end, which then generates or triggers candidate state transitions based on the returned results.

[0122] In the data collaboration method and apparatus based on a unified state diagram as described in the above embodiments of this disclosure, by simultaneously evaluating task complexity and domain classification during the semantic analysis phase, and by having the collaboration processing end uniformly maintain the task state while the task processing end executes complex tasks, a unified state can be maintained. Figure 1 While maintaining consistency, it also improves the scalability for handling complex tasks.

[0123] Example 7 This embodiment, based on embodiment five, further defines the blocking state analysis, critical path recommendation, and state synchronization health assessment.

[0124] In this embodiment, when the collaborative processing terminal detects that multiple unified state nodes in the unified state graph are in a blocked state and meet any of the following conditions: the global information entropy exceeds a preset global entropy threshold, or the increase in global information entropy within a preset time window exceeds a preset increase threshold, the collaborative processing terminal analyzes the connection edge dependencies between the unified state nodes in the blocked state, identifies the critical path causing the blockage, and recommends a state transition path to unblock the blockage based on the critical path.

[0125] Specifically, the collaborative processing end can extract the set of blocked nodes from the unified state diagram. And trace back the preceding nodes along the task dependency edges and document hierarchy edges.

[0126] For any path It can calculate path congestion weights. :

[0127] in, Representing a path The number of nodes that are currently in a blocked state. Indicates the depth of path dependency. This indicates the number of nodes subsequently affected by this path. to Preset weighting coefficients. The collaborative processing end can... The highest path is identified as the critical path, or... The set of paths that exceed a preset path threshold is identified as the set of critical paths.

[0128] Based on the critical path, the collaborative processing end can generate state transition paths to unblock the process. For example, if multiple chapters are blocked because a certain preceding annotation is unresolved, it is recommended to process the subtask corresponding to that annotation first; if multiple chapters are blocked because the same institutional basis is missing, it is recommended to supplement the corresponding institutional basis first; if multiple nodes are blocked because the same person in charge is overloaded, it is recommended to reallocate the tasks.

[0129] In this embodiment, the collaborative processing terminal also records the entropy change data corresponding to each state transition event to generate an entropy change curve.

[0130] Let the first The entropy corresponding to the next state transition becomes The state transition time is Then the entropy change curve can be expressed as: The time series that make up the time series.

[0131] The collaborative processing end can assess the state synchronization health of the unified state graph based on the slope of the entropy change curve. If the slope is negative or lower than the preset slope threshold within a continuous time window, it indicates that the unified state graph is becoming more ordered; if the slope is positive and exceeds the preset slope threshold within a continuous time window, it indicates that blocking, rollback, or parallel conflicts may increase, and the collaborative processing end can generate suggestions for process intervention, knowledge push, or task reassignment.

[0132] In the data collaboration method and apparatus based on a unified state graph in the above embodiments of this disclosure, by combining the connection edge dependency relationship and entropy change curve of the blocked node, the critical path causing the blockage can be identified from the overall level of the unified state graph, and suggestions for unblocking can be generated, thereby improving the state diagnosis capability in the process of complex document collaboration.

[0133] Example 8 This embodiment, based on Embodiment 1, further defines the knowledge push and structured accumulation of the knowledge context field. For example... Figure 3 As shown, this embodiment includes steps S301 to S303.

[0134] Step S301: In response to the task status identifier of the target unified status node being updated to the in progress state, extract the document context information associated with the content anchor identifier of the target unified status node, as well as the task attribute information recorded by the task skeleton or determined based on the structured directory information, and combine the document context information and task attribute information into a query vector.

[0135] Specifically, document context information may include chapter titles, parent headings, subheadings, current paragraph content, adjacent paragraph content, table titles, chart descriptions, and existing annotations. Task attribute information may include task type, expected deliverables, suggested responsible party, domain classification, and execution complexity. The collaborative processing end can use a text embedding model to encode the above information into query vectors. .

[0136] Step S302: Based on the query vector, perform semantic similarity retrieval in the knowledge index library, write knowledge entries with similarity higher than a preset threshold into the knowledge context field of the target unified status node and push them to the corresponding responsible person.

[0137] Suppose that the first [item] in the knowledge index base The vector of each knowledge item is Then query vector With knowledge item vectors The semantic similarity between them can be calculated using cosine similarity:

[0138] when Greater than the preset similarity threshold At that time, the collaborative processing end writes the corresponding knowledge item's identifier, summary, source, similarity score, and applicable instructions into the knowledge context field of the target unified status node, and presents it to the responsible person in the user interface in the form of an associated panel.

[0139] Preferably, a preset similarity threshold is used. The value can be set according to the size of the knowledge base and the business scenario, for example, between 0.65 and 0.85, or dynamically adjusted based on historical click-through rates or human feedback.

[0140] In this embodiment, the knowledge index base includes an enterprise knowledge base and a personal knowledge base.

[0141] A corporate knowledge base can include corporate archives such as internal policies, internal documents, internal news networks, work contact forms, instructions from leaders, requests for approval, template documents, audit question databases, and regulatory rule databases.

[0142] A personal knowledge base can include personal business experience, personal work records, historical data entry, personal collections, frequently used reference snippets, and task examples that the individual has handled. The personal knowledge base is dynamically updated based on task execution content and is primarily used for knowledge generation and output. The enterprise knowledge base is mainly used as reference input for compliance verification.

[0143] Therefore, the knowledge context field can both proactively push knowledge during task execution and form a structured accumulation after the task is completed.

[0144] Step S303: In response to the fact that the task status identifiers of a set of unified state nodes in the unified state graph have all reached the completed state and the associated documents have been marked as finalized, extract the state transition history of a set of unified state nodes, the knowledge entry reference records in the knowledge context field, and the document content fragments pointed to by the content anchor identifiers. Then, perform structured encapsulation according to the preset graph data pattern and store it in the knowledge index library.

[0145] The preset graph data pattern can include process instance nodes, task status nodes, document content fragment nodes, knowledge entry nodes, responsible person nodes, and connecting edges. Connecting edges can include relationships such as "originating from," "referenced," "resolved," "approved," and "belonging to the same process." The encapsulated collaborative process instance is written to the knowledge index library as a knowledge retrieval source for subsequent collaborative tasks.

[0146] In the data collaboration method and apparatus based on a unified state diagram in the above embodiments of this disclosure, by actively retrieving and writing the knowledge context field when the task enters the in-process state, the manual retrieval of knowledge materials by the responsible person can be reduced; by structurally encapsulating the state transition history, knowledge references and document content fragments after the document is finalized, a collaboration process can be precipitated into a subsequently searchable knowledge asset.

[0147] Example 9 This embodiment, based on Embodiment 2, further defines the compliance verification, verification suspension, and back-to-back collaborative task generation. For example... Figure 4 As shown, this embodiment includes steps S401 to S405.

[0148] Step S401: In response to the task status identifier flowing to the preset verification status, perform compliance verification on the task data associated with the target unified status node based on the preset enterprise knowledge base, and update the compliance status field according to the verification result.

[0149] In this embodiment, the preset verification status can be submission for review, pending completion, pre-completion, or pre-finalization. The input for compliance verification can include document content fragments associated with the target unified status node, task type, expected deliverables, compliance status field, enterprise knowledge base entries bound to the knowledge context field, and supplementary entries retrieved from the enterprise knowledge base.

[0150] Compliance verification can be performed by a verification rule engine. This engine can integrate keyword matching, semantic similarity, template version comparison, rule condition judgment, and coverage calculation. As one implementation method, the set of essential requirements related to the task in the enterprise knowledge base is... The set of covered requirements identified in the target document content is This requires coverage. It can be represented as:

[0151] The validation rule engine can also calculate the semantic similarity between document content and enterprise knowledge entries. And determine the hard rules Pass or fail. Overall verification score. It can be represented as:

[0152] in, , , For preset weighting coefficients, It can take the value 1 or 0, or it can be determined proportionally based on multiple rules. If Not lower than the preset compliance threshold And if there are no hard rules that fail, then the verification is considered successful; if Below If any hard rule fails, the verification is deemed unsuccessful.

[0153] Step S402: When the verification result is passed, update the compliance status field to the verification passed status.

[0154] Specifically, the collaborative processing end can Update to pass, and allow the corresponding task status to continue to progress to done or subsequent approval status.

[0155] Step S403: When the verification result is unsuccessful, update the compliance status field to the verification pending status, generate compliance difference description information, generate a collaborative task based on the compliance difference description information, and distribute the collaborative task to a new responsible person who is different from the original responsible person corresponding to the target unified status node.

[0156] Specifically, the collaborative processing end can Update to suspended or fail, output a blocking signal to prevent the task node from transitioning to the completed state, and append compliance difference description information to the knowledge context field.

[0157] The discrepancy description information may include missing requirements, potentially outdated templates, content inconsistent with company policies, uncovered knowledge items, and departments or responsible roles that need to be added. The new responsible person can be determined by the domain classification in the discrepancy description information, the department to which the company knowledge item belongs, the organizational structure database, and historical processing records.

[0158] Step S404: Set permission isolation flags for collaborative tasks, and control the new responsible person and the original responsible person to not see each other's submitted information based on the permission isolation flags.

[0159] Specifically, the original responsible person can only view the information they submitted and the task prompts that they are authorized to view, while the new responsible person can only view the collaborative task requirements and their own submitted information. The two cannot view each other's submitted information.

[0160] Step S405: Based on the permission isolation identifier, the initiator of the collaborative task is controlled to have the permission to view the content filled in by the original person in charge and the new person in charge, and in response to the initiator, the merge operation is performed on each content filled in, and the merged content is updated to the task data corresponding to the target unified status node.

[0161] The creator can be the lead person, the reviewer, or an administrator with merge permissions. After the creator performs the merge operation in the merge interface, the collaborative processing end writes the merged content into the document content or task data corresponding to the target unified status node, and records the merged version, the merger, the merge time, and the corresponding logical clock.

[0162] In the data collaboration method and apparatus based on a unified state diagram in the above embodiments of this disclosure, by performing compliance verification based on the enterprise knowledge base before the task is completed or submitted for review, problems such as incomplete content, outdated templates, or missing institutional basis can be intercepted before the task flow enters the completion state; by generating collaborative tasks with isolated permissions when the verification fails, multiple responsible persons can supplement content back-to-back, and the creator can merge them uniformly, thereby improving the integrity and security of collaborative filling of complex documents.

[0163] Example 10 This embodiment, based on embodiment nine, further limits the handling of high-frequency rollbacks and local graph chaos caused by failure of compliance verification.

[0164] In this embodiment, when the collaborative processing terminal detects that the target unified state node has failed the compliance verification, causing the task status identifier to roll back between different states, and the number of rollbacks of the task status identifier within the preset time window reaches the preset rollback number threshold, the local information entropy of the local graph structure to which the target unified state node belongs is calculated.

[0165] High-frequency rollback can be determined by the number of state rollbacks. Judgment. Let the preset time window be... ,exist Internal target unified state node from or Revert to previous state or The number of times ,when Greater than the preset backoff threshold At that time, the collaborative processing end determines that the node has a high frequency of rollback.

[0166] Local graph structure This can include the target unified state node, its parent node, child nodes, predecessor dependent nodes, successor dependent nodes, and nodes with knowledge reference relationships with it. Let the number of nodes within the local graph structure be... ,state The number of nodes appearing in the local graph structure is Then the local frequency for:

[0167] Local information entropy It can be represented as:

[0168] when Exceeding the preset disorder threshold At that time, the collaborative processing end extracts the document context information and compliance difference description information associated with the content anchor point identifier of the target unified status node, and retrieves semantically matching reference knowledge entries in the enterprise knowledge base and personal knowledge base based on the document context information and compliance difference description information. The reference knowledge entries are then written into the knowledge context field of the target unified status node and pushed to the current responsible person.

[0169] If the local information entropy of the local graph structure does not decrease within the preset time window, it indicates that the current responsible person may lack relevant domain knowledge, the prerequisite tasks have not been resolved, or the existing processing path cannot be unblocked. The collaborative processing end can generate state transition suggestions or task reallocation suggestions based on the domain classification of the reference knowledge items. For example, it can suggest first supplementing a certain corporate policy basis, first resolving a certain prerequisite annotation subtask, or assigning the task to a new responsible person whose domain matches the reference knowledge item.

[0170] In the data collaboration method and apparatus based on a unified state diagram in the above embodiments of this disclosure, by calculating the local information entropy of the local graph structure that frequently reverts due to compliance verification failure, local collaboration disorder can be identified from a single failure event. Furthermore, by combining the enterprise knowledge base and personal knowledge base, reference knowledge can be pushed or redistribution suggestions can be generated, thereby reducing the risk of repeated rework.

[0171] Example 11 This embodiment, based on Embodiment 1, further defines the generation of subtasks when the content operation event is an annotation event. For example... Figure 5 As shown, this embodiment includes steps S501 to S503.

[0172] Step S501: In response to an annotation event triggered within the area corresponding to any content anchor point in the document to be processed, extract the annotation location information, annotation content, and mentioner identifier corresponding to the annotation event, and determine the target unified state node based on the association between the annotation location information and each content anchor point identifier.

[0173] Specifically, annotation location information may include paragraph ID, element ID, table cell position, chart object ID, start and end character offsets, or document path. Annotation content may include natural language descriptions and mentioner identifiers.

[0174] For example, the annotation could be something like "@Zhang San, please verify the source of this data".

[0175] Step S502: Based on the annotation content and annotation location information, generate subtasks belonging to the target unified status node, and assign the subtasks to the responsible persons corresponding to the mentioner identifiers.

[0176] Specifically, the title of a subtask can be generated based on the annotation content and location information, such as "Verify the data source of a certain paragraph". The task description of the subtask can retain the original annotation text, the task anchor can point to the original annotation location information, and the responsible person can be set to the user corresponding to the mentioner identifier. The subtask is bound to the original annotation, content anchor, and target unified status node. The responsible person can process it in the task panel or reply in the document annotation. The reply content and processing status are synchronously written to the unified status graph.

[0177] Step S503: When all subtasks associated with the target unified status node are marked as resolved, update the task status identifier of the target unified status node from the audited status to the completed status.

[0178] Specifically, the collaborative processing end can statistically analyze the status of each subtask under the target unified status node. If all forced subtasks are... or Then the collaborative processing end generates by Circulation to The candidate state transitions are processed, and state transition decisions are made according to the method of Example 1 or Example 5. If there are still unclosed annotations or unresolved subtasks, the state transition is blocked and the reason is recorded in the knowledge context field.

[0179] In the data collaboration method and apparatus based on a unified state diagram in the above embodiments of this disclosure, by converting document annotation events into subtasks bound to specific content anchors, the review comments, responsible persons, and processing status within the document can form a structured closed loop, avoiding the difficulty in tracking annotation comments which are scattered between the document and the task system.

[0180] Example 12 This embodiment, based on embodiment one, further limits the permission linkage between tasks and documents and the document finalization process.

[0181] In this embodiment, in response to the task status identifier of the target unified status node being changed to the preset review status, the collaborative processing terminal switches the permission status identifier of the area corresponding to the content anchor identifier of the target unified status node in the document to be processed to retain only the annotation permission.

[0182] Specifically, the preset review status can be: , or .when When the status is changed to the preset review status, the collaborative processing terminal calls the document system interface through the permission adapter to... Corresponding area Updated to or .

[0183] Preferably, this permission switching can be reflected in the interface control of the document editing terminal, such as disabling editing operations for the corresponding chapter and only allowing reviewers to add comments. As a safety net, the backend can also perform a final verification based on the permission status when the document is saved.

[0184] In response to the task status indicator of the target unified status node reverting from the review status to the in-process status, the collaborative processing end restores the permission status indicator of the target unified status node from the restricted editing status (only retaining annotation permissions) to the editable status. For example, when the reviewer returns the revised version, Depend on Back to , Depend on Restore to This allows the responsible party to continue making modifications.

[0185] When the task status indicators of all unified status nodes corresponding to the document to be processed meet the preset completion conditions, the collaborative processing end triggers the finalization or submission approval process for the document to be processed. The preset completion conditions may include all chapter task statuses being... or All mandatory annotations have been resolved; all compliance statuses are now met. The following conditions must be met: no blocking state nodes, global information entropy below the preset draft threshold, and completion of critical path tasks. When the final draft or approval process is triggered, the collaborative processing end can package the document content, unified state diagram, task execution traces, permission change records, knowledge reference records, and compliance verification records into a process submission data package.

[0186] In the data collaboration method and apparatus based on a unified state diagram in the above embodiments of this disclosure, by automatically switching local document permissions according to the task status, the document editing permissions can be kept consistent with the task flow stage; by triggering the finalization or submission for approval process when all unified state nodes meet the completion conditions, the process advancement driven by the deliverable status can be realized.

[0187] Example 13 This embodiment provides a data collaboration device based on a unified state diagram. For example... Figure 6 As shown, the device includes: The document parsing and skeleton generation module 61 is used to parse the structured directory information of the document to be processed and generate a task skeleton corresponding to the directory structure of the document to be processed based on the hierarchical relationship in the structured directory information. The state graph construction module 62 is used to construct a unified state graph based on the task skeleton. The unified state graph includes unified state nodes and connecting edges. The unified state node contains at least a content anchor identifier, a task state identifier, a permission state identifier, and a knowledge context field to form a four-tuple basic data structure. The content anchor identifier is used to establish an association between the unified state node and the corresponding directory position in the document to be processed. The knowledge context field is used to record the knowledge entry reference or difference description information related to the task state identifier. The connecting edges are used to represent the document hierarchy relationship and / or task dependency relationship between unified state nodes. The state transition decision module 63 is used to monitor content operation events of the document to be processed, determine the target unified state node based on the correlation between the operation position of the content operation event and the content anchor point identifier, and generate candidate state transitions for the target unified state node; calculate graph state parameters that characterize the disorder of the state distribution of the unified state graph based on the task state distribution of multiple unified state nodes in the unified state graph; predict the expected graph state parameters after executing the candidate state transitions, and determine whether to trigger the task state identifier transition of the target unified state node based on the change between the expected graph state parameters and the current graph state parameters; when the change meets the preset ordering conditions, the task state identifier transition is triggered, and the permission state identifier and knowledge context field of the target unified state node are updated or maintained according to the transitioned task state identifier; when the change does not meet the preset ordering conditions, the corresponding state transition is blocked, and the knowledge context field of the target unified state node is updated or maintained according to the blocking result.

[0188] The apparatus in this embodiment can be used to perform the steps described in any of the above method embodiments.

[0189] In one possible implementation of the above embodiments, on the basis of the quadruple basic data structure, the unified state node also includes a logical clock identifier and a compliance status field to form a six-tuple data structure. There is a causal relationship between the task status identifier, permission status identifier, knowledge context field, and compliance status field; Changes to identifiers or fields associated with the same state transition event in a six-tuple data structure are executed atomically within the same logical unit; Logical clock identifiers are incremented based on cooperative operation events and are used to maintain causal consistency of state updates among unified state nodes in a distributed environment.

[0190] In one possible implementation of the above embodiments, the document parsing and skeleton generation module 61 is specifically used to perform semantic analysis on the text of each level of headings in the structured directory information to extract task attributes, and generate a task skeleton based on the hierarchical relationship and task attributes in the structured directory information. Among them, the document parsing and skeleton generation module 61 is also specifically used to perform dependency parsing on the title text and identify the syntactic structure type of the title text; In response to the syntactic structure type containing verb-object structure, the core verb and object are extracted from the title text. The core verb is matched with the preset task type library to determine the task type in the task attributes, and the object is extracted as the expected deliverable in the task attributes. In response to noun phrases with no verbs in their syntactic structure, the core nouns in the title text are extracted as expected deliverables, and the task type is inferred based on the semantic attributes of the expected deliverables or the preset default mapping rules. Match the expected deliverables with a pre-defined organizational structure library to determine the recommended responsible party in the task attributes.

[0191] In one possible implementation of the above embodiments, the state transition decision module 63 is specifically used to generate candidate state transitions in response to content operation events satisfying preset valid operation conditions. The valid operation conditions include: the content operation event is the first editing operation on the area corresponding to the content anchor point identifier of the target unified state node, and the content increment corresponding to the first editing operation exceeds the preset character number threshold.

[0192] In one possible implementation of the above embodiments, the graph state parameter is the global information entropy, and the preset ordering conditions include entropy reduction conditions or entropy increase allowable conditions. The state transition decision module 63 is specifically used to count the frequency of each task's state identifier appearing in all unified state nodes. The global information entropy is calculated using the information entropy algorithm based on the frequency of occurrence of each task status identifier. The state transition decision module 63 is specifically used to calculate the difference between the expected global information entropy after executing the candidate state transition and the current global information entropy as the entropy change. When the entropy change is less than or equal to zero, or when the entropy change is greater than zero and less than or equal to the preset entropy increase allowable threshold, the preset ordering condition is determined to be met. When the entropy change exceeds the preset entropy increase tolerance threshold, it is determined that the preset ordering condition is not met.

[0193] In one possible implementation of the above embodiments, the document parsing and skeleton generation module 61 is further used to evaluate the execution complexity of the task based on the semantic analysis results and perform domain classification. The device also includes a dynamic routing module, which is used to perform the construction of the unified state graph and the maintenance of task state identifiers in the collaborative processing terminal. Dynamically route tasks based on execution complexity and domain classification; When the execution complexity is determined to be lower than the preset complexity threshold, the corresponding collaborative processing operation is directly executed in the collaborative processing terminal. When the execution complexity is determined to be greater than or equal to a preset complexity threshold, a dispatch request containing execution instructions and context parameters is generated and sent to a task processing end that is independent of the co-processing end. The task processing end also asynchronously listens for the execution result to trigger an update of the task status flag.

[0194] In one possible implementation of the above embodiments, the device further includes a blocking path analysis module, which is used to analyze the connection edge dependency relationship between the unified state nodes in the blocking state, identify the critical path causing the blocking, and recommend a state transition path to unblock based on the critical path when multiple unified state nodes in the unified state graph are detected to be in a blocked state and meet any of the following conditions: the global information entropy exceeds a preset global entropy threshold, or the increase of the global information entropy within a preset time window exceeds a preset increase threshold. Record the entropy change data corresponding to each state transition event to generate an entropy change curve, and evaluate the state synchronization health of the unified state graph based on the slope of the entropy change curve.

[0195] In one possible implementation of the above embodiments, the device further includes a knowledge retrieval and accumulation module, which is used to extract the document context information associated with the content anchor identifier of the target unified state node and the task attribute information recorded by the task skeleton or determined based on the structured directory information in response to the task status identifier of the target unified state node being updated to the in progress state, and combine the document context information and the task attribute information into a query vector. Based on the query vector, semantic similarity retrieval is performed in the knowledge index. Knowledge entries with similarity higher than a preset threshold are written into the knowledge context field of the target unified status node and pushed to the corresponding responsible person. The knowledge index base includes an enterprise knowledge base and a personal knowledge base. The personal knowledge base is dynamically updated according to the task execution content and is used for knowledge generation and push output. The enterprise knowledge base is used as a reference input for compliance verification. In response to the fact that the task status identifiers of a set of unified state nodes in the unified state graph have all reached the completed state and the associated documents have been marked as finalized, the state transition history of a set of unified state nodes, the knowledge entry reference records in the knowledge context field, and the document content fragments pointed to by the content anchor identifiers are extracted, structured and encapsulated according to the preset graph data pattern, and stored in the knowledge index library.

[0196] In one possible implementation of the above embodiments, the device further includes a compliance verification and collaborative distribution module, which is used to perform compliance verification on the task data associated with the target unified status node based on a preset enterprise knowledge base in response to the task status identifier flowing to a preset verification status, and update the compliance status field according to the verification result. When the verification result is passed, the compliance status field will be updated to the verification passed status. When the verification result is unsuccessful, the compliance status field is updated to the verification pending status, compliance difference description information is generated, and a collaborative task is generated based on the compliance difference description information. The collaborative task is then distributed to a new responsible person who is different from the original responsible person corresponding to the target unified status node. Set permission isolation markers for collaborative tasks, and based on the permission isolation markers, control the new person in charge and the original person in charge to not see each other's submitted information; Based on the permission isolation identifier, the initiator of the collaborative task is given the permission to view the content filled in by the original and new responsible persons. In response to the initiator, the initiator performs a merging operation on the content filled in, and updates the merged content to the task data corresponding to the target unified status node.

[0197] In one possible implementation of the above embodiments, the device further includes a local entropy-assisted decision-making module, which is used to calculate the local information entropy of the local graph structure to which the target unified state node belongs when it is detected that the task status identifier rolls back between different states due to the failure of compliance verification of the target unified state node, and the number of rollbacks of the task status identifier within a preset time window reaches a preset rollback number threshold. When the local information entropy exceeds the preset disorder threshold, extract the document context information associated with the content anchor identifier of the target unified state node and the compliance difference description information. Based on document context information and compliance difference description information, semantically matching reference knowledge entries are retrieved from the preset enterprise knowledge base and preset personal knowledge base. The reference knowledge entries are written into the knowledge context field of the target unified status node and pushed to the current responsible person. If the local information entropy of the local graph structure does not decrease within the preset time window, then state transition or task reallocation suggestions are generated based on the domain classification of the reference knowledge entries.

[0198] In one possible implementation of the above embodiments, the content manipulation events include annotation events; The device also includes an annotation subtask generation module, which is used to respond to annotation events triggered in the area corresponding to any content anchor point in the document to be processed, extract annotation location information, annotation content and mentioner identifier corresponding to the annotation event, and determine the target unified state node based on the association between the annotation location information and each content anchor point identifier. Based on the annotation content and annotation location information, subtasks belonging to the target unified status node are generated, and the subtasks are assigned to the responsible persons corresponding to the mentioner identifier; When all subtasks associated with the target unified status node are marked as resolved, the task status identifier of the target unified status node will be updated from the review status to the completed status.

[0199] In one possible implementation of the above embodiments, the device further includes a permission and delivery control module, which is used to switch the permission status identifier of the area corresponding to the content anchor identifier of the target unified status node in the document to be processed to retain only the annotation permission in response to the task status identifier of the target unified status node being changed to a preset review status. In response to the task status identifier of the target unified status node being rolled back from the review status to the in-process status, the permission status identifier of the target unified status node is restored from the restricted editing status with only annotation permissions to the editable status; When the task status indicators of all unified status nodes corresponding to the document to be processed meet the preset completion conditions, the finalization or submission approval process for the document to be processed is triggered.

[0200] It should be noted that the data collaboration device provided in the above embodiments is only illustrated by the division of the above program modules when implementing the corresponding data collaboration method. In practical applications, the above processing can be assigned to different program modules as needed, or multiple modules can be merged into one module to complete all or part of the processing described in the embodiments of this disclosure. In addition, the above device embodiments and method embodiments belong to the same concept, and their specific implementation process can be found in the limitations of the method embodiments above.

[0201] Example 14 This embodiment also provides an electronic device having the above-described features. Figure 6 The data collaboration device based on a unified state diagram is shown.

[0202] Figure 7 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present disclosure.

[0203] The following is a detailed reference. Figure 7 The diagram illustrates a structural schematic suitable for implementing an electronic device according to embodiments of the present disclosure. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 701, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 702 or a program loaded from memory 708 into random access memory (RAM) 703. The RAM 703 also stores various programs and data required for the operation of the electronic device. The processor 701, ROM 702, and RAM 703 are interconnected via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0204] Typically, the following devices can be connected to I / O interface 705: input devices 706 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 707 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 708 including, for example, magnetic tapes, hard disks, etc.; and communication devices 709. Communication device 709 allows electronic devices to exchange data via wireless or wired communication with other devices. Although Figure 7Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0205] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication device 709, or installed from memory 708, or installed from ROM 702. When the computer program is executed by processor 701, it performs the functions defined in the unified state diagram-based data collaboration method of embodiments of this disclosure.

[0206] Figure 7 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0207] This disclosure also provides a computer-readable storage medium in which the methods described in this disclosure can be implemented in hardware or firmware, or implemented as recordable on a storage medium, or implemented as computer code downloaded over a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and subsequently stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium may also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the unified state diagram-based data collaboration method shown in the above embodiments.

[0208] A portion of this disclosure can be applied to computer program products, such as computer program instructions, which, when executed by a computer, can invoke or provide methods and / or technical solutions according to this disclosure through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, and installation package files. Accordingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions; the computer compiling the instructions and then executing the corresponding compiled program; the computer reading and executing the instructions; or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0209] Optional application examples This embodiment provides an optional supplementary implementation to illustrate a further extension of the connection edges and state synchronization in the unified state graph. This embodiment does not change the basic technical solution of any of the foregoing embodiments one to twelve.

[0210] In one implementation, multi-source dependency reasoning can be introduced to supplement the connection edges in the unified state graph. The collaborative processing end can perform semantic dependency analysis on document titles, body text, and annotations to identify logical dependencies expressed by words such as "based on," "therefore," and "in summary"; it can perform behavioral dependency analysis on historical editing logs to identify temporal dependencies where a certain type of user typically processes another type of chapter only after completing another type of chapter; and it can also inject mandatory process dependencies based on project templates or process rules.

[0211] The aforementioned dependencies can be encapsulated as RichDependency objects. A RichDependency object can include a starting node, a target node, dependency type, dependency strength, and the context fragments generated by the dependency. When performing influence domain analysis and graph state parameter prediction, the state mapping engine can use RichDependency objects as input to improve the accuracy of determining the influence range of candidate state transitions.

[0212] In another implementation, an adaptive distributed synchronization mechanism can be set up. When there are network fluctuations between the collaborative processing end and the document editing end, task processing end, or external task system, the collaborative processing end can choose between immediate synchronization, lazy synchronization, and optimistic synchronization strategies based on network conditions, data criticality, and user activity. Immediate synchronization is used for critical states such as submission for review, compliance verification, and permission locking; lazy synchronization is used for knowledge entry browsing records or non-critical logs; and optimistic synchronization can be used for editing states with low probability of conflict, and then merged later through logical clocks and conflict resolution rules.

[0213] In a specific application example, after a user creates an annual report document, the collaborative processing end parses the document title style to obtain a raw node tree including chapters such as "Policy Implementation Status," "Project Implementation Status," "Data Verification Status," and "Risk and Rectification Status." The collaborative processing end performs semantic analysis on each title, identifies the task type and expected deliverables, and generates a task skeleton based on the organizational structure library. Subsequently, the collaborative processing end constructs a unified state diagram based on the task skeleton, creating a unified state node for each chapter and writing content anchor identifiers, task status identifiers, permission status identifiers, and knowledge context fields. When the responsible person first enters content exceeding a preset character threshold in the "Project Implementation Status" chapter, the document editing end captures the editing event and reports it to the collaborative processing end. The collaborative processing end determines the corresponding unified state node and generates a... arrive The system calculates the graph state parameters before and after the candidate state transitions. If the change satisfies the preset ordering conditions, the state transition is submitted, and historical cases, relevant regulations, and past personal information are retrieved from the knowledge index based on the document context and task attributes of the node. Knowledge entries with similarity higher than the threshold are written into the knowledge context field and pushed to the responsible person.

[0214] When the responsible person submits the chapter for review, the collaborative processing end changes the task status to review and switches the chapter's permissions to only allow annotations via the permission adapter. When the reviewer annotates a section with "Supplementary data source," the document editor captures the annotation event, and the collaborative processing end generates a subtask bound to the anchor point of that section and assigns it to the corresponding responsible person. After the responsible person completes the task and marks the annotation as resolved, the collaborative processing end generates subsequent candidate status transitions based on the annotation closure status.

[0215] When a chapter is about to enter the completion state, the collaborative processing end triggers a compliance check. The check rule engine extracts relevant policy requirements, template requirements, and historical audit issues from the enterprise knowledge base, calculating requirement coverage and semantic similarity. If it finds that the current content lacks a certain type of necessary evidence, and the original responsible person's submission cannot cover that requirement, the collaborative processing end prevents the task from entering the completion state, updates the compliance status to suspended, and generates a difference description. Simultaneously, the collaborative processing end generates a collaborative task based on the difference domain, distributes it to the new responsible person, and sets an access control barrier to ensure that the submissions from the new and original responsible persons are not visible to each other. The task creator can view and merge the submissions from both parties; the merged content is written to the corresponding document chapter, and then the compliance check is triggered again. After the check passes, the task status for that chapter transitions to the completion state.

[0216] When all chapter tasks in the document have reached the completion status, all mandatory annotations have been resolved, and the compliance status of each chapter is "passed," the collaborative processing end extracts the state transition history, knowledge entry reference records, and document content fragments from the unified state diagram, encapsulates them into a reusable collaborative process instance, and writes it into the knowledge index library. When similar document tasks are initiated subsequently, the system can retrieve this collaborative process instance based on the query vector of the new task and push relevant historical experience and reference materials to the responsible party.

[0217] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but may be executed at different times, or may be executed in turn or alternately with other steps or at least some of the steps or stages of other steps.

[0218] Although embodiments of the present disclosure have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the present disclosure, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A data collaboration method based on a unified state diagram, characterized in that, include: The structured directory information of the document to be processed is parsed, and a task skeleton corresponding to the directory structure of the document to be processed is generated based on the hierarchical relationship in the structured directory information. A unified state graph is constructed based on the task skeleton. The unified state graph includes unified state nodes and connecting edges. The unified state node contains at least a content anchor identifier, a task state identifier, a permission state identifier, and a knowledge context field to form a four-tuple basic data structure. The content anchor identifier is used to associate the unified state node with the corresponding directory location in the document to be processed. The knowledge context field is used to record knowledge entry references or difference description information related to the task state identifier. The connecting edges are used to represent the document hierarchy relationship and / or task dependency relationship between the unified state nodes. Listen for content operation events on the document to be processed, determine the target unified state node based on the correlation between the operation position of the content operation event and the content anchor point identifier, and generate candidate state transitions for the target unified state node. Based on the task state distribution of multiple unified state nodes in the unified state graph, calculate the global information entropy that characterizes the disorder of the state distribution in the unified state graph, and use it as the current global information entropy. Predict the expected global information entropy after executing the candidate state transition, and subtract the current global information entropy from the expected global information entropy to obtain the entropy change. Determine whether to trigger the task state identifier flow of the target unified state node based on the entropy change. When the entropy change is less than or equal to zero, or when the entropy change is greater than zero and less than or equal to a preset entropy increase tolerance threshold, the task status identifier is triggered to transition, and the permission status identifier and knowledge context field of the target unified status node are updated or maintained according to the transitioned task status identifier; when the entropy change is greater than the preset entropy increase tolerance threshold, the corresponding state transition is blocked, and the knowledge context field of the target unified status node is updated or maintained according to the blocking result.

2. The method according to claim 1, characterized in that, Based on the quadruple basic data structure, the unified state node also includes a logical clock identifier and a compliance status field to form a six-tuple data structure. There is a causal constraint relationship between the task status identifier, the permission status identifier, the knowledge context field, and the compliance status field; Change operations on identifiers or fields associated with the same state transition event in the six-tuple data structure are executed atomically within the same logical unit; The logical clock identifier is incremented based on cooperative operation events and is used to maintain the causal consistency of state updates among the unified state nodes in a distributed environment.

3. The method according to claim 1, characterized in that, The process of parsing the structured directory information of the document to be processed and generating a task skeleton corresponding to the directory structure of the document to be processed based on the hierarchical relationship in the structured directory information includes: Semantic analysis is performed on the text of each level heading in the structured directory information to extract task attributes, and the task skeleton is generated based on the hierarchical relationship in the structured directory information and the task attributes. The step of performing semantic analysis on the text of each level of headings in the structured directory information to extract task attributes includes: Dependency parsing is performed on the title text to identify the syntactic structure type of the title text; In response to the fact that the syntactic structure type contains a verb-object structure, the core verb and object in the title text are extracted, the core verb is matched with a preset task type library to determine the task type in the task attribute, and the object is extracted as the expected deliverable in the task attribute; In response to the syntactic structure type being a noun phrase without a verb, the core nouns in the title text are extracted as the expected deliverables, and the task type is inferred based on the semantic attributes of the expected deliverables or a preset default mapping rule. The expected deliverables are matched with a preset organizational structure database to determine the recommended responsible person in the task attributes.

4. The method according to claim 1, characterized in that, The generation of candidate state transitions for the target unified state node includes: In response to the content operation event satisfying the preset valid operation conditions, the candidate state transition is generated; The effective operation conditions include: the content operation event is the first editing operation on the area corresponding to the content anchor point identifier of the target unified state node, and the content increment corresponding to the first editing operation exceeds a preset character number threshold.

5. The method according to claim 1, characterized in that, The calculation of the global information entropy, which characterizes the disorder of the state distribution of the unified state graph, based on the task state distribution of multiple unified state nodes in the unified state graph, includes: Statistical analysis of the frequency of each task status identifier appearing in all the unified status nodes; The global information entropy is calculated using the information entropy algorithm based on the occurrence frequency of each task status identifier.

6. The method according to claim 3, characterized in that, The step of performing semantic analysis on the text of each level of headings in the structured directory information to extract task attributes also includes: evaluating the execution complexity of the task based on the semantic analysis results and performing domain classification. The method further includes: The construction of the unified state diagram and the maintenance of the task state identifier are performed in the collaborative processing terminal; The task is dynamically routed based on the execution complexity and the domain classification. When the execution complexity is determined to be lower than a preset complexity threshold, the collaborative processing operation corresponding to the task is directly executed in the collaborative processing terminal. When the execution complexity is determined to be greater than or equal to the preset complexity threshold, a dispatch request containing execution instructions and context parameters is generated, and the dispatch request is sent to a task processing end independent of the collaborative processing end. The execution result fed back by the task processing end is asynchronously listened to to trigger the update of the task status identifier.

7. The method according to claim 5, characterized in that, The method further includes: When multiple unified state nodes in the unified state graph are detected to be in a blocked state, and any of the following conditions are met: the global information entropy exceeds a preset global entropy threshold, or the increase in global information entropy within a preset time window exceeds a preset increase threshold, the connection edge dependency relationship between the unified state nodes in the blocked state is analyzed, the critical path causing the blockage is identified, and a state transition path to unblock the blockage is recommended based on the critical path. Record the entropy change data corresponding to each state transition event to generate an entropy change curve, and evaluate the state synchronization health of the unified state diagram based on the slope of the entropy change curve.

8. The method according to claim 1, characterized in that, The method further includes: In response to the task status identifier of the target unified status node being updated to an in-progress state, the document context information associated with the content anchor identifier of the target unified status node and the task attribute information recorded by the task skeleton or determined based on the structured directory information are extracted, and the document context information and the task attribute information are combined into a query vector. Based on the query vector, semantic similarity retrieval is performed in the knowledge index library. Knowledge entries with similarity higher than a preset threshold are written into the knowledge context field of the target unified status node and pushed to the corresponding responsible person. The knowledge index base includes an enterprise knowledge base and a personal knowledge base. The personal knowledge base is dynamically updated according to the task execution content. The personal knowledge base is used for knowledge generation and push output, while the enterprise knowledge base is used as a reference input for compliance verification. In response to the fact that the task status identifiers of a group of unified state nodes in the unified state graph have all reached the completed state and the associated documents have been marked as finalized, the state transition history of the group of unified state nodes, the knowledge entry reference records in the knowledge context field, and the document content fragments pointed to by the content anchor identifiers are extracted, structured and encapsulated according to the preset graph data pattern, and stored in the knowledge index library.

9. The method according to claim 2, characterized in that, The method further includes: In response to the task status identifier flowing to a preset verification status, compliance verification is performed on the task data associated with the target unified status node based on a preset enterprise knowledge base, and the compliance status field is updated according to the verification result; When the verification result is passed, the compliance status field is updated to the verification passed status; When the verification result is unsuccessful, the compliance status field is updated to the verification pending status, compliance difference description information is generated, and a collaborative task is generated based on the compliance difference description information. The collaborative task is then distributed to a new responsible person who is different from the original responsible person corresponding to the target unified status node. Set an access control identifier for the collaborative task, and based on the access control identifier, control the new person in charge and the original person in charge to be unable to see each other's submitted information; Based on the permission isolation identifier, the initiator of the collaborative task is controlled to have the permission to view the content filled in by the original person in charge and the new person in charge, and in response to the initiator, a merging operation is performed on each piece of content, and the merged content is updated to the task data corresponding to the target unified status node.

10. The method according to claim 9, characterized in that, The method further includes: When it is detected that the target unified state node causes the task status identifier to fall back between different states due to the failure of the compliance verification, and the number of times the task status identifier falls back within a preset time window reaches a preset fallback number threshold, the local information entropy of the local graph structure to which the target unified state node belongs is calculated. When the local information entropy exceeds the preset disorder threshold, extract the document context information and compliance difference description information associated with the content anchor identifier of the target unified state node; Based on the document context information and the compliance difference description information, semantically matching reference knowledge entries are retrieved from the preset enterprise knowledge base and the preset personal knowledge base. The reference knowledge entries are written into the knowledge context field of the target unified status node and pushed to the current responsible person. If the local information entropy of the local graph structure does not decrease within the preset time window, then state transition or task reallocation suggestions are generated based on the domain classification of the reference knowledge entries.

11. The method according to claim 1, characterized in that, The content manipulation events include annotation events; The method further includes: In response to an annotation event triggered within the area corresponding to any content anchor point in the document to be processed, the annotation location information, annotation content, and mentioner identifier corresponding to the annotation event are extracted, and the target unified state node is determined based on the association between the annotation location information and each content anchor point identifier. Based on the annotation content and the annotation location information, subtasks belonging to the target unified status node are generated, and the subtasks are assigned to the responsible persons corresponding to the mentioner identifier; When all the subtasks associated with the target unified status node are marked as resolved, the task status identifier of the target unified status node is updated from the review status to the completed status.

12. The method according to claim 1, characterized in that, The method further includes: In response to the task status identifier of the target unified status node being changed to a preset review status, the permission status identifier of the area corresponding to the content anchor identifier of the target unified status node in the document to be processed is switched to retain only annotation permissions. In response to the task status identifier of the target unified status node being rolled back from the review status to the in progress status, the permission status identifier of the target unified status node is restored from the restricted editing status with only annotation permissions to the editable status; When the task status identifiers of all the unified status nodes corresponding to the document to be processed meet the preset completion conditions, the finalization or submission approval process for the document to be processed is triggered.

13. A data collaboration device based on a unified state diagram, characterized in that, include: The document parsing and skeleton generation module is used to parse the structured directory information of the document to be processed, and generate a task skeleton corresponding to the directory structure of the document to be processed based on the hierarchical relationship in the structured directory information. A state graph construction module is used to construct a unified state graph based on the task skeleton. The unified state graph includes unified state nodes and connecting edges. The unified state node includes at least a content anchor identifier, a task state identifier, a permission state identifier, and a knowledge context field to form a four-tuple basic data structure. The content anchor identifier is used to associate the unified state node with the corresponding directory position in the document to be processed. The knowledge context field is used to record knowledge entry references or difference description information related to the task state identifier. The connecting edges are used to represent the document hierarchy relationship and / or task dependency relationship between the unified state nodes. The state transition decision module is used to monitor content operation events on the document to be processed, determine the target unified state node based on the correlation between the operation position of the content operation event and the content anchor point identifier, and generate candidate state transitions for the target unified state node; based on the task state distribution of multiple unified state nodes in the unified state graph, calculate the global information entropy characterizing the disorder of the state distribution of the unified state graph, as the current global information entropy; predict the expected global information entropy after executing the candidate state transition, and subtract the current global information entropy from the expected global information entropy to obtain the entropy change, and determine whether to trigger the task state identifier transition of the target unified state node based on the entropy change; when the entropy change is less than or equal to zero, or when the entropy change is greater than zero and less than or equal to a preset entropy increase tolerance threshold, trigger the transition of the task state identifier, and update or maintain the permission state identifier and knowledge context field of the target unified state node based on the transitioned task state identifier; when the entropy change is greater than the preset entropy increase tolerance threshold, block the corresponding state transition, and update or maintain the knowledge context field of the target unified state node based on the blocking result.

14. An electronic device, characterized in that, include: A memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, the processor executing the computer instructions to perform the method of any one of claims 1 to 12.

Images