A tablet computer terminal digitized office file collaborative management system

Abstract

The application discloses a tablet computer terminal digitized office file cooperative management system and relates to the technical field of computer software and network communication.The tablet computer terminal digitized office file cooperative management system comprises a cloud main workspace twin, an edge shadow twin and a terminal shadow twin.The cloud main workspace twin is arranged on a cloud server and is used for building a global knowledge graph and modeling user behaviors.The tablet computer terminal digitized office file cooperative management system shortens the response time of file opening and interface switching, provides adaptive protection for sensitive files in different use environments, simplifies the starting process of multi-person cooperative editing, reduces manual configuration operations of users, enables users to seamlessly switch current work among different devices, reduces the data transmission amount between the cloud, edge nodes and terminal devices, and improves the operation stability of the overall system in a weak network environment.

Description

Technical Field

[0001] This invention relates to the field of computer software and network communication technology, specifically to a digital office document collaborative management system for tablet computers. Background Technology

[0002] With the rapid development of mobile internet technology and the widespread adoption of smart terminal devices, tablet computers have gradually become an important platform for mobile office work due to their portability, touch operation advantages, and powerful processing capabilities. Especially against the backdrop of digital office transformation, enterprises' demands for file collaboration, data sharing, and access control are increasing, driving in-depth research and development of mobile office systems.

[0003] In the existing technology, some mobile office collaboration systems have already been put into use. For example, patent document with authorization publication number "CN106469361A" describes a "collaboration system suitable for mobile office". This system is based on a private cloud storage architecture, deploys storage servers locally, and supports functions such as web access, online document preview, Office plugin integration, and internal and external link sharing, effectively improving the security and collaboration efficiency of enterprise document management. The core advantage of this system lies in its data encryption and permission configuration capabilities under its enterprise-level architecture, enabling cross-terminal document access and sharing, such as on PCs and smartphones.

[0004] However, the aforementioned system still has certain limitations: First, its design is mainly centered around PCs or general web interfaces. Although it can be accessed via a browser on smartphones or tablets, it has not been deeply optimized for tablets in terms of interface adaptation, touch interaction, and offline operation, resulting in a less than ideal user experience. Second, the system's file collaboration function relies heavily on online status, and document access and editing capabilities are limited in weak or no network environments. Furthermore, although the system supports permission configuration, it lacks support for advanced collaboration functions such as version control, operation logging, and task workflow in real-time multi-user collaboration scenarios, making it difficult to meet the refined management needs of complex office scenarios.

[0005] In conclusion, existing mobile office systems still need improvement in terms of tablet compatibility, offline collaboration capabilities, and intelligent management functions. Therefore, there is an urgent need for a digital office document collaboration management system specifically designed for tablets, offering efficient collaboration, intelligent control, and a superior user experience. Summary of the Invention

[0006] The purpose of this invention is to provide a digital office document collaborative management system for tablet computers to solve the problems mentioned in the background art.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a digital office document collaborative management system for tablet computers, including a cloud-based main workspace twin, an edge shadow twin, and a terminal shadow twin; The cloud-based main workspace twin is deployed on a cloud server and is used to construct a global knowledge graph and model user behavior. The edge shadow twin is deployed on the enterprise's local server or mobile edge computing node to provide temporary data collaborative arbitration services in environments with weak network or no network coverage. The terminal shadow twin is deployed locally on the tablet computer to collect user interaction data, generate local operation logs, and preload file content according to the prediction instructions issued by the cloud main workspace twin. The cloud-based main workspace twin, the edge shadow twin, and the terminal shadow twin maintain data consistency through an adaptive synchronization protocol.

[0008] The three twins are equipped with the same core data model. The adaptive synchronization protocol can adjust the data synchronization strategy and priority according to the network transmission status, and synchronize only the incremental change information of data atoms among the three.

[0009] Preferably, the system further includes an offline collaborative consistency engine; The offline collaborative consistency engine is set locally on the tablet computer and is used to generate operation atoms when the network is disconnected, and to sign and sort the operation atoms. After the network is restored, the edge shadow twin receives the operation atoms, performs spatiotemporal alignment on the operation atoms from different terminals, and merges the aligned data into the cloud main workspace twin. The operation atom is a data unit carrying a timestamp, spatial coordinates, semantic tags and operator digital signature, and the file content is composed of at least one of the operation atoms.

[0010] The offline collaborative consistency engine performs asymmetric cryptographic signatures on the generated operation atoms, and the edge shadow twin can complete the spatiotemporal alignment of operation atoms through temporal relationships and logical dependencies.

[0011] Preferably, the system further includes an intelligent semantic control engine; The intelligent semantic management engine is set in the main workspace twin in the cloud and is used to construct a multimodal cognitive graph; The nodes of the multimodal cognitive graph include file content, user roles, tablet sensor data, operation timing data, and environmental data. The edges of the multimodal cognitive graph are used to represent the relationships between the nodes. The intelligent semantic control engine predicts the user's operational intent through a spatiotemporal graph convolutional network based on the multimodal cognitive graph, and generates the prediction instruction based on the prediction result and sends it to the terminal shadow twin.

[0012] Multimodal cognitive graphs can dynamically update node attributes and edge weights based on newly added synchronous data, while spatiotemporal graph convolutional networks can extract spatiotemporal features of user actions to predict intent.

[0013] Preferably, the system further includes an adaptive touch twin rendering engine; The adaptive touch twin rendering engine is set locally on the tablet computer and is used to generate a three-dimensional interactive interface that conforms to the current ergonomics in real time based on the holding posture, screen size or ambient light of the tablet computer. The adaptive touch twin rendering engine is also used to build a virtual document object model in the tablet computer's memory in advance according to the prediction instructions issued by the intelligent semantic management engine, and to complete the rendering of the next screen content before the user performs a swipe operation.

[0014] This engine can call on various sensors on the tablet to collect scene parameters and quickly generate the interactive interface through a lightweight model. The pre-rendered content will be cached in the tablet's memory.

[0015] Preferably, the offline collaborative consistency engine is further used to record the operation atoms as a version control structure based on a directed acyclic graph (DAG), where each node in the version control structure corresponds to one operation atom, and the edges in the version control structure are used to represent the temporal relationship or dependency relationship between the operation atoms. The DAG can completely record the logical association of offline operations through the construction of nodes and edges, effectively avoiding the problem of operation temporal disorder caused by relying solely on timestamps.

[0016] Preferably, the edge shadow twin is also used to receive the operation atoms generated by multiple tablet terminals during network disconnection, and to perform consensus confirmation on the receiving order of the operation atoms through lightweight distributed ledger technology to generate a consensus result; After the network is restored, the edge shadow twin sends the operation atoms and the consensus results to the cloud main workspace twin.

[0017] The consensus result records the unique identifier of the operation atom and the confirmation node information, which are stored locally in the edge shadow twin along with the operation atom.

[0018] Preferably, the intelligent semantic control engine is also used to calculate the user's operational risk value in the current environment in real time based on the multimodal cognitive graph; When the operational risk value exceeds a preset threshold, the intelligent semantic control engine automatically adjusts the user's file operation permissions on the tablet and sends a biometric authentication request to the tablet. The operational risk value is calculated by weighting multi-dimensional scene features. After successful biometric authentication, the user's original file operation permissions are restored, and the relevant authentication results are synchronized to the multimodal cognitive graph.

[0019] Preferably, the intelligent semantic control engine is also used to detect the editing behavior of multiple users on the same file based on the multimodal cognitive graph; When at least two users are detected simultaneously editing the same file, the intelligent semantic control engine automatically creates a temporary virtual collaboration space and grants collaborative editing permissions to the at least two users within this space. The temporary virtual collaboration space is a dynamically generated, logically isolated area in the cloud. User editing operations within this space are synchronized in real-time as atomic operations, and each user's operation area is uniquely marked.

[0020] Preferably, the intelligent semantic control engine is also used to detect the matching degree between the current task and the state of the user's terminal device based on the prediction result of the user's operation intention; When the matching degree is lower than a preset value, the intelligent semantic control engine generates a task migration instruction, synchronizing the current task's editing status to other terminal devices associated with the user. The task migration instruction includes the latest edited version of the file and the precise editing location; the original terminal retains a read-only copy of the file for the user to view at any time.

[0021] Preferably, the cloud-based main workspace twin, the edge shadow twin, and the terminal shadow twin all store the same core data model; The core data model is used to uniformly represent the file content, user operation behavior and context information as data atoms. The data atoms include at least the timestamp dimension, spatial coordinate dimension and semantic tag dimension. The adaptive synchronization protocol is used to synchronize incremental changes of data atoms among the cloud main workspace twin, the edge shadow twin, and the terminal shadow twin. Operation atoms are subclasses of data atoms, inheriting the basic dimensional information of data atoms and adding operator digital signatures and hash verification values ​​to achieve unified encapsulation and parsing of data units within the system.

[0022] This invention provides a digital office document collaborative management system for tablet computers. It has the following beneficial effects: This tablet-based digital office document collaboration management system constructs a three-tiered collaborative digital twin architecture—a cloud-based main workspace twin, an edge shadow twin, and a terminal shadow twin—and configures a unified spatiotemporal data atomic encapsulation standard for the core data model. This enables unified expression and precise synchronization of document content and user operations across different network environments. The system's operation atomic processing mechanism and version control structure based on a directed acyclic graph (DAG) allow edited data generated in offline environments to carry complete temporal, spatial, and semantic information. After network recovery, the spatiotemporal alignment capability of the edge shadow twin achieves accurate merging with cloud data, avoiding data version chaos caused by concurrent editing. The construction of a multimodal cognitive graph and the deployment of a spatiotemporal graph convolutional network enable the system to understand office scenarios from multiple dimensions, including user behavior, device status, and environmental data, providing a data foundation for predicting user operation intentions and dynamically adjusting resources.

[0023] This tablet-based digital office document collaboration management system ensures that offline edited content remains completely consistent with the cloud version after network recovery, eliminating the need for manual version conflict resolution. Cognitive graph-based operation intent prediction commands allow the tablet to preload content before user actions, shortening response time for file opening and interface switching. Dynamic risk value calculation and operation permission adjustment mechanisms provide adaptive protection for sensitive files in different usage environments without disrupting normal office workflows. Automatic creation and release of temporary virtual collaboration spaces simplify the startup process for multi-user collaborative editing, reducing manual configuration operations. Task and terminal device status matching analysis and task migration command generation enable users to seamlessly switch between different devices, maintaining work continuity. A unified data atomic model and incremental synchronization protocol reduce data transmission volume between the cloud, edge nodes, and terminal devices, improving the overall system's operational stability in weak network environments. Attached Figure Description

[0024] Figure 1 This is a data flow diagram between modules of a digital office document collaborative management system for tablet computers according to the present invention; Figure 2 This is a sequence diagram of task migration for a digital office document collaborative management system on a tablet computer according to the present invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] Please see Figure 1 and Figure 2 The present invention provides a technical solution: a cloud main workspace twin, an edge shadow twin, and a terminal shadow twin; The cloud-based main workspace twin is deployed on a cloud server to build a global knowledge graph and model user behavior. Edge shadow twins are deployed on enterprise local servers or mobile edge computing nodes to provide temporary data collaborative arbitration services in environments with weak networks or no network coverage. The terminal shadow twin is deployed locally on the tablet computer to collect user interaction data, generate local operation logs, and preload file content according to the predicted instructions issued by the main workspace twin in the cloud. Among them, the cloud-based main workspace twin, the edge shadow twin, and the terminal shadow twin maintain eventual data consistency through an adaptive synchronization protocol.

[0027] It should be further explained that, in the specific implementation, the system constructs a cloud-edge-device three-level collaborative digital twin architecture. Its core is to provide users with a virtual workspace that maintains continuity across different network environments and terminal devices through a main workspace twin deployed on a cloud server, a shadow twin deployed on an enterprise local server or mobile edge computing node, and a shadow twin deployed on a tablet computer.

[0028] The cloud-based main workspace twin is responsible for analyzing global file data and users' long-term operational behavior patterns, and constructing a knowledge graph that includes file content, user roles, and interaction history. Based on this knowledge graph, it predicts the user's possible next action and generates corresponding preloading or task scheduling instructions.

[0029] Edge shadow twins on enterprise local servers or mobile edge computing nodes, as an extension of cloud capabilities, can take over some collaborative management functions when the connection between tablets and cloud networks is unstable or interrupted. They can provide low-latency data exchange and operation arbitration services for multiple tablets in the same local area network or within the coverage area of ​​edge nodes, ensuring the continuity of local collaboration.

[0030] The terminal shadow twin deployed locally on the tablet is responsible for collecting the user's specific interaction behaviors in real time, such as touch position, pen pressure, browsing path, etc., and converting these behaviors into local operation logs with timestamps and semantic tags. At the same time, the terminal shadow twin receives and executes predictive instructions from the main workspace twin in the cloud, and builds the file content or interface elements that the user is about to access in advance in local memory.

[0031] The cloud-based main workspace twin, the edge shadow twin, and the terminal shadow twin communicate through an adaptive synchronization protocol. This protocol can dynamically adjust the data synchronization strategy and priority based on the current network quality. For example, it can synchronize the complete file version when the network conditions are good, and only synchronize key operation logs or metadata when the network conditions are limited. This ensures the eventual consistency of the data in different states. That is, when the network is restored or the user switches devices, all operations and file versions can be automatically merged, enabling the user's work on different devices and at different times to be seamlessly connected.

[0032] The cloud-based main workspace twin is deployed on a public or private cloud server cluster with elastic computing capabilities. It adopts a microservice architecture to decompose core functional modules, including a global data management module, a user behavior modeling module, a multimodal cognitive graph construction module, an intelligent semantic control engine core module, and a synchronization protocol service module. Each module communicates internally through a RESTful interface and externally establishes long-term communication channels with the edge shadow twin and the terminal shadow twin through the WebSocket protocol. The server cluster's hardware configuration meets a minimum of 8 cores and 16GB of computing resources and more than 1TB of solid-state storage resources. It can be elastically expanded according to the number of concurrent users. The core data model of the cloud-based main workspace twin is completely consistent with that of the edge and terminal sides, ensuring the uniformity of data parsing.

[0033] The edge shadow twin is deployed on a rack server or 5G mobile edge computing node within the enterprise's local area network. The hardware configuration must meet at least 4 cores and 8GB of computing resources, and more than 500GB of solid-state storage resources. The software architecture adopts a lightweight microservice architecture, including a local collaborative arbitration module, an operation atomic spatiotemporal alignment module, a lightweight consensus module, and an edge-side synchronization protocol adaptation module. The edge shadow twin communicates with the cloud main workspace twin via a wide area network and with the terminal shadow twin within the same local area network via a local area network. When the connection between the terminal and the cloud wide area network is interrupted but the local area network connection is normal, the edge shadow twin takes over the collaborative arbitration function. When the network connection between the terminal and the cloud and the edge node is interrupted, the offline collaborative consistency engine on the terminal side completes the recording and storage of local operations.

[0034] The terminal shadow twin is deployed locally on a tablet running Android, HarmonyOS, or iOS operating systems. It is installed and runs as an independent application. The application integrates a local data storage module, a user interaction data collection module, an operation log generation module, a file preloading module, a local rendering adaptation module, an offline collaborative consistency engine, and an adaptive touch twin rendering engine. The application obtains sensor data, screen parameters, grip posture data, network status data, and battery power data of the device through the tablet system API. It collects user interaction data such as touch position, touch pressure, swipe trajectory, and editing operations through the touch event callback interface. All collected data is encapsulated into corresponding data atoms according to the requirements of the core data model and stored in the embedded database on the tablet. At the same time, it interacts with the edge and cloud sides through an adaptive synchronization protocol.

[0035] The adaptive synchronization protocol adopts an incremental synchronization architecture, consisting of five layers: the protocol handshake layer, the data parsing layer, the synchronization strategy decision layer, the data transmission layer, and the conflict resolution layer. Upon protocol startup, both communicating parties first complete identity authentication and core data model version verification through the handshake layer, ensuring that the core data models used by both parties are completely consistent. After successful verification, an encrypted communication channel is established. The data parsing layer is responsible for breaking down the content to be synchronized into the smallest granularity of data atoms, extracting only the incremental information of the changed data atoms, rather than the full file data.

[0036] The synchronization strategy decision layer monitors the network bandwidth, latency, and packet loss rate parameters of the current communication link in real time, and dynamically adjusts the synchronization strategy according to the network status. When the network status is good and the bandwidth is sufficient, a real-time full incremental synchronization mode is adopted to synchronize all changed data atoms to the peer in real time. In a weak network environment, a priority synchronization mode is adopted, prioritizing the synchronization of small-volume key data atoms such as metadata and operation logs, and delaying the synchronization of large-volume data atoms such as file content and attachments. In a network-free environment, the synchronization task is paused, and the changed data atoms are cached locally, and the transmission is automatically triggered to resume after the network is restored.

[0037] The data transmission layer adopts a breakpoint resume mechanism, which segments data atoms that exceed a preset size for transmission in a single transmission. If a transmission interruption occurs, the transmission will resume from the breakpoint after the connection is restored, without having to retransmit the full amount of data.

[0038] When the conflict resolution layer receives synchronized data atoms from the peer, it checks whether there are concurrent modification conflicts in the same location and time window based on the timestamp, spatial coordinates, and semantic tag dimension information of the data atoms. When a conflict is detected, it prioritizes retaining the operation atom data with digital signature, and simultaneously marks the conflict content and synchronizes it to both ends for users to confirm for a second time, ensuring eventual data consistency.

[0039] The core data model encapsulates all information that needs to be managed and synchronized within the system into data atoms. Each data atom is the smallest, universally applicable data unit within the system, containing three basic dimensions: timestamp, spatial coordinates, and semantic tags. The timestamp dimension uses globally unified UTC time, accurate to the millisecond, recording the occurrence time of the event corresponding to the data atom. The spatial coordinate dimension defines corresponding location information based on the data atom type. For file content data atoms, the spatial coordinates are the paragraph number, character offset, and cell coordinates within the file. For user operation data atoms, the spatial coordinates are the two-dimensional coordinates of the operation on the tablet screen. For device status data atoms, the spatial coordinates are the device's geographic location coordinates. The semantic tag dimension assigns standardized business semantics to the data atom, including but not limited to inserting text, deleting content, formatting adjustments, device posture changes, permission changes, and file preloading, ensuring that each data atom carries a business meaning that can be parsed by the system.

[0040] Operation atoms are a subclass of data atoms. They are dedicated data atoms specifically encapsulated for user file editing operations. In addition to the three basic dimensions of data atom information, they also carry the operator's digital signature and the hash verification value of the edited content. The file content is composed of one or more operation atoms combined according to the temporal relationship. All operation atoms conform to the encapsulation standard of data atoms and can be directly incrementally synchronized through the adaptive synchronization protocol.

[0041] The system also includes an offline collaborative consistency engine; The offline collaborative consistency engine is set up locally on the tablet to generate operation atoms when the network is disconnected, and to sign and sort the operation atoms. After the network is restored, the edge shadow twin receives operation atoms, performs spatiotemporal alignment on operation atoms from different terminals, and merges the aligned data into the main workspace twin in the cloud. Among them, the operation atom is a data unit carrying a timestamp, spatial coordinates, semantic tags and operator digital signature, and the file content consists of at least one operation atom.

[0042] It should be further explained that, in the specific implementation, the system deploys an offline collaborative consistency engine locally on the tablet computer. This engine is deeply coupled with the terminal shadow twin and is used to handle file editing and data generation tasks in environments without network coverage.

[0043] When the tablet detects a network connection loss, the offline collaborative consistency engine does not put the file into a read-only state, but allows the user to continue editing the file. During this process, the engine encapsulates each user's editing action—such as modifying a paragraph in a document or updating a cell in a table—as an independent data unit, i.e., an atomic operation.

[0044] Each operation atom encapsulates the precise timestamp of the edit, its position coordinates within the file content, semantic tags for the editing action (e.g., insertion, deletion, formatting), and the current user's digital signature, ensuring the uniqueness and traceability of each operation atom. These operation atoms are recorded locally in the order of their generation and simultaneously written into a version control structure built on a directed acyclic graph. This structure maintains the sequential dependencies between different operation atoms through the pointing relationships between nodes, avoiding logical confusion that might result from simply relying on timestamps.

[0045] Once the tablet's network is restored and it reconnects to the edge shadow twin, the offline collaborative consistency engine uploads all locally stored operation atoms and their corresponding directed acyclic graph structures to the edge shadow twin, either all at once or in batches. Upon receiving concurrent operation atoms from multiple tablets, the edge shadow twin does not simply overwrite or merge them. Instead, based on the timestamp and spatial coordinates of each operation atom, combined with the dependencies revealed by the directed acyclic graph, it performs a spatiotemporal alignment operation. This identifies which operation atoms target the same file and the same location, and automatically resolves potential editing conflicts based on their temporal order and logical dependencies, generating a new sequence of operation atoms with a clear order and logical consistency.

[0046] Subsequently, the edge shadow twin synchronizes this aligned sequence to the main workspace twin in the cloud, where the cloud completes the final data merging and version update, allowing all offline editing results to be fully and accurately integrated into the global workspace.

[0047] The offline collaborative consistency engine is integrated into the local terminal shadow twin of the tablet computer, and works in real time with the local embedded database and touch event acquisition module. When the tablet computer detects that the network connection with the main workspace twin and the edge shadow twin in the cloud is interrupted, it automatically starts the offline editing mode. At this time, all editing operations of the user on the file will not be restricted. The offline collaborative consistency engine will break down each touch editing operation of the user into the smallest granularity of editing action, generate corresponding operation atoms for each editing action, and complete the encapsulation according to the core data model standard. At the same time, the operation atoms are asymmetric encrypted and signed using the digital certificate pre-stored locally by the current user to ensure the immutability and traceability of the operation atoms. All generated operation atoms are stored in the local embedded database according to the generation time sequence, and are also written into the version control structure built based on the directed acyclic graph.

[0048] The system also includes an intelligent semantic control engine; The intelligent semantic control engine is located in the main workspace twin in the cloud and is used to build a multimodal cognitive graph; The nodes of the multimodal cognitive graph include file content, user roles, tablet sensor data, operation timing data, and environmental data. The edges of the multimodal cognitive graph are used to represent the relationships between nodes. The intelligent semantic control engine predicts the user's operational intent through a spatiotemporal graph convolutional network based on a multimodal cognitive graph, and generates prediction instructions based on the prediction results to send to the terminal shadow twin.

[0049] It should be further explained that, in the specific implementation, the system integrates an intelligent semantic management engine in the main workspace twin in the cloud. This engine is responsible for building a multimodal cognitive graph to achieve a deep understanding of users' office behavior.

[0050] The data sources for constructing this graph are not limited to traditional file attributes and user account information, but extend to multiple dimensions: semantic themes and key entities extracted from the file content itself, user roles and reporting relationships imported from the enterprise organizational structure database, device posture and ambient light data collected in real time from sensors such as accelerometers, gyroscopes and cameras embedded in the tablet, operation timing data such as touch pressure, editing pause frequency and page dwell time recorded from the user interface, and environmental data such as current signal strength and connection stability obtained from the network module.

[0051] The aforementioned data points are abstracted as nodes in a graph, and the relationships between nodes, such as a user frequently editing a file during a certain period, a user's device being in motion while performing a certain operation, or the modification of a file being consecutive in time with an attachment uploaded by another user, are constructed as edges in the graph, thus forming a dynamically evolving semantic network that reflects the complex interactions between people, devices, content, and the environment.

[0052] The intelligent semantic control engine is integrated into the core services of the cloud-based main workspace twin. It is responsible for the construction, updating, and inference application of the multimodal cognitive graph. When constructing the multimodal cognitive graph, the intelligent semantic control engine connects to the global database in the cloud, extracting multi-dimensional data sources. These include semantic topics, key entities, and paragraph association information extracted from document content using natural language processing technology; user roles, permission levels, reporting relationships, and collaboration history information synchronized from the enterprise organizational structure database; device grip posture, geographical location, ambient light, and motion status data collected from the tablet computer's sensors synchronized from the terminal shadow twin; user operation timing data synchronized from the terminal, including touch pressure, editing pause frequency, page dwell time, and operation path information; and the terminal's network status and battery status. Battery power and equipment load environmental data are used to abstract each independent data entity into a node in a multimodal cognitive graph. Standardized attribute fields are configured for each node to record the specific parameter information of the entity. At the same time, the relationships between different nodes are analyzed, including the relationship between users and file editing, the relationship between users and users collaboration, the relationship between operation behavior and device status, and the relationship between file content references. All kinds of relationships are abstracted into edges between nodes in the graph, and a corresponding weight value is configured for each edge. The weight value is dynamically updated according to the frequency and tightness of the relationship, and finally a dynamically evolving multimodal cognitive graph is formed. The graph will be updated in real time according to a preset period, combined with newly added synchronized data, to ensure that the information of nodes and edges can accurately reflect the current office scene status.

[0053] The intelligent semantic control engine's built-in spatiotemporal graph convolutional network model is specifically designed for predicting user operation intentions based on multimodal cognitive graphs. This model employs a lightweight network structure with two graph convolutional layers and one fully connected layer. The input consists of the adjacency matrix and node attribute matrix of the multimodal cognitive graph. The adjacency matrix represents the relationships and edge weights between nodes in the graph, while the node attribute matrix represents the multidimensional attribute information of each node. The model extracts the spatial topological features of the graph through graph convolutional layers, capturing the association features between users, files, devices, and the environment. Simultaneously, it extracts the temporal change features of node attributes through temporal convolutional kernels, capturing the changing patterns of user operation behavior over time. Finally, the fully connected layer outputs the predicted operation intentions of the user within a preset future time window, including the file identifier that the user may access, the type of operation that may be performed, and the probability information of the operation.

[0054] The model is pre-trained using massive amounts of historical operation data from enterprise office scenarios. The training dataset includes user operation behavior data and corresponding subsequent operation labels for different user roles, different office scenarios, and different device states. During training, the cross-entropy loss function is used to optimize model parameters, ensuring that the model's prediction accuracy meets the preset accuracy requirements. After the model is deployed in the cloud, it will be incrementally fine-tuned in conjunction with real-time user operation data to continuously optimize prediction accuracy. After the intelligent semantic control engine obtains the operation intent prediction results output by the model, when the prediction probability exceeds a preset threshold, it generates the corresponding prediction instruction and sends it to the corresponding user's terminal shadow twin through an adaptive synchronization protocol. The instruction contains the file identifier that the user may access, the range of preloaded content, and pre-rendering requirements information, so that the terminal can execute the corresponding preloading and pre-rendering operations.

[0055] Based on this continuously updated multimodal cognitive graph, the intelligent semantic control engine invokes a spatiotemporal graph convolutional network model deployed in the cloud to perform periodic or event-triggered inference. This model captures patterns in user behavior across time and space by analyzing changes in node attributes and edge weights and directions within the graph. For example, it predicts user intent based on behaviors such as launching related applications after reviewing certain types of documents, or handling specific types of approval processes at specific locations or time periods.

[0056] The intelligent semantic control engine encodes the prediction results into specific prediction instructions. These instructions include information such as the file identifiers that the user may access, the types of operations that may be performed, and the predicted time window. They are then sent to the terminal shadow twin on the user's tablet via an adaptive synchronization protocol, allowing the tablet to preload content or adjust the interface.

[0057] The system also includes an adaptive touch twin rendering engine; The adaptive touch twin rendering engine is set up locally on the tablet and is used to generate a 3D interactive interface that conforms to the current ergonomics in real time based on the tablet's grip posture, screen size, or ambient lighting. The adaptive touch twin rendering engine is also used to build a virtual document object model in the tablet's memory in advance based on the predictive instructions issued by the intelligent semantic control engine, and to complete the rendering of the next screen's content before the user performs a swipe operation.

[0058] It should be further explained that, in the specific implementation, the system integrates an adaptive touch twin rendering engine locally on the tablet computer. This engine works in conjunction with the terminal shadow twin and the intelligent semantic control engine to generate a personalized 3D interactive interface in real time based on the user's current physical environment and operating habits.

[0059] The engine uses the tablet's built-in accelerometer, gyroscope, and front-facing camera to collect real-time data on the device's current holding posture, such as whether the device is held by the left or right hand, whether the screen is in landscape or portrait orientation, and the tilt angle of the device relative to the user's face. It also combines the screen's physical size and ambient light intensity to construct a multi-dimensional parameter vector describing the current interaction scenario.

[0060] Based on this parameter vector, the adaptive touch twin rendering engine calls a pre-installed neural radiation field model on the tablet. This model is pre-trained and can dynamically generate a matching 3D interactive interface layout according to the input scene parameters. For example, when the engine detects that the user is holding the device with their left hand in a bright light environment, the engine will arrange commonly used navigation buttons and operation controls in the lower left corner to the center of the screen to adapt to the natural range of motion of the left thumb, while increasing the contrast of the control edges and increasing the effective size of the clickable area. When the engine detects that the user is holding the device with both hands and the device is placed horizontally, the engine will reconstruct the interface elements into a symmetrical layout suitable for two-handed operation and unfold the virtual keyboard into a split form suitable for two-handed thumb input.

[0061] The adaptive touch twin rendering engine is integrated into the local terminal shadow twin of the tablet computer. It works in real time with the tablet system's graphics rendering components, sensor modules, and touch event modules to generate and pre-render the interactive interface in real time. The engine obtains the device's screen size, resolution, and screen orientation parameters in real time through the tablet system's API. It collects the device's grip posture and tilt angle data through gyroscope and accelerometer sensors, collects the current ambient light intensity data through the ambient light sensor of the front camera, and collects the user's grip shape and touch point distribution data through touch event callbacks. All of the above parameters are integrated into a multi-dimensional feature vector of the current interaction scene and input into the built-in lightweight neural radiation field model.

[0062] This neural radiation field model is a lightweight model adapted and optimized for tablet interactive interface generation scenarios. It is pre-trained using human-computer interaction interface sample data under different screen sizes, different holding postures, and different ambient lighting conditions. The redundant network layers used for 3D scene reconstruction in the original model have been removed, while the core network structure for scene feature extraction and interface layout generation has been retained. It can quickly output ergonomic interface layout parameters based on the input interactive scene feature vector, including the position, size, contrast, and clickable area size of interactive controls, the font size and line spacing of the interface, and the layout of the virtual keyboard.

[0063] Based on the layout parameters output by the model, the engine calls the tablet system's graphics rendering components to generate the corresponding interactive interface in real time. For example, when it detects that the user is holding the tablet with one hand on the left and the screen is in portrait mode, the engine will arrange frequently used navigation controls and operation buttons in the lower left corner to the center of the screen to adapt to the natural range of motion of the left thumb. At the same time, it will increase the clickable area of ​​the controls to avoid accidental touches. When it detects that the ambient light intensity is high, it will automatically increase the contrast of the interface elements and adjust the screen brightness to ensure that the interface content is clear and readable. When it detects that the user is holding the tablet with both hands and the screen is in landscape mode, it will automatically adjust the virtual keyboard to a split layout to adapt to the input habits of both thumbs.

[0064] Furthermore, the rendering engine continuously listens for predictive commands issued by the intelligent semantic control engine through the terminal shadow twin. These commands encapsulate the file identifiers or possible operation types that the user might access next. Upon receiving a command, the rendering engine does not wait for the user to actually initiate an action. Instead, it pre-builds a virtual document object model in the tablet's memory, following the content the user is currently viewing. This model contains the complete document structure and visual elements that will be presented on the next screen.

[0065] When a user swipes their finger across the screen, triggers page turning or scrolling, the rendering engine directly calls the pre-built virtual model from memory to draw it, thus achieving real-time presentation of the interface content in the user's visual perception.

[0066] The adaptive touch twin rendering engine also continuously receives predictive instructions from the cloud-based intelligent semantic management engine forwarded by the terminal's shadow twin. It parses the file identifier and preloaded content range information contained in the instructions. During the idle period of the user's current operation, it retrieves the content data of the corresponding file from local storage or edge nodes in advance and builds a virtual document object model of the file in the tablet's memory. This model contains the complete paragraph structure, format information, and visual element data of the file. At the same time, based on the user's current browsing progress, it predicts the user's next swipe direction and browsing range, and completes the rendering calculation of the next screen or multiple screens of content in advance. The rendered screen is cached in memory. When the user actually triggers a swipe or page-turning operation, the engine directly retrieves the cached rendered screen from memory for display, without the need for real-time rendering calculation, which greatly shortens the response time of interface switching and improves the smoothness of operation.

[0067] The offline collaborative consistency engine is also used to record operation atoms as a version control structure based on a directed acyclic graph. Each node in the version control structure corresponds to an operation atom, and the edges in the version control structure are used to represent the temporal relationship or dependency relationship between operation atoms.

[0068] It should be further explained that, in the specific implementation, the system builds a data version control structure based on a directed acyclic graph within the offline collaborative consistency engine to manage the operation atoms generated in the offline environment.

[0069] When a user edits a file offline on a tablet, each editing action is encapsulated as an independent operation atom. Instead of being stored as a simple list or queue, it is recorded as a node in a directed acyclic graph (DAG). Each newly generated operation atom node contains not only its edited content, timestamp, and location coordinates, but also automatically resolves and establishes directed edges with existing nodes in the graph. Specifically, if the current operation atom modifies a specific location in the document, and that location has been modified by other operation atoms before, the engine will point the current node to its predecessor node based on the order of timestamps and the logical relationships of the edited content, forming a directed edge representing a causal dependency.

[0070] This process continues locally, allowing the entire directed acyclic graph to fully record the order and dependencies of all offline operations. For example, if a user deletes a paragraph offline and then inserts new text in the original position of that paragraph, the node corresponding to the insertion operation will generate a directed edge pointing to the node that deleted the paragraph, indicating that the former is logically dependent on the latter.

[0071] When the tablet computer's network is restored and it synchronizes data with the edge shadow twin, the engine uploads the structure of the directed acyclic graph (DAG) along with all nodes. After receiving multiple DAGs from various devices, the edge shadow twin accurately identifies the true dependencies between concurrent operations on different devices by traversing the directed edges of the nodes in the graph, rather than relying solely on timestamps. This avoids timing discrepancies caused by asynchronous device clocks or network latency, providing a structured logical foundation for subsequent spatiotemporal alignment and conflict resolution.

[0072] When constructing the directed acyclic graph version control structure, the offline collaborative consistency engine treats each operation atom as an independent node in the graph. Each time a new operation atom node is generated, it automatically parses the spatial coordinate dimension information of the operation atom, retrieves the existing predecessor operation atom nodes under the spatial coordinates, and establishes a directed edge from the current node to the predecessor node based on the temporal sequence and the logical dependency relationship of the edited content. If the current operation atom has no logical dependency relationship with the existing node, a directed edge is established pointing to the root node of the file. This ensures that the entire directed acyclic graph can completely record the temporal relationship and causal dependency of all offline operations, avoiding logical confusion caused by simply relying on timestamps.

[0073] Once the tablet's network is restored and a new connection is established with the edge shadow twin, the offline collaborative consistency engine uploads all locally stored operation atoms and their corresponding directed acyclic graph (DAG) structures to the edge shadow twin. Upon receiving the operation atoms and DAG structures uploaded from multiple terminals, the edge shadow twin performs a spatiotemporal alignment operation. First, it traverses all nodes and directed edges of the DAG to clarify the logical dependencies of all operation atoms. Then, combining the timestamp and spatial coordinate dimensions of the operation atoms, it identifies concurrent operations targeting the same location in the same file. Following a rule prioritizing logical dependencies and secondarily considering timestamp order, it sorts all operation atoms, automatically resolving concurrent operations without logical conflicts and marking those with logical conflicts. Finally, it generates a set of operation atom sequences with clear temporal order and logical consistency, and then synchronizes this sequence to the cloud-based main workspace twin, completing the final data merging and version update.

[0074] The edge shadow twin is also used to receive operation atoms generated by multiple tablet terminals during network outages, and to confirm the order of receiving operation atoms through lightweight distributed ledger technology to generate consensus results. Once the network is restored, the edge shadow twin will send the operation atoms and consensus results to the cloud main workspace twin.

[0075] It should be further explained that, in the specific implementation, the system configures the edge shadow twin as a temporary collaborative arbitration node in environments with weak network coverage or no network coverage.

[0076] When the connection between multiple tablet terminals and the cloud-based main workspace twin is interrupted, but they are still within the coverage area of ​​the enterprise LAN or mobile edge computing nodes, the operation atoms generated by the offline collaborative consistency engine on these tablets no longer wait directly for a response from the cloud, but are first sent to the local edge shadow twin.

[0077] Upon receiving operation atoms from different terminals, the edge shadow twin does not simply sort them by reception timestamp. Instead, it initiates an embedded lightweight distributed ledger consensus module. This module employs a consensus algorithm based on a directed acyclic graph structure, packaging each newly received operation atom into a transaction record. Through simplified consensus exchange with other edge nodes or terminal devices within the region, the order in which these operation atoms were received by the edge shadow twin is confirmed across the entire network.

[0078] For example, in an office scenario, multiple tablets simultaneously edit the same project document offline. The modification operations generated by each device are broadcast in real time to the local enterprise server, i.e., the edge shadow twin. The consensus module on this server exchanges data with other backup servers or a few designated tablets in several rounds to form an immutable consensus result that contains the receiving order of all operation atoms. This result is essentially a time-series ledger jointly recognized by all nodes participating in the consensus.

[0079] Subsequently, the edge shadow twin packages and stores this consensus result, which includes the order information of the operation atoms, together with the original operation atom data. When the network connection is restored and the edge shadow twin re-synchronizes with the main workspace twin in the cloud, it uploads all the operation atoms accumulated during this period and their corresponding consensus results to the cloud.

[0080] After receiving this data, the cloud-based main workspace twin directly merges the operations according to the established order of operations in the consensus results, without the need for re-confrontational conflict detection and timing analysis. This ensures that during offline periods without centralized server coordination, all concurrent editing operations can still maintain a globally consistent and verifiable sequence.

[0081] The lightweight distributed ledger consensus module built into the edge shadow twin is specifically designed to handle concurrent atomic consensus confirmation of multiple terminals within a local area network during WAN outages. This module employs a lightweight consensus algorithm based on a directed acyclic graph structure, eliminating the need for complex computing power competition and achieving consensus solely through simple data exchange between multiple nodes. When multiple tablet terminals lose their WAN connection with the main workspace twin in the cloud but maintain a local area network connection with the edge shadow twin, the atomic operations generated by each terminal are uploaded to the edge shadow twin in real time. The consensus module packages each received atomic operation into an independent transaction record, writes it into the local distributed ledger structure, and broadcasts the transaction record to other backup edge nodes and participating terminal nodes within the local area network, receiving confirmation acknowledgments from each node. Once confirmation acknowledgments are received from more than half of the valid nodes, the consensus confirmation of the atomic operation reception order is completed, generating an immutable consensus result. This consensus result contains the unique identifier of the atomic operation, the reception sequence, and the list of confirming nodes, and is stored locally at the edge along with the corresponding atomic operation. Once the WAN is restored, the edge shadow twin uploads the operation atoms and corresponding consensus results to the cloud main workspace twin. The cloud directly merges the data based on the timing confirmed by the consensus results, without needing to re-perform conflict detection and timing analysis, ensuring that concurrent editing operations during offline periods have a globally consistent and verifiable order.

[0082] The intelligent semantic control engine is also used to calculate the user's operational risk value in the current environment in real time based on the multimodal cognitive graph; When the operational risk value exceeds the preset threshold, the intelligent semantic control engine automatically adjusts the user's file operation permissions on the tablet and sends a biometric authentication request to the tablet.

[0083] It should be further explained that, in the specific implementation, the system utilizes an intelligent semantic management engine deployed on a cloud-based main workspace twin, and based on the multimodal cognitive graph it constructs, performs real-time risk quantification assessment of every file access and editing behavior of the user on the tablet computer.

[0084] The engine continuously extracts multi-dimensional features related to the current operation from the cognitive graph, including the user's current geographical location, the type of network environment, the ambient light and motion status reported by the device's sensors, the user's operation frequency and error rate in the previous period, and the sensitivity tags contained in the file being operated on. These features are then input into a pre-set risk assessment model to calculate a dynamic operation risk value.

[0085] When the risk value rises to a preset threshold, for example, when the system detects that a user is in an unfamiliar public network environment and attempts to access a financial document with a highly sensitive label, and at the same time the analysis of the tablet's camera footage shows frequent movement of people behind the user, the intelligent semantic control engine will determine that there is a hidden danger in the current operating environment and automatically trigger the permission adjustment mechanism.

[0086] This mechanism does not directly cut off file access. Instead, based on the degree to which the risk value exceeds the threshold, it dynamically downgrades the user's file operation permissions on the tablet from "full editing" to "read-only browsing" or "disable printing and exporting." At the same time, the intelligent semantic control engine sends a biometric authentication request to the local terminal shadow twin on the tablet via an adaptive synchronization protocol. Upon receiving the request, the tablet immediately pops up an authentication window on the front-end interface, requiring the user to complete secondary authentication through fingerprint recognition or facial recognition.

[0087] Only after a user successfully passes biometric authentication will the intelligent semantic control engine restore the user's operating permissions to their original level and feed the risk event and authentication result as new behavioral data into the multimodal cognitive graph for subsequent optimization analysis of the user's behavior patterns.

[0088] The intelligent semantic control engine also calculates the risk value of the user's current operation in real time based on a multimodal cognitive graph and a built-in risk assessment model. The risk assessment model adopts a multi-feature weighted scoring algorithm. Input features include whether the user's current geographical location is a frequently used office location, whether the current network accessed is a trusted enterprise intranet, whether there are any anomalies in the surrounding environment detected by the device sensors, whether the user's current operation behavior is consistent with historical habits, the sensitivity level of the file being operated on, and the matching degree between the user's role permissions and file access permissions. The model assigns a corresponding weight value to each input feature. The weight values ​​are preset according to the enterprise's security management strategy. Among them, the file sensitivity level and the trustworthiness of the network environment have the highest weight. The model multiplies the quantitative score of each feature by its corresponding weight and then sums them to obtain the final operation risk value. The risk value ranges from 0 to 100, with higher values ​​indicating higher operation risks.

[0089] The system pre-sets two levels of risk thresholds: a warning threshold and a control threshold. When the calculated operational risk value exceeds the warning threshold, the system sends a risk warning message to the user's tablet terminal. When the operational risk value exceeds the control threshold, the intelligent semantic control engine automatically triggers a permission adjustment mechanism. Based on the magnitude of the risk value, the system dynamically downgrades the user's operation permission for the file from full editing permission to read-only browsing, prohibition of export, and prohibition of printing, respectively. Simultaneously, it sends a biometric secondary authentication request to the tablet terminal via an adaptive synchronization protocol. Upon receiving the request, the terminal immediately pops up an authentication window on the current interface and uses the tablet system's fingerprint and facial recognition modules to complete the secondary verification of the user's identity. Once the user successfully passes biometric authentication, the intelligent semantic control engine immediately restores the user's original operation permissions and simultaneously updates the risk event and authentication result to the multimodal cognitive graph for subsequent model optimization and user behavior pattern updates.

[0090] The intelligent semantic control engine is also used to detect the editing behavior of multiple users on the same file based on the multimodal cognitive graph; When at least two users are detected editing the same file simultaneously, the intelligent semantic control engine automatically creates a temporary virtual collaboration space and grants collaborative editing permissions to at least two users within the temporary virtual collaboration space.

[0091] It should be further explained that, in the specific implementation, the system utilizes an intelligent semantic management engine deployed on a cloud-based main workspace twin, based on its continuously constructed multimodal cognitive graph, to monitor the operational behavior of multiple users on the same file in real time.

[0092] This engine identifies whether at least two users are editing a file by analyzing the dynamic changes of the edges between nodes representing different users and nodes representing specific files in the cognitive graph, as well as the operation timing data attached to these nodes. When the engine detects that the editing operations of multiple users highly overlap in the time dimension and the operation positions point to the same paragraph or data area in the file content, the system automatically triggers the creation process of a temporary virtual collaboration space without relying on users to manually create a shared folder or send a collaboration invitation.

[0093] This temporary virtual collaboration space is a dynamically generated, logically isolated area in the cloud, designed to accommodate users participating in collaborative editing, copies of the files they are editing, and a real-time workflow. Simultaneously with space creation, an intelligent semantic control engine dynamically assigns appropriate collaborative editing permissions to users entering the space based on user roles, historical collaboration frequency, and the probability of conflict in the current editing behavior, analyzed from a multimodal cognitive graph.

[0094] For example, for two users from the same project team who have a history of frequent collaboration, the engine may grant them concurrent editing permissions for all parts of the file; while for a project member and an external visitor, the engine may only grant the visitor permission to add comments or track revisions, or only grant them access to specific paragraphs in the file that do not involve core secrets for editing.

[0095] During collaborative editing, all user editing operations are performed atomically and synchronized in real time within a temporary virtual collaboration space via an adaptive synchronization protocol, ensuring that each participant can see the changes made by other members. When the engine detects through its multimodal cognitive graph that all users have stopped editing the file and no new editing actions have occurred within a preset time, the temporary virtual collaboration space is automatically released, the cloud resources it occupied are reclaimed, the temporarily granted collaborative editing permissions expire, and the file status reverts to normal storage and management mode.

[0096] The intelligent semantic control engine monitors the editing behavior of multiple users on the same file in real time based on a multimodal cognitive graph. By analyzing the real-time changes of the edges connecting user nodes and file nodes in the graph, and combining the temporal data of user operations and the spatial coordinates of the operation locations, it identifies whether two or more users are simultaneously performing editing operations on the same file. When it detects that the editing operations of multiple users highly overlap within a preset time window and that the operation locations point to the same content area of ​​the file, the intelligent semantic control engine automatically triggers the creation process of a temporary virtual collaboration space. A logically isolated temporary virtual collaboration space is dynamically generated on the cloud server, the latest version of the file is synchronized to the space, and all users who performed editing operations are included in the collaboration space. Based on the user roles recorded in the multimodal cognitive graph, the permission association with the file, and historical collaboration records, corresponding collaborative editing permissions are dynamically assigned to each user. For example, the file creator and project leader are granted full content editing permissions, project participants are granted editing permissions for the paragraphs they are responsible for and full content comment permissions, and external collaborators are granted only comment permissions and editing permissions for specified paragraphs.

[0097] Within the temporary virtual collaboration space, all users' editing operations generate corresponding operation atoms in real time. Through an adaptive synchronization protocol, these atoms are synchronized in real time across all terminals within the space, ensuring that each user can see the editing content of other members in real time. At the same time, the editing areas and content of different users are marked on the interface to avoid editing conflicts.

[0098] The intelligent semantic control engine continuously monitors the editing behavior of all users within the collaboration space. When it detects that all users have stopped editing the file and there are no new editing or commenting operations within a preset time period, it automatically triggers the release process of the temporary virtual collaboration space. The final version of the file in the space is synchronized to the global storage of the main workspace in the cloud, the cloud resources occupied by the space are reclaimed, and the temporary collaborative editing permissions granted to users are revoked. The file is restored to the normal permission management mode. The preset time for stopping editing can be configured by the enterprise administrator according to the office needs. The default configuration is 30 minutes. If a user initiates editing operations on the file again before the space is released, the system will automatically reactivate the temporary virtual collaboration space without having to create it again.

[0099] The intelligent semantic control engine is also used to detect the matching degree between the current task and the state of the user's terminal device based on the prediction results of the user's operation intention; When the matching degree is lower than the preset value, the intelligent semantic control engine generates a task migration instruction to synchronize the editing status of the current task to other terminal devices associated with the user.

[0100] It should be further explained that, in the specific implementation, the system utilizes the intelligent semantic management engine of the main workspace twin deployed in the cloud, and based on the prediction results of the user's operation intentions from the multimodal cognitive graph it constructs, performs a matching degree analysis on the task currently being performed by the user and the state of the user's terminal device.

[0101] The engine continuously receives device status data uploaded by the local terminal shadow twin of the tablet computer, including but not limited to the remaining battery power, the signal strength and stability of the current network connection, and the real-time load of the device's central processing unit. It combines this data with the prediction results of the user's historical behavior patterns extracted from the multimodal cognitive graph and the current operation intention to comprehensively determine whether the current task is suitable to continue to be executed on the current device.

[0102] When the engine detects that the match between the user's document editing task and the device status is lower than a preset value, such as when the user is writing a long report with many pictures on a tablet, but the device battery is about to run out and there are no charging facilities nearby, and the engine predicts through cognitive graph that the user may need to show the first draft of the report to the team within an hour, the intelligent semantic control engine determines that the current state of the tablet cannot support the user to complete the expected task.

[0103] Based on this determination, the engine automatically generates a task migration instruction, which includes the unique identifier of the current task, the latest version information of the document, and the user's precise editing position in the current document. Subsequently, the engine sends this instruction to other terminal devices that the user has pre-registered and are online, such as the user's office personal computer, via an adaptive synchronization protocol.

[0104] Upon receiving a task migration instruction, the office PC automatically retrieves the corresponding document from its cloud-based main workspace twin and opens it in the same editing location as on the tablet. Simultaneously, it restores all unsaved changes made by the user on the tablet, allowing the user to seamlessly continue their previous work on the PC. After the task migration is complete, the local terminal shadow twin on the tablet marks the relevant data for the current task as migrated and retains a read-only copy locally for the user to review.

[0105] The intelligent semantic management engine detects the matching degree between the user's current office task and the current terminal device status in real time based on the prediction results of the user's operation intent. The matching degree detection adopts a multi-dimensional comprehensive evaluation mechanism. The evaluation indicators include the terminal device's remaining battery power, the stability and bandwidth of the current network connection, the real-time load of the device's processor, the adaptability of the device's screen size to the current task type, and the device's remaining storage space. Combined with the predicted task duration and resource requirements for task processing from the prediction results of the user's operation intent, the matching degree score between the current task and the terminal device is calculated. The matching degree score ranges from 0 to 100, with a higher score indicating a higher matching degree.

[0106] The system pre-sets a matching threshold. When the calculated matching score is lower than the preset threshold, the intelligent semantic control engine determines that the current terminal device cannot support the user to complete the current office task. It automatically generates a task migration instruction, which includes the unique file identifier corresponding to the current task, the latest edited version of the file, the user's current editing position, unsaved editing operations, and interface layout status information. At the same time, it searches for other terminal devices that are pre-bound to the user account and are online, including office computers, laptops, and other mobile terminals, and sends the task migration instruction to the target terminal device through an adaptive synchronization protocol.

[0107] After receiving the task migration instruction, the target terminal device automatically retrieves the latest version of the corresponding file from the cloud, opens the file in the corresponding application, and automatically jumps to the editing position that is completely consistent with the original terminal, restoring all the user's editing state, interface layout, and unsaved changes, so that the user can seamlessly continue the office tasks on the target terminal. After the task migration is completed, the terminal shadow twin of the original tablet terminal will mark the relevant data of the task as migrated, retain a read-only copy of the file locally, and send a prompt message to the user that the task migration is complete.

[0108] The cloud-based main workspace twin, the edge shadow twin, and the terminal shadow twin all store the same core data model; The core data model is used to uniformly represent file content, user operation behavior, and contextual information as data atoms. Data atoms include at least the timestamp dimension, spatial coordinate dimension, and semantic tag dimension. The adaptive synchronization protocol is used to synchronize incremental changes of data atoms among the main workspace twin, edge shadow twin, and terminal shadow twin in the cloud.

[0109] It should be further explained that, in the specific implementation, the system configures the same core data model for the cloud main workspace twin, the edge shadow twin, and the terminal shadow twin to ensure that the data can be uniformly understood and processed among the three.

[0110] The core of this data model is to abstract and encapsulate all information that needs to be managed and synchronized in the system, whether it be specific file content, user operations, or device environmental status, into data atoms. Each data atom is an independent, indivisible data unit, and its internal structure contains at least three dimensions of information to describe the attributes of that data.

[0111] The first dimension, timestamp, records the precise time of the event represented by the data atom, such as the moment a user completes a keyboard input or the moment a sensor collects a location data point. The second dimension, spatial coordinates, locates the specific position of the data atom within its file or virtual space. For document content, this coordinate could be a paragraph number, a table cell coordinate, or the character offset of a piece of text within the entire document; for user behavior, it could be the two-dimensional coordinates of a touch event on the screen. The third dimension, semantic tagging, assigns a business meaning to the data atom, such as "insert text," "delete image," "move cursor," or "device tilt," enabling the data atom itself to carry business logic that the system can understand.

[0112] Based on the aforementioned unified core data model, the cloud main workspace twin, edge shadow twin, and terminal shadow twin no longer synchronize the entire file or complex data structure during operation. Instead, they synchronize only the incremental changes of the data atoms they generate or receive through an adaptive synchronization protocol.

[0113] For example, when a user modifies a document on a tablet, the terminal shadow twin only sends the data atoms describing the modification and their incremental changes, such as "insert text Y at coordinate X," to the edge shadow twin or the cloud main workspace twin via an adaptive synchronization protocol. Upon receiving this data atom, the receiver, based on its timestamp, spatial coordinates, and semantic tags, systematically merges it into its local core data model copy, thus achieving eventual consistency across the three at the data level without requiring a full file transfer each time.

[0114] This system constructs a three-tiered collaborative digital twin architecture—a cloud-based main workspace twin, an edge shadow twin, and a terminal shadow twin—and configures a unified spatiotemporal data atomic encapsulation standard for the core data model. This enables unified expression and precise synchronization of file content and user operations across different network environments. The system's operation atomic processing mechanism and version control structure based on a directed acyclic graph (DAG) allow edited data generated in network-off-grid environments to carry complete temporal, spatial, and semantic information. After network recovery, the spatiotemporal alignment capability of the edge shadow twin achieves accurate merging with cloud data, avoiding data version chaos caused by concurrent editing. The construction of a multimodal cognitive graph and the deployment of a spatiotemporal graph convolutional network enable the system to understand office scenarios from multiple dimensions, including user behavior, device status, and environmental data, providing a data foundation for predicting user operation intentions and dynamically adjusting resources.

[0115] The system ensures that offline edited content remains completely consistent with the cloud version after network recovery, eliminating the need for manual version conflict resolution. Operation intent prediction commands based on cognitive graphs enable tablets to preload content before user actions are initiated, reducing response time for file opening and interface switching. Dynamic risk value calculation and operation permission adjustment mechanisms provide adaptive protection for sensitive files in different usage environments without disrupting normal office workflows. Automatic creation and release of temporary virtual collaboration spaces simplify the startup process for multi-user collaborative editing, reducing manual configuration operations. Task and terminal device status matching analysis and task migration command generation allow users to seamlessly switch between different devices, maintaining workflow continuity. A unified data atomic model and incremental synchronization protocol reduce data transmission volume between the cloud, edge nodes, and terminal devices, improving the overall system's operational stability in weak network environments.

[0116] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0117] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A tablet-based digital office document collaborative management system, characterized in that, This includes a cloud-based main workspace twin, an edge shadow twin, and a terminal shadow twin; The cloud-based main workspace twin is deployed on a cloud server and is used to construct a global knowledge graph and model user behavior. The edge shadow twin is deployed on the enterprise's local server or mobile edge computing node to provide temporary data collaborative arbitration services in environments with weak network or no network coverage. The terminal shadow twin is deployed locally on the tablet computer to collect user interaction data, generate local operation logs, and preload file content according to the prediction instructions issued by the cloud main workspace twin. The cloud-based main workspace twin, the edge shadow twin, and the terminal shadow twin maintain data consistency through an adaptive synchronization protocol.

2. The tablet-based digital office document collaborative management system according to claim 1, characterized in that: The system also includes an offline collaborative consistency engine; The offline collaborative consistency engine is set locally on the tablet computer and is used to generate operation atoms when the network is disconnected, and to sign and sort the operation atoms. After the network is restored, the edge shadow twin receives the operation atoms, performs spatiotemporal alignment on the operation atoms from different terminals, and merges the aligned data into the cloud main workspace twin. The operation atom is a data unit carrying a timestamp, spatial coordinates, semantic tags and operator digital signature, and the file content is composed of at least one of the operation atoms.

3. The tablet-based digital office document collaborative management system according to claim 2, characterized in that: The system also includes an intelligent semantic control engine; The intelligent semantic management engine is set in the main workspace twin in the cloud and is used to construct a multimodal cognitive graph; The nodes of the multimodal cognitive graph include file content, user roles, tablet sensor data, operation timing data, and environmental data. The edges of the multimodal cognitive graph are used to represent the relationships between the nodes. The intelligent semantic control engine predicts the user's operational intent through a spatiotemporal graph convolutional network based on the multimodal cognitive graph, and generates the prediction instruction based on the prediction result and sends it to the terminal shadow twin.

4. The tablet-based digital office document collaborative management system according to claim 3, characterized in that: The system also includes an adaptive touch twin rendering engine; The adaptive touch twin rendering engine is set locally on the tablet computer and is used to generate a three-dimensional interactive interface that conforms to the current ergonomics in real time based on the holding posture, screen size or ambient light of the tablet computer. The adaptive touch twin rendering engine is also used to build a virtual document object model in the tablet computer's memory in advance according to the prediction instructions issued by the intelligent semantic management engine, and to complete the rendering of the next screen content before the user performs a swipe operation.

5. The tablet-based digital office document collaborative management system according to claim 2, characterized in that: The offline collaborative consistency engine is also used to record the operation atoms as a version control structure based on a directed acyclic graph, where each node in the version control structure corresponds to one operation atom, and the edges in the version control structure are used to represent the temporal relationship or dependency relationship between the operation atoms.

6. The tablet-based digital office document collaborative management system according to claim 2, characterized in that: The edge shadow twin is also used to receive the operation atoms generated by multiple tablet terminals during network disconnection, and to perform consensus confirmation on the receiving order of the operation atoms through lightweight distributed ledger technology to generate a consensus result; After the network is restored, the edge shadow twin sends the operation atoms and the consensus results to the cloud main workspace twin.

7. The tablet-based digital office document collaborative management system according to claim 3, characterized in that: The intelligent semantic control engine is also used to calculate the user's operational risk value in the current environment in real time based on the multimodal cognitive graph; When the operational risk value exceeds a preset threshold, the intelligent semantic control engine automatically adjusts the user's file operation permissions on the tablet computer and sends a biometric authentication request to the tablet computer.

8. The tablet-based digital office document collaborative management system according to claim 3, characterized in that: The intelligent semantic control engine is also used to detect the editing behavior of multiple users on the same file based on the multimodal cognitive graph; When at least two users are detected editing the same file simultaneously, the intelligent semantic control engine automatically creates a temporary virtual collaboration space and grants collaborative editing permissions to the at least two users within the temporary virtual collaboration space.

9. A tablet-based digital office document collaborative management system according to claim 3, characterized in that: The intelligent semantic control engine is also used to detect the matching degree between the current task and the state of the user's terminal device based on the prediction result of the user's operation intention. When the matching degree is lower than a preset value, the intelligent semantic management engine generates a task migration instruction to synchronize the editing status of the current task to other terminal devices associated with the user.

10. A tablet-based digital office document collaborative management system according to claim 1, characterized in that: The cloud-based main workspace twin, the edge shadow twin, and the terminal shadow twin all store the same core data model; The core data model is used to uniformly represent the file content, user operation behavior and context information as data atoms. The data atoms include at least the timestamp dimension, spatial coordinate dimension and semantic tag dimension. The adaptive synchronization protocol is used to synchronize the incremental changes of the data atoms among the cloud main workspace twin, the edge shadow twin, and the terminal shadow twin.

Images