Intelligent office digital interaction method and system based on adaptive decision

Abstract

The application discloses an intelligent office digital interaction method and system based on adaptive decision, relates to the technical field of human-computer interaction, and through real-time monitoring of signal strength in a working environment, makes the system capture a time point at which a current task situation may change in real time, and improves system response immediacy; through multidimensional quantification of the task situation when it changes, and joint determination of the current task intention by using historical data weight and user habit preference, makes the system automatically and accurately perceive the working stage and task intention of the user, and realizes dynamic matching of the current task situation and the past operation habit of the user; through task intention description, an adjustment instruction sequence is derived, an initial adjustment scheme is regenerated, and finally the initial adjustment scheme is optimized according to the user habit preference, thereby effectively connecting the historical behavior and the current task situation, making the system quickly make an interface adjustment decision most conforming to the user intention under the current situation, and significantly improving operation efficiency and comfort.

Description

Technical Field

[0001] This invention relates to the field of human-computer interaction, and in particular to an intelligent office digital interaction method and system based on adaptive decision-making. Background Technology

[0002] With the deepening of digital transformation, intelligent office systems have become core tools for improving organizational efficiency, and the quality of their interactive experience directly determines office efficiency and user satisfaction. In this field, a key challenge in achieving intelligent upgrades is how to enable systems to understand and adapt to the different work needs of users, much like a human assistant. Currently, many systems offer rich functionality, but their interaction modes are often preset and fixed, unable to dynamically adjust according to the user's specific task context. For example, an interface layout designed for routine administrative tasks may become ineffective when the user transitions to creative planning requiring divergent thinking; the original simple interface may limit the accessibility of tools, forcing the user to perform numerous tedious searches and switching operations.

[0003] Existing methods also fail to accurately determine the most efficient interaction configuration when faced with complex tasks that change suddenly or periodically, often making one-sided or delayed decisions. For example, they may fail to pre-configure key functions when users urgently need to process batch reports, or display unnecessary shortcut prompts when users are engaged in in-depth reading. This rigid interaction design ignores the dynamic matching relationship between task type and user operating habits, resulting in unstable system efficiency in actual applications.

[0004] This technical challenge has led to a specific contradiction in actual business operations: frequent mismatches have arisen between the general interactive processes provided by the system and the personalized, real-time work needs of users. Users must constantly manually adjust system settings to adapt to different tasks, and this process itself becomes a point of inefficiency, disrupting the natural flow of work. Summary of the Invention

[0005] This invention provides an intelligent office digital interaction method and system based on adaptive decision-making. It abandons the preset and fixed interface configuration method in traditional solutions. By automatically sensing the user's work stage and task intent, it dynamically adjusts according to changes in the current specific task context, optimizes interface elements in real time, and achieves dynamic matching between the current task context and the user's operating habits. This improves the system's ability to collaboratively process the continuity of user behavior and the variability of task contexts, significantly enhances the personalization and real-time nature of the interaction mode, and comprehensively optimizes the user experience and work efficiency in office scenarios.

[0006] This invention provides an intelligent office digital interaction method based on adaptive decision-making, executed by a computer, comprising: Real-time monitoring of signal strength in the current dynamic context, wherein the signal strength is represented by at least external environmental signals and / or user operation signals; If the signal strength meets the preset situation change conditions, then multi-dimensional quantization is performed based on the signal strength to generate a situation feature vector; Based on the context feature vector, the weights of pre-collected historical data, and user habit preferences, a user task intent description corresponding to the signal strength is determined. Based on the user task intent description, an adjustment instruction sequence is generated, and the adjustment instruction sequence is transformed into an initial adjustment scheme for interface elements by combining the instruction execution sequence and the granularity of the adjustment instructions. Based on the user's habits and preferences, the initial adjustment scheme of the interface elements is optimized to generate interface optimization adjustment parameters, so as to adaptively adjust the current interface based on the interface optimization adjustment parameters.

[0007] This invention provides an intelligent office digital interaction method and system based on adaptive decision-making. By monitoring the signal strength from the external environment and / or user's autonomous operations in the work environment in real time, the system can capture the time points when the current specific task situation may change, improving the immediacy of the system response. When the signal strength reflects a change in the current specific task situation, it quantifies it in multiple dimensions and combines it with the user's historical behavior patterns gradually formed in a series of previous coherent work processes to jointly determine the user's current task intention. This allows the system to automatically and accurately perceive the user's work stage and task intention, achieving dynamic matching between the current task situation and the user's past operating habits. The system derives an adjustment instruction sequence from the user's task intention description, generates an initial adjustment plan from the adjustment instruction sequence, and finally optimizes the initial adjustment plan based on historical behavior preferences containing stable user behavior patterns. This effectively connects the user's historical behavior with the current task situation, enabling the system to quickly make interface adjustment decisions that best match the user's intention in the current situation. This solves the problem of frequent mismatches between the general interaction process provided by existing systems and the personalized, real-time work needs of users, and the inefficiency caused by delayed behavior response. Attached Figure Description

[0008] Figure 1 This is one of the flowcharts illustrating an intelligent office digital interaction method based on adaptive decision-making provided in an embodiment of the present invention; Figure 2 This is a second flowchart illustrating an intelligent office digital interaction method based on adaptive decision-making provided in an embodiment of the present invention. Figure 3 This is the third flowchart of an intelligent office digital interaction method based on adaptive decision-making provided in this embodiment of the invention; Figure 4 This is the fourth flowchart of an intelligent office digital interaction method based on adaptive decision-making provided in an embodiment of the present invention; Figure 5 This is the fifth flowchart illustrating an intelligent office digital interaction method based on adaptive decision-making provided in this embodiment of the invention. Figure 6 This is the sixth flowchart of an intelligent office digital interaction method based on adaptive decision-making provided in this embodiment of the invention; Figure 7 This is the seventh flowchart of an intelligent office digital interaction method based on adaptive decision-making provided in an embodiment of the present invention. Detailed Implementation

[0009] 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.

[0010] Reference Figure 1 This invention provides an intelligent office digital interaction method based on adaptive decision-making, comprising the following steps: Step 100: Monitor the signal strength in the current dynamic situation in real time, wherein the signal strength is represented by at least external environmental signals and / or user operation signals; This step aims to improve the immediacy of system response, enabling the system to capture in real-time changes in the current task context. Many traditional office systems, while offering rich functionality, often have preset and fixed interaction patterns, unable to dynamically adjust based on the user's current task context. The underlying reason for this limitation lies in the system's lack of ability to collaboratively process the two closely related core attributes: "user behavior continuity" and "task context variability." User behavior continuity refers to the gradual formation and manifestation of user operating habits and preferences through a series of coherent work processes, while task context variability refers to the potentially drastically different work goals and content requirements that users may face at different points in time. How to enable the system to automatically perceive the user's work stage and task intent, and accordingly optimize interaction paths and interface elements in real-time, thereby maintaining a high level of operational efficiency and comfort in changing task scenarios, has become a key issue in enhancing the practical value of intelligent office systems.

[0011] In this step, the system monitors the current external environmental signal strength and / or user operation signal strength in real time, thereby enabling it to promptly capture the moment when the user intends to change the current work task, avoiding the drawbacks of traditional office system models that cannot respond to user task intention changes in a timely manner.

[0012] External environmental signals refer to signals corresponding to the objective attributes of the external environment, such as temperature, humidity, light intensity, environmental noise level, time, geographical coordinates, equipment type, etc., which can be obtained through devices such as temperature sensors, humidity sensors, light sensors, sound level meters or noise detectors, or by accessing log records.

[0013] User operation signals refer to the signals corresponding to the interactive operations performed by the user on the system, such as the number of clicks, cursor movement trajectory, page dwell time, application switching frequency, application program interface (API) call sequence, etc., and may also include text information entered by the user.

[0014] Signal strength refers to the fluctuation intensity corresponding to external environmental signals and / or user operation signals, which can be obtained through signal analysis under preset conditions. This fluctuation intensity reflects the probability of contextual fluctuations and task switching during user interaction with the office system.

[0015] Real-time monitoring can be achieved by the system periodically collecting the aforementioned signals at preset short initial time intervals, or by the system calculating one or more time intervals based on the user's past operation data to match the user's usage habits, and then flexibly collecting the signals according to these time intervals. Furthermore, if the system detects a significant change in the user's usage habits, it can automatically adjust the real-time monitoring frequency based on the user's real-time feedback to further improve work efficiency and enhance the user experience.

[0016] Step 200: If the signal strength meets the preset situation change conditions, then perform multi-dimensional quantization based on the signal strength to generate a situation feature vector; This step is used to determine the timing of contextual fluctuations and the trigger points for task switching. Preset contextual change conditions refer to the conditions that determine when the current user task has changed, requiring adjustments to the interface elements. Specifically, this can be set as follows: using real-time analysis methods to calculate the strength values ​​of external environmental signals and / or user operation signals, determining the specific manifestation of signal strength. If the signal strength value exceeds a preset adaptation threshold, it is determined that the preset contextual change condition is met. Alternatively, it can be set as follows: after calculating that the signal strength value exceeds the preset adaptation threshold, historical behavior data is obtained to determine whether data weight adjustments are needed. If so, based on historical behavior data and the current signal strength, the data weight ratio is calculated, and behavior fusion processing is completed to obtain a comprehensive behavior feature value. Based on the comprehensive behavior feature value and the specific background of the task context, the variability assessment level of the task context is determined. Finally, based on the variability assessment level, it is determined whether interface element adjustments are needed. Additionally, this contextual change condition can also be set by the user according to their own needs.

[0017] Multidimensional quantization refers to quantification based on two or more dimensions, such as heterogeneous characteristics, behavioral response latency, and historical behavioral data weights.

[0018] When the system detects external environmental signals and / or user operation signals, it processes the data to obtain a standardized signal dataset. Using this standardized dataset, the system calculates the signal strength value using real-time analysis methods. It then determines whether the signal strength value meets preset context change conditions. If the signal strength meets these conditions, a dynamic context detection mechanism is triggered, indicating a user need to adjust interface elements. The system continues to collect current external environmental signals and / or user operation signals, converting them into context feature vectors. If the signal strength value does not meet the preset context change conditions, the system determines that the current context fluctuation intensity is insufficient to trigger the dynamic context detection mechanism, and interface elements do not require adjustment at this time.

[0019] Specifically, the system collects environmental data in real time through sensors, such as temperature and humidity, obtaining the current task input signal as a temperature of 28.5 degrees Celsius and a humidity of 65%. First, the system calculates the signal strength S using the signal strength calculation formula: (Where T is temperature and H is humidity) Calculate the real-time signal strength, and substitute the values ​​to obtain: .

[0020] Next, the system compares the signal strength with a preset scenario adaptation threshold of 40. It finds that 43.1 is greater than 40, indicating that the signal strength is above the threshold, thus allowing for further integration of historical behavior data weights. The historical behavior data weights are calculated based on the temperature and humidity trends over the past week. Assuming a temperature weight of 0.7 and a humidity weight of 0.3, the system combines the current signal strength with the historical behavior data weights to calculate the weighted signal value W. , Subsequently, the system evaluates task situation variability based on the weighted signal value W, and uses a variability evaluation formula to evaluate task situation variability V: , Right now , The scenario variability is 7.75%, falling within the mildly variable range (0-10% is considered mild). Within this range, the system can determine that the trigger point for task switching and scenario changes has not yet arrived, eliminating the need for immediate adjustments to interface elements. Close monitoring of input signals is still possible. To ensure logical rigor, the system also considers business scenarios. For example, if the variability exceeds 5%, the system automatically increases the environmental monitoring frequency from once per hour to once every 30 minutes to dynamically adapt to scenario changes. The data analysis and calculation processes described above rely on real-time processing by the backend system, ensuring accurate results and logical consistency. Signal strength and weighting are integrated into the variability assessment to form a complete chain.

[0021] Step 300: Based on the context feature vector, the weights of pre-collected historical data, and user habit preferences, determine the user task intent description corresponding to the signal strength; In this step, the pre-collected historical data weights and user habit preferences refer to the data weights and habit preferences summarized by the system through analysis of users' continuous operational behavior data over a period of time. The user's task intent description can be directly derived by the system through the analysis of contextual feature vectors, pre-collected historical data weights, and user habit preferences, or it can be further derived by combining it with the text information directly input by the user.

[0022] Specifically, the system obtains basic information about user behavior patterns by comparing contextual feature vectors with historical data weights, thus determining preliminary predictive intent. Based on the preliminary predictive intent, in-depth analysis is performed in conjunction with user habits and preferences. If the deviation of the predicted intent exceeds a preset threshold, the weight distribution of the feature vector comparison is readjusted to obtain more accurate predictive intent data. For the obtained predictive intent data, relevant patterns are extracted from the user's historical behavior records. If the matching degree between the extracted pattern and the current contextual feature vector is lower than a preset threshold, the specific direction of the deviation is analyzed. After obtaining the specific direction of the deviation, the current user's true intent is determined based on the specific direction of the deviation, generating a task intent description as the user's task intent corresponding to the signal strength that satisfies the contextual change conditions, thus completing the closed-loop processing of the entire prediction process.

[0023] Step 400: Generate an adjustment instruction sequence based on the user task intent description, and transform the adjustment instruction sequence into an initial adjustment scheme for interface elements by combining the instruction execution timing and adjustment instruction granularity. In this step, the adjustment instruction sequence refers to the sequence of instructions for the interface elements that need to be adjusted. The system can generate the adjustment instruction sequence directly based on the user task intent description obtained in the previous step, or it can generate the adjustment instruction sequence based on the text information that the user is currently inputting to express the task intent, or it can generate the adjustment instruction sequence by combining the user task intent obtained in the previous step with the text information input by the user.

[0024] Specifically, the system parses the user's task intent description to obtain core requirement information and determine the classification criteria for intent tags. Based on the classification criteria and priority order, the system judges the importance of each tag and obtains a priority ranking result. For the priority ranking result, a corresponding adjustment instruction sequence is generated, and the execution order of the instructions is determined using an instruction sequence arrangement method. Based on the instruction execution sequence, specific parameters for the adjustment granularity are obtained, and the adjustment instruction sequence is layered to obtain a layered instruction set. Based on the layered instruction set and the attribute information of the interface elements, a preliminary interface layout is generated as the initial adjustment scheme for the interface elements.

[0025] Step 500: Optimize the initial adjustment scheme of the interface elements based on the user's habits and preferences, generate interface optimization adjustment parameters, and adaptively adjust the current interface based on the interface optimization adjustment parameters.

[0026] In this step, the system can determine the interface optimization adjustment parameters based on user habits and preferences, or it can determine the interface optimization adjustment parameters by combining preset values ​​of user habits and preferences and element interaction weights.

[0027] Specifically, the system acquires basic data of interface elements and analyzes user operation frequency and preference data from previous interface adjustments to extract key patterns and determine the preference weights of user habits. By combining the preset values ​​of user habit preference weights and interaction weights, the system performs comprehensive calculations based on the direction of interface adjustments, categorizes the results of the weight analysis, and obtains an optimized allocation scheme for interaction weights. Based on this optimized allocation scheme, the system combines the weight analysis results with the priority of element distribution, generates specific values ​​using a parameter mapping tool, and thus determines the interface element adjustment parameters. The system then adjusts the current interface elements according to these parameters to ensure that the interface layout is adapted to the current task context.

[0028] It should be noted that in dynamic scenarios, the external environmental signals and / or user operation signals monitored by the system in real time may change at any time, and the corresponding signal strength will also change accordingly. When the system detects a change in signal strength, it can dynamically adjust the context feature vector and then adapt the interface optimization adjustment parameters accordingly. This completes the closed-loop mechanism of dynamic adjustment of context feature vector and intent prediction, which significantly improves the personalization and real-time performance of the interaction mode and comprehensively optimizes the user experience and work efficiency in office scenarios.

[0029] The intelligent office digital interaction method based on adaptive decision-making provided by this invention abandons the traditional preset and fixed interface configuration method. By monitoring the signal strength from the external environment and / or user's autonomous operation in the work environment in real time, the system can capture the switching time points when the current specific task situation may change, improving the immediacy of the system response. When the signal strength reflects a change in the current specific task situation, it quantifies it in multiple dimensions and combines it with the user's historical behavior patterns gradually formed in a series of previous coherent work processes to jointly determine the user's current task intention. This allows the system to automatically and accurately perceive the user's work stage and task intention, and can adjust its response according to the current specific task situation. The system dynamically adjusts to match the current task context with user operating habits. It derives an adjustment instruction sequence from the user's task intent description, generates an initial adjustment plan from this sequence, and finally optimizes the initial adjustment plan based on historical behavioral preferences that reflect stable user behavior patterns. This effectively connects the user's historical behavior with the current task context, enhancing the system's ability to collaboratively handle the continuity of user behavior and the variability of task contexts. It ensures that the interface layout adapts to the task context, resolving the frequent mismatch between the general interaction flow provided by existing systems and the personalized, real-time work needs of users. This allows the system to maintain a high level of operational efficiency and comfort in ever-changing task contexts.

[0030] In one embodiment, please refer to Figure 2 Before step 500, the following are also included: Step 001: Collect continuous user operation data, perform feature analysis on the continuous user operation data, and obtain task switching density and context fluctuation frequency. Step 002: Based on the task switching density and context fluctuation frequency, construct context dynamic change features that reflect the continuity of user historical behavior; In this step, continuous user operation data refers to the user's operation data throughout a series of consecutive tasks. The system continuously captures and understands the stable patterns inherent in the user behavior sequence by analyzing this operation data. Contextual dynamic change characteristics refer to the feature descriptions of dynamic changes in the context, used to reflect the changing trends of user behavior under different contexts.

[0031] The system can pre-obtain continuous user operation data from user interaction records, perform preliminary cleaning and formatting on this data to obtain a structured behavior dataset. The system then uses time series analysis on this structured dataset, labeling the occurrence times of contextual fluctuations and task switching to determine the time points of contextual fluctuations and the trigger points of task switching. If the labeled time points show a contextual fluctuation frequency higher than a preset threshold, these time points are clustered to obtain high-frequency intervals of contextual fluctuations and identify areas of dense fluctuation. The system performs in-depth data mining within these densely fluctuating areas to extract user behavior response patterns under different contexts, obtaining a correlation pattern between behavior and context. Based on this correlation pattern, the density of task switching is segmented to obtain the distribution of task switching intensity, determining the regularity characteristics of switching behavior. If the regularity characteristics of switching behavior show a periodic distribution, the system divides these characteristics into time windows to obtain the range of periodic switching intervals, determining the continuity of user behavior. By integrating the data of continuous behavior, the system constructs a feature description of dynamic contextual changes, deriving the changing trends of user behavior under different contexts.

[0032] Specifically, the system uses information technology to collect and analyze user behavior data. First, it automatically collects user operation data on mobile devices using sensors and logging systems. For example, it records the number of clicks, page dwell time, and application switching frequency on a particular application. Assuming that a user clicks 150 times on a social media application in a day, with an average page dwell time of 30 seconds and 80 application switches, then behavioral features are extracted from this data. Feature engineering methods can be used to standardize the number of clicks, dwell time, and switching frequency, and calculate the density of user operations, for example, using the formula: , The density is calculated as follows: , This reflects the activity level of user operations. Subsequently, the frequency of contextual fluctuations and task switching density are analyzed. Time series analysis algorithms, such as Fourier transform, can be used to calculate the frequency of user behavior fluctuations within a day. Assuming the analysis shows a fluctuation frequency of 3.5 times per hour, the task switching density is obtained by dividing the number of switches by the total time. , This indicates that users switch tasks frequently. Further, a continuous description of user behavior can be constructed. A Hidden Markov Model (HMM) can be used to model user behavior states. Assuming the state transition probability matrix shows a probability of 0.25 for a user switching from a focused state to a distracted state, a continuous description curve can be generated by combining time series data to reflect the stability of user behavior. Finally, the dynamic change characteristics of the context are obtained. A clustering algorithm is used to classify the fluctuation frequency and switching density. Assuming user behavior is divided into two categories—high dynamic (frequency greater than 3.0 times / hour) and low dynamic—the current user is classified as belonging to the high dynamic category. In addition, the system can be combined with business scenarios such as online learning platforms to further analyze the potential decline in learning efficiency of highly dynamic users due to distraction, and push focus reminder functions, thus forming a complete logical chain from data collection to feature analysis and then to business application.

[0033] Step 500 includes: Step 501: Determine the interface adjustment preference weight based on the user's habit preferences, and perform weight analysis by combining the preset interaction weight with the interface adjustment preference weight to obtain the weight analysis result; Step 502: Optimize the initial adjustment scheme of the interface elements according to the weight analysis results to obtain the interface optimization adjustment parameters; In this step, the preset interaction weights refer to the weight values ​​that the system pre-assigns to each element in the interface based on the user's work needs in daily office scenarios. The system pre-acquires relevant records of user habits, such as historical behavior logs, analyzes the frequency of user operations and preference data during interface adjustments, and sets preference weights that reflect user habits and preferences. The system performs weight analysis by combining the user habit preference weights with the preset interaction weight values, obtaining the weight analysis results. It then uses classification algorithms such as Support Vector Machines to classify the weight analysis results, resulting in an optimized allocation scheme for the interaction weights. Based on this optimized allocation scheme, the weight analysis results are combined with the priority of element distribution, and specific values ​​are generated using parameter mapping tools. These values, combined with the initial interface element adjustment scheme, determine the final interface optimization adjustment parameters.

[0034] Step 503: Adjust the interface optimization parameters according to the dynamic change characteristics of the scenario to obtain the target interface layout scheme; Step 504: Determine dynamic interaction path signals according to the target interface layout scheme, so as to adaptively adjust the current interface through the dynamic interaction path signals.

[0035] In this step, the dynamic interaction path refers to the interaction path predicted by the system based on the target interface layout scheme, which conforms to the user's task context. The dynamic interaction path signal refers to the type of input signal that the system determines needs to monitor based on the aforementioned dynamic interaction path, and monitors dynamic input signals that match this signal type, such as click count, cursor movement trajectory, and response time. The system calls a pre-established parameter mapping table to determine the interface adjustment strategy and optimized parameter combination corresponding to the dynamic changes in the context. If the determined optimized parameter combination deviates from the current system configuration, an adjustment instruction is generated for the deviation, and the adjusted interface layout scheme, i.e., the target interface layout scheme, is obtained. The element layout of the current interface is then adjusted according to the target interface layout scheme to improve the user's operational efficiency.

[0036] Subsequently, the system, in conjunction with the target interface layout scheme, predicts interaction path planning that aligns with the user's task context based on task switching density and context fluctuation frequency, and determines whether this interaction path covers the main task scenarios required for the user's work. If the interaction path covers the main task scenarios, the corresponding interaction signal type is determined based on this path planning, and input signals matching this type are monitored as dynamic interaction path signals. Then, based on the dynamic interaction path signals, the system's interaction mode is updated in real time to complete the adaptation processing of the mode configuration, thereby improving the real-time performance and adaptability of the interaction mode configuration in the user's current task context. Specifically, the system optimizes and adjusts parameters through interface design to adapt to task switching density and context fluctuation frequency, predicting interaction paths suitable for the user's task context. First, the system analyzes user operation logs to calculate task switching density. Assuming 5 task switches per minute, exceeding the threshold of 3 switches per minute is considered high-density switching. The system automatically adjusts interface layout parameters, reducing the response time of frequently used functional modules from the default 0.5 seconds to 0.3 seconds to improve operational efficiency. The analysis process is based on historical data statistics, using a sliding window algorithm to calculate the switching frequency to ensure real-time performance.

[0037] Secondly, in response to the frequency of situational fluctuations, the system collects environmental sensor data. For example, if the light intensity changes by more than 50 lux every 10 minutes, it is determined to be a high-fluctuation situation. The system automatically adjusts the screen brightness parameters and smoothly transitions from the current brightness of 300 nits to the target brightness of 500 nits through a linear interpolation algorithm. The fluctuation trend is analyzed to avoid frequent adjustments that may cause user discomfort.

[0038] Subsequently, based on the aforementioned task density and contextual fluctuation frequency, the system employs a decision tree algorithm, combined with a user behavior preference database, to predict the optimal interaction path. For example, when task density is high and fluctuation frequency is large, a simplified interface adapted to the target interface layout scheme is prioritized for dynamic interaction path planning. This reduces the number of visual elements from 20 to 10, lowering cognitive load. The analysis process calculates path selection probabilities to ensure signal generation accuracy. Finally, based on this dynamic interaction path planning, the system determines the dynamic interaction path signals that need to be monitored, such as user response intervals and cursor movement trajectories. By monitoring these dynamic interaction path signals, the system updates interface optimization parameters in real time, such as increasing touch sensitivity from the default value of 50 to 70. Additionally, a feedback loop mechanism evaluates user satisfaction scores every 30 minutes; if the score falls below 80, parameter fine-tuning is triggered, and satisfaction data is analyzed to optimize configuration effectiveness. All these steps form a closed loop through data-driven and algorithmic support, ensuring system adaptability and a high degree of matching with user task contexts.

[0039] This embodiment addresses the inefficiency caused by the variability of task contexts, delayed behavioral responses, and mismatched interaction paths in office scenarios, proposing a comprehensive solution. By collecting user behavior data and extracting dynamic contextual change features, combined with real-time signal strength and historical behavior weights, the variability of task contexts is accurately assessed, and contextual feature vectors are generated to predict user intent, thus forming a description of the user's task intent. Furthermore, the system dynamically generates and adjusts instruction sequences, optimizes interface element configuration schemes, and integrates user habits and preferences with interaction weights to ensure interface layout adapts to the task context. Simultaneously, the intent description and configuration parameters are updated in real-time for behavioral deviations, ultimately achieving intelligent adaptation of the interaction path. Through the closed-loop mechanism of dynamic adjustment of contextual feature vectors and intent prediction, the personalization and real-time performance of interaction modes are significantly improved, comprehensively optimizing user experience and work efficiency in office scenarios, maintaining a high level of operational efficiency and comfort in ever-changing task contexts.

[0040] In one embodiment, please refer to Figure 2 and 3 After step 500, the following steps are also included: Step 600: Monitor the dynamic interaction path signal in real time; Step 700: If a behavioral deviation is identified in the dynamic interaction path signal, then the context feature vector is corrected according to the behavioral deviation. Step 800: Update the user task intent description based on the corrected context feature vector, and update the interface optimization adjustment parameters based on the updated user task intent description.

[0041] In this step, the system uses signal analysis to process dynamic interaction path signals, obtain path signal features, and determine the differences between these features and preset interaction mode configuration data to identify behavioral deviations. If the system determines that no behavioral deviation is detected, no interface element adjustments are made; if a behavioral deviation is detected, the system corrects the context feature vector based on the deviation data. The system uses the adjusted context feature vector to further update the user task intent description, generating a corresponding adjustment instruction sequence to obtain a preliminary instruction set. For the preliminary instruction set, the system determines if the instruction sequence matches preset module rules, and then outputs an optimized element configuration scheme. Using the optimized element configuration scheme, interaction weights are calculated to obtain the priority ranking of each interface element. Based on the priority ranking and the weight calculation results, the final adjustment parameters are determined, forming a complete parameter dataset. The complete parameter dataset is then applied to the interface optimization process to obtain the final interface presentation scheme. In addition, the system also generates new interaction mode configuration data based on the updated user task intent description to complete the adaptive update of the dynamic interaction mode configuration.

[0042] Specifically, prior to this step, the system initializes a configuration model that includes dimensions such as response latency threshold, operation sequence compliance, and UI element attention by parsing historical interaction logs and application preset rules. For example, the normal range of response latency is set to 100 milliseconds to 500 milliseconds, and the operation sequence is modeled using a hidden Markov model, whose state transition probability matrix is ​​obtained based on training from the past 10,000 successful sessions.

[0043] The system monitors dynamic interaction path signals in real time, collecting user click streams, cursor movement trajectories, and API call sequences 10 times per second. It uses a dynamic time warping algorithm to calculate the similarity between the current operation path and a standard path template. When behavioral deviations are detected, such as a calculated similarity below a preset threshold of 0.7, or a single-step response delay exceeding 800 milliseconds three consecutive times, an adjustment mechanism is triggered. Adjusting the contextual feature vector involves multi-dimensional quantification of the real-time context. The system immediately extracts features from the session data, such as the current page dwell time, the number of error pop-ups, and the types of the last five operations. It analyzes the root causes of deviations using a pre-trained random forest classifier (with an F1 score of 0.92 on the training set) and fine-tunes the 128-dimensional feature vector representing the user's current context accordingly, for example, increasing the "confusion index" feature value from 0.3 to 0.8. Finally, based on the updated contextual feature vector, the system uses an attention-based sequence-to-sequence model to regenerate or revise the user's task intent description. This model takes the adjusted feature vector and the current dialogue history as input and outputs a structured intent description. For example, the initial "query information" intent is updated to "After multiple unsuccessful searches, the user may be trying to find specific entries about the financial statements for the second quarter of 2023 through advanced filtering functions," thereby providing more accurate guidance for downstream services.

[0044] Furthermore, as another specific embodiment, during the interface optimization and adjustment process, the system first analyzes the updated user task intent description and automatically extracts user preference data. For example, the user's expected interface response speed is 0.5 seconds, and their preferred color contrast is 1.2. Natural language processing algorithms are used to semantically parse the intent text, generating a keyword weight matrix. "Fast response" has a weight of 0.8, and "visual comfort" has a weight of 0.6. Combined with historical user behavior data, the user task priority is calculated to be 0.75, forming an initial adjustment instruction sequence. For example, instruction 1 is "shorten loading time to 0.4 seconds," and instruction 2 is "adjust main color contrast to 1.3." Subsequently, based on the instruction sequence, the system calls a preset interface element adaptation algorithm, compares the loading time optimization target with the current server performance parameters, and determines that the actual adjustable time is 0.45 seconds. This is then adjusted using the color adjustment formula: , The optimized contrast ratio was calculated to be 1.43, and an adjusted element configuration scheme was generated, such as reducing button response time to 0.45 seconds and increasing background color contrast to 1.43. Next, the element interaction weights were integrated. Based on user click frequency data, the system calculated the interaction weight of the core button to be 0.9 and that of secondary buttons to be 0.5, using a weighted average algorithm. , The overall weight was determined to be 0.78. Based on the interface element configuration scheme, the priority of core buttons was increased by 20%, and the response time of secondary buttons was allowed to be delayed by 0.1 seconds. Finally, based on the above data, the system used a multi-objective optimization algorithm to balance response speed and visual effects, calculating the final interface optimization adjustment parameters. For example, the loading time was locked at 0.45 seconds, the contrast was fine-tuned to 1.4, and an additional 10% resource loading priority was allocated to core buttons to ensure a 15% improvement in user experience. The effectiveness of the parameter adjustments was verified through correlation analysis with business goals (such as a 5% increase in user retention rate).

[0045] This embodiment monitors dynamic interaction path signals in real time and responds promptly when behavioral deviations are identified. Based on the aforementioned dynamic interaction mode configuration, it adaptively adjusts interface optimization parameters, thereby forming a complete optimization logic chain and improving the personalization and real-time performance of the interaction mode.

[0046] In one embodiment, please refer to Figure 4 Step 300 includes: Step 301: Based on the historical data weights and user habit preferences, perform comparative analysis on the context feature vector to generate predicted intent data; Step 302: Extract intent label priorities based on the predicted intent data, and determine the label combination most relevant to the context feature vector according to the intent label priorities; Step 303: Perform semantic association analysis by combining the tag combination with the context feature vector to obtain the user task intent description.

[0047] In this step, the system first obtains basic information about user behavior patterns by comparing contextual feature vectors with historical data weights, thus determining preliminary behavior matching results. Then, based on these preliminary matching results, in-depth analysis is performed in conjunction with user habits and preferences. If the predicted intent deviation exceeds a preset threshold, the weight distribution of the feature vector comparison is readjusted to obtain more accurate predicted intent data. For the obtained predicted intent data, relevant patterns are extracted from historical behavior records. If the matching degree between the extracted patterns and the current contextual feature vector is lower than a preset threshold, the specific direction of the deviation is determined.

[0048] After identifying the specific direction of the deviation source, the system sorts the tags based on their priority and uses a pre-established tag semantic association library for matching to determine the tag combination most relevant to the user's task intent. The system then analyzes the semantic association of the tags with the feature vector of the current context to obtain a preliminary description of the user's task intent. Based on this preliminary description, the system integrates the weights of historical data and the influence of user habits and preferences to generate the final description of the user's task intent, thus completing the closed-loop processing of the entire prediction process.

[0049] Specifically, in the process of predicting user task intent, the system first quantifies the user's current environmental data, such as time, location, and device type, by constructing a context feature vector. For example, time is divided into hourly segments in a 24-hour format, location is accurate to 0.01 degrees using latitude and longitude coordinates, and device type is encoded as a value from 1 to 5, generating a vector containing 10 dimensions of features, such as [3, 45.23, 126.67, 2, ...]. The system then uses the Euclidean distance algorithm to calculate the similarity between the current context vector and the historical context vector. Assuming there are 1000 records in the historical data, the system takes the top 10 records with the highest similarity, with an average similarity of 0.85. If the similarity is higher than the preset threshold of 0.8, the system proceeds to the next step of analysis.

[0050] Next, a weighted average algorithm is used to compare the weights of historical data and user habits and preferences. The frequency of historical behavior (e.g., a certain behavior occurred 50 times in the past 30 days) is assigned a weight of 0.6, and user preferences (e.g., a user's preference for a certain type of task rating of 4.5 / 5) is assigned a weight of 0.4. The comprehensive score is calculated. If the score is 0.82, which is lower than the preset deviation standard of 0.9, the prediction intention is considered to be consistent with the historical behavior.

[0051] Furthermore, if the deviation meets the standard, the priority and semantic association of the intent tags are extracted. By constructing a tag priority matrix, for example, the priorities of tags A, B, and C are 0.5, 0.3, and 0.2, respectively, and combined with semantic association analysis, the semantic distance between tags is calculated using a word vector model. For example, the distance between tags A and B is 0.1, indicating a high degree of relevance. Finally, the tag A with the highest priority is determined as the main intent.

[0052] Finally, based on the above results, a user task intent description is generated. Combining the semantic content and contextual features of tag A, the output description is as follows: "At the current time 3 o'clock, at location 45.23, 126.67, the user may need to perform a task related to tag A." This description is then stored in the database for subsequent business calls, such as input data for a task recommendation system.

[0053] This embodiment compares and analyzes the context feature vector by weighting historical data and user habits and preferences to obtain predicted intent data. Then, it combines the predicted intent data with the context feature vector to perform intent tag priority ranking and tag semantic association analysis to obtain the user task intent description, thereby completing the closed-loop processing of the entire prediction process and forming a complete logical chain from context analysis to intent determination, ensuring the automation and accuracy of the technology implementation.

[0054] In one embodiment, please refer to Figure 4 and 5 Step 400 includes, Step 401: Map the user task intent description to generate the adjustment instruction sequence according to the intent tag priority; Step 402: Determine the instruction execution timing using an instruction sequence arrangement method, and obtain the adjustment instruction granularity based on the instruction execution timing; Step 403: According to the instruction execution timing and adjustment instruction granularity, the adjustment instruction sequence is processed into layers to obtain a layered instruction set; Step 404: Combine the preset interface element configuration rules with the instruction set to generate the initial adjustment scheme for the interface elements.

[0055] In this embodiment, the system obtains core requirement information by parsing the user's task intent description, determines the classification basis of intent tags, and then judges the importance of each tag based on the classification basis of intent tags and the priority order. If the association information of a certain tag exceeds a preset threshold, it is marked as high priority, and the priority ranking result is obtained.

[0056] Based on the priority ranking results, corresponding adjustment instructions are generated, and the execution order of the instructions is determined using an instruction sequence arrangement. Based on the instruction execution sequence, specific parameters for the adjustment granularity are obtained, and the instructions are processed in layers to obtain a layered instruction set. The system uses the layered instruction set, combined with the attribute information of the interface elements, and employs pre-established mapping rules to determine the matching degree between each element and the instruction. If the matching degree reaches a preset standard, it is categorized as a configuration object, determining the list of elements to be configured. Using the list of elements to be configured, combined with the logical rules of the module configuration, a preliminary interface layout scheme is generated, resulting in the final configuration scheme, i.e., the initial adjustment scheme for the aforementioned interface elements.

[0057] Specifically, when processing user task intent descriptions, the system first performs semantic analysis on the user-input text using natural language processing technology to extract core intent tags. For example, if the user inputs the text "quickly find the nearest restaurant," the system identifies "find" and "restaurant" as core intents through word segmentation and semantic vector matching, calculates the intent weight as 0.85, and sets the priority to 1. Next, the system generates an adjustment instruction sequence based on the intent tag priority. According to pre-established mapping rules, the system maps the priority 1 intent to the "search module," generating an instruction sequence such as "activate search interface - set geographic location parameters - limit search radius to 5 kilometers." The system uses timestamp sorting to ensure the execution sequence of instructions and sets the execution interval to 0.2 seconds to avoid system overload.

[0058] Subsequently, by combining the instruction execution timing and adjusting the instruction granularity, the system refines the instructions into subtasks. For example, "setting geographic location parameters" is broken down into "obtaining user latitude and longitude" and "writing search conditions," with each subtask allocated 0.1 seconds of execution time to ensure granularity is controlled at the millisecond level to improve response speed. Next, the system dynamically adjusts the instruction execution order based on the current network latency (assumed to be 50ms) and device performance (CPU utilization of 30%), prioritizing the execution of "obtaining user latitude and longitude" and generating a preliminary configuration scheme for interface elements. For example, the search box is placed at the top of the interface, the default value of the search radius option is set to 5 kilometers, and the button color is adjusted to RGB(255, 128, 0) for highlighting. Through these steps, a complete logical chain from intent recognition to interface configuration is formed. If the system detects a network latency exceeding 100ms, it automatically switches to a backup interface to ensure a response time of no more than 1 second. Simultaneously, it correlates business data such as user historical search records (assumed to be "Chinese food") and adjusts the default search result sort to "Chinese food priority," increasing its weight to 0.9.

[0059] This embodiment forms a complete logical chain from intent prediction to interface configuration through the above steps, thereby ensuring the accuracy of personalized configuration and the continuity of user experience.

[0060] In one embodiment, please refer to Figure 6 Step 200 includes: Step 201: Combine the historical data weights and signal strength to perform variability classification and generate variability assessment results; In this embodiment, the system acquires historical behavior data, determines whether data weight adjustment is needed, calculates the data weight ratio based on the historical behavior data and the current signal strength, completes behavior fusion processing, and obtains a comprehensive behavior feature value. Based on the comprehensive behavior feature value and the specific background of the task context, the system uses classification algorithms such as support vector machines to perform variability classification and determine the variability assessment level of the task context. Through the variability assessment level, a final assessment result of the dynamic changes in the context is generated, providing a basis for judging the adaptability of task execution.

[0061] Step 202: Based on the variability assessment results, analyze the heterogeneous characteristics and behavioral response attributes corresponding to the signal strength, and then generate the context feature vector through vector mapping dimension transformation.

[0062] In this embodiment, the system obtains raw records of task scenarios from multiple sources, including heterogeneous characteristics and timestamp information of behavioral responses, to obtain a preliminary scenario dataset. Based on this preliminary scenario dataset, the system standardizes the heterogeneous characteristics and integrates data from different sources using a unified format to determine the processed structured dataset. If the behavioral response timestamps in the structured dataset exceed a preset threshold, delay issues are marked, and relevant records are extracted to form a delay feature subset. Using this delay feature subset, a vector mapping method is applied to transform the high-dimensional data, constructing a low-dimensional scenario feature vector to obtain the transformed feature representation. For the transformed feature representation, the correlation between the task scenario and intent prediction is analyzed. If some dimension values ​​of the feature vector deviate from a preset range, their weights are adjusted to obtain an optimized feature vector. Based on the optimized feature vector and the contextual information of the variability evaluation results, an initial basis for intent prediction is constructed to determine potential behavioral trends. After obtaining the initial basis for intent prediction, input data for a scenario-adaptive prediction model is generated for the potential behavioral trends, and this is determined as the final scenario feature vector.

[0063] Specifically, in the task context variability assessment phase, uncertainty can be quantified by calculating the entropy value of the context data sequence. For example, by collecting the environmental noise level (unit: decibel) sequence of 10 consecutive user operations [45.2, 67.8, 50.1, 70.5, 46.0, 68.9, 49.5, 72.1, 47.3, 69.4], the Shannon entropy formula can be applied: , The calculations first discretize the data into three intervals (40-55, 55-70, and above 70) and count the frequencies, resulting in a probability distribution of approximately [0.4, 0.4, 0.2]. The calculated entropy value is approximately 1.52, indicating a moderate degree of variability in the situation. To address data heterogeneity, such as the simultaneous existence of numerical light intensity (unit: lux, sequence [320, 580, 210]) and categorical device types (coded as [1, 3, 2], representing mobile phone, tablet, and computer respectively), multi-dimensional processing is employed. Specifically, numerical data is standardized; assuming the mean of the light intensity sequence is 370 and the standard deviation is approximately 185.2, the first standardized value is approximately -0.27. Categorical data is one-hot encoded, with the device type "mobile phone" (coded 1) converted into a vector [1, 0, 0]. To address the behavior response delay, a time decay factor is introduced. For example, if the time difference sequence between the recorded user click event and the context acquisition is [0.3, 1.2, 0.8] seconds, and the decay coefficient λ = 0.5 is set, an exponential decay function is applied: , The weights are calculated to obtain a weight vector [0.86, 0.55, 0.67]. This weight is then weighted and fused with the corresponding standardized contextual data (such as standardized light intensity). Subsequently, through vector mapping dimensionality transformation, the processed heterogeneous features (such as weighted light values, one-hot encoded vectors, etc.) are concatenated into a unified high-dimensional vector, and dimensionality reduction is performed using Principal Component Analysis (PCA). Assuming the original concatenated vector has a dimension of 15, PCA retains 95% of the variance, reducing the dimension to 8, generating the final contextual feature vector. For example, a sample vector after dimensionality reduction might be [0.12, -0.45, 0.78, 0.02, -0.33, 0.19, 0.61, -0.05]. This contextual feature vector serves as the basic input data for the intent prediction model, for example, inputting it into a support vector machine or neural network model for subsequent intent classification prediction.

[0064] This embodiment forms a complete logical chain by monitoring signal strength, weighting and fusion, and then evaluating variability. The variability evaluation results are then mapped and converted into the required context feature vectors, ensuring the logical consistency of the entire process and improving the accuracy of the results.

[0065] In one embodiment, please refer to Figure 7 Before step 500, the following are also included: Step 010: Determine whether the initial adjustment scheme of the interface elements conforms to the preset basic layout specifications; Step 020: If yes, determine whether the initial adjustment scheme of the interface elements has passed the configuration compatibility test; Step 030: If not, then revise the initial adjustment scheme of the interface elements until the revised initial adjustment scheme of the interface elements passes the test, so as to perform the step of optimizing the initial adjustment scheme of the interface elements based on the user's habit preferences according to the revised initial adjustment scheme of the interface elements.

[0066] This implementation primarily focuses on the initial analysis of element distribution and layout constraints. A comparison with a pre-defined rule base is used to determine whether the layout conforms to basic layout specifications. These specifications are based on the general layout of the system interface and can specifically include: checking whether the spacing between different interface elements is within a preset range, and whether the arrangement of different interface elements is aligned, etc. This yields a preliminary layout scheme determination result. If the determination result indicates that the initial adjustment scheme of the interface elements does not conform to the basic layout specifications, the initial adjustment scheme is adjusted according to these specifications until the adjusted scheme conforms to the specifications, allowing for further steps.

[0067] If the judgment result shows that the initial adjustment scheme of this interface element conforms to the basic layout specifications, the system will continue to perform compatibility testing on it, and use device environment simulation tools to test the performance of the layout in different environments.

[0068] If the test results show incompatibility, the element distribution parameters in the initial adjustment scheme of the interface elements are corrected until the corrected initial adjustment scheme of the interface elements passes the above test; if the test results show no incompatibility, the subsequent steps can be executed directly.

[0069] Specifically, in the initial stage of obtaining the interface element configuration scheme, the current interface is first detected using an image recognition algorithm to identify components such as buttons and text boxes and their coordinates. For example, the coordinates of the login button are identified as (120, 300, 200, 50). Next, for element layout constraint verification, a rule-based detection algorithm is used to calculate the relative distance and alignment between elements. Constraint rules are set, such as the button spacing must be greater than 10 pixels. If two buttons are detected with a spacing of 8 pixels, it is determined to be a constraint violation. For configuration compatibility detection, a hash algorithm is used to compare the configuration fingerprint of the current layout with the standard fingerprint in the scheme library. If the similarity exceeds a preset threshold, such as 90%, it is determined to be a match.

[0070] Then, after the layout is matched, the system incorporates user habits and preferences. For example, by analyzing historical interaction logs, it uses a collaborative filtering algorithm to calculate the click weight of specific elements. For instance, the interaction weight of the "search box" element is 0.85, while the interaction weight of the "help link" element is 0.2. Simultaneously, combining the element interaction weights (obtained through test data statistics, such as the "submit button" having a weight coefficient of 1.5), a weighted fusion algorithm is employed. , Finally, the interface optimization adjustment parameters were determined. For example, the size of the login button was adjusted from 200×50 to 220×55, and the color saturation was increased by 15% to ensure that the adjustment not only conforms to the technical solution but also adapts to user behavior patterns.

[0071] This embodiment further modifies the initial adjustment scheme of interface elements based on the basic layout specifications and configuration compatibility of the interface layout, so as to ensure that the modified initial adjustment scheme of interface elements not only conforms to the technical solution but also adapts to user behavior patterns, thereby improving the rationality of the adjustment scheme.

[0072] The following describes an intelligent office digital interaction system based on adaptive decision-making provided by the present invention. The intelligent office digital interaction system based on adaptive decision-making described below and the intelligent office digital interaction method based on adaptive decision-making described above can be referred to and correspond to each other.

[0073] This invention provides an intelligent office digital interaction system based on adaptive decision-making, comprising: A signal strength monitoring module is used to monitor the signal strength in real time under the current dynamic situation, wherein the signal strength is represented by at least external environmental signals and / or user operation signals; The signal multidimensional quantization module is used to perform multidimensional quantization based on the signal strength and generate a context feature vector if the signal strength meets the preset context change conditions. The task intent determination module is used to determine the user task intent description corresponding to the signal strength based on the context feature vector, the weight of pre-collected historical data, and user habit preferences. The adjustment scheme conversion module is used to generate an adjustment instruction sequence based on the user task intent description, and to convert the adjustment instruction sequence into an initial adjustment scheme for interface elements by combining the instruction execution sequence and the granularity of the adjustment instruction. The optimization parameter generation module is used to optimize the initial adjustment scheme of the interface elements based on the user's habits and preferences, generate interface optimization adjustment parameters, and adaptively adjust the current interface based on the interface optimization adjustment parameters.

[0074] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A smart office digital interaction method based on adaptive decision-making, characterized in that, Executed by a computer, including: Real-time monitoring of signal strength in the current dynamic context, wherein the signal strength is represented by at least external environmental signals and / or user operation signals; If the signal strength meets the preset situation change conditions, then multi-dimensional quantization is performed based on the signal strength to generate a situation feature vector; Based on the context feature vector, the weights of pre-collected historical data, and user habit preferences, a user task intent description corresponding to the signal strength is determined. Based on the user task intent description, an adjustment instruction sequence is generated, and the adjustment instruction sequence is transformed into an initial adjustment scheme for interface elements by combining the instruction execution sequence and the granularity of the adjustment instructions. Based on the user's habits and preferences, the initial adjustment scheme of the interface elements is optimized to generate interface optimization adjustment parameters, so as to adaptively adjust the current interface based on the interface optimization adjustment parameters.

2. The intelligent office digital interaction method based on adaptive decision-making according to claim 1, characterized in that, Before optimizing the initial adjustment scheme of the interface elements based on the user's habits and preferences and generating interface optimization adjustment parameters, the method further includes: Collect continuous user operation data, perform feature analysis on the continuous user operation data, and obtain task switching density and context fluctuation frequency; Based on the task switching density and context fluctuation frequency, a context dynamic change feature reflecting the continuity of user's historical behavior is constructed; The step of optimizing the initial adjustment scheme of the interface elements based on the user's habits and preferences, generating interface optimization adjustment parameters, and adaptively adjusting the current interface based on the interface optimization adjustment parameters includes: Based on the user's habit preferences, determine the interface adjustment preference weights, and combine the preset interaction weights with the interface adjustment preference weights to perform weight analysis and obtain the weight analysis results; Based on the weight analysis results, the initial adjustment scheme of the interface elements is optimized to obtain the interface optimization adjustment parameters; The interface optimization parameters are adjusted according to the dynamic changes in the scenario to obtain the target interface layout scheme; The dynamic interaction path signal is determined based on the target interface layout scheme, so as to adaptively adjust the current interface through the dynamic interaction path signal.

3. The intelligent office digital interaction method based on adaptive decision-making according to claim 2, characterized in that, After optimizing the initial adjustment scheme of the interface elements based on the user's habits and preferences, generating interface optimization adjustment parameters, and adaptively adjusting the current interface based on the interface optimization adjustment parameters, the method further includes: Real-time monitoring of the dynamic interaction path signals; If a behavioral deviation is identified in the dynamic interaction path signal, the context feature vector is corrected based on the behavioral deviation. The user task intent description is updated based on the corrected context feature vector, and the interface optimization adjustment parameters are updated based on the updated user task intent description.

4. The intelligent office digital interaction method based on adaptive decision-making according to claim 1, characterized in that, The step of determining the user task intent description corresponding to the signal strength based on the context feature vector, pre-collected historical data weights, and user habit preferences includes: Based on the historical data weights and user habit preferences, the context feature vectors are compared and analyzed to generate predictive intent data; Based on the predicted intent data, intent label priorities are extracted, and the label combination most relevant to the context feature vector is determined according to the intent label priorities. By combining the aforementioned tag combinations with contextual feature vectors to perform semantic association analysis, the user's task intent description is obtained.

5. The intelligent office digital interaction method based on adaptive decision-making according to claim 4, characterized in that, The step of generating an adjustment instruction sequence based on the user task intent description, and then converting the adjustment instruction sequence into an initial adjustment scheme for interface elements by combining the instruction execution timing and adjustment instruction granularity, includes: The user task intent description is mapped to generate the adjustment instruction sequence according to the intent tag priority; The instruction execution timing is determined by an instruction sequence arrangement method, and the adjustment instruction granularity is obtained based on the instruction execution timing. According to the instruction execution timing and adjustment instruction granularity, the adjustment instruction sequence is layered to obtain a layered instruction set; By combining the preset interface element configuration rules with the instruction set, an initial adjustment scheme for the interface elements is generated.

6. The intelligent office digital interaction method based on adaptive decision-making according to claim 1, characterized in that, The step of performing multi-dimensional quantization based on the signal strength to generate a context feature vector includes: By combining the historical data weights and signal strength, variability classification is performed to generate variability assessment results; Based on the variability assessment results, the heterogeneous characteristics and behavioral response attributes corresponding to the signal strength are analyzed, and then the context feature vector is generated through vector mapping dimension transformation.

7. The intelligent office digital interaction method based on adaptive decision-making according to claim 1, characterized in that, Before optimizing the initial adjustment scheme of the interface elements based on the user's habits and preferences, the method further includes: Determine whether the initial adjustment scheme of the interface elements conforms to the preset basic layout specifications; If so, determine whether the initial adjustment scheme of the interface elements has passed the configuration compatibility test; If not, then the initial adjustment scheme of the interface elements is revised until the revised initial adjustment scheme of the interface elements passes the test, so as to perform the step of optimizing the initial adjustment scheme of the interface elements based on the user's habit preferences according to the revised initial adjustment scheme of the interface elements.

8. An intelligent office digital interaction system based on adaptive decision-making, characterized in that, include: A signal strength monitoring module is used to monitor the signal strength in real time under the current dynamic situation, wherein the signal strength is represented by at least external environmental signals and / or user operation signals; The signal multidimensional quantization module is used to perform multidimensional quantization based on the signal strength and generate a context feature vector if the signal strength meets the preset context change conditions. The task intent determination module is used to determine the user task intent description corresponding to the signal strength based on the context feature vector, the weight of pre-collected historical data, and user habit preferences. The adjustment scheme conversion module is used to generate an adjustment instruction sequence based on the user task intent description, and to convert the adjustment instruction sequence into an initial adjustment scheme for interface elements by combining the instruction execution timing and the adjustment instruction granularity. The optimization parameter generation module is used to optimize the initial adjustment scheme of the interface elements based on the user's habits and preferences, generate interface optimization adjustment parameters, and adaptively adjust the current interface based on the interface optimization adjustment parameters.

Images