Data display method and system for service suitable for aging, medium and program product

Abstract

The invention discloses a data display method and system for service suitable for aging, a medium and a program product, and relates to the field of electrical digital data processing.The method comprises the steps that historical interaction data of a target user is collected, and a user interaction mode is determined; dividing the user interaction instruction into an offline instruction set and an online instruction set based on the user interaction mode; voice input of a user is responded, and the voice input is converted into an interaction instruction based on a local language model; when the interaction instruction belongs to the offline instruction set, generating and displaying a local response result; when the interaction instruction belongs to the online instruction set, determining a network connection state; when the network connection state is normal connection, an interaction instruction is sent to the server, and an online response result returned by the server is received and displayed; and when the network connection state is connection abnormity, generating a temporary response result based on the historical interaction data and displaying the temporary response result. According to the invention, the problem of poor voice interaction result display effect under the condition of unstable network can be solved.

Description

Technical Field

[0001] The present application relates to the field of electronic digital data processing, and in particular to a data display method, system, medium and program product for aging-friendly services. Background Art

[0002] As the population ages, demand for intelligent, elderly-friendly services is growing. Seniors often interact with smart devices in a habitual and repetitive manner, relying more on simple voice interaction. Furthermore, they may occasionally encounter network instability and other issues when using smart devices, placing higher demands on service continuity and reliability.

[0003] In related technologies, aging-friendly service systems primarily utilize cloud computing architectures, uploading user voice commands to cloud servers for processing. After receiving the user's voice input, the system converts it into text using speech recognition technology. The cloud server then performs semantic understanding and task processing, ultimately returning the results to the user for feedback.

[0004] However, in actual applications, the cloud computing-based service model has high requirements for network quality. When the network fluctuates or is interrupted, the voice commands issued by users cannot be transmitted to the cloud for processing in time, which will cause service response delays and result in disordered display results. Especially in scenarios where the network is unstable and specified information cannot be loaded, elderly users may repeatedly query the same information. Each query requires re-network transmission and cloud processing, which will cause serious conflicts between previous and subsequent operations and affect the user experience. Summary of the Invention

[0005] The present application provides a data display method, system, medium and program product for aging-friendly services, which are used to optimize the problem of poor display effect of voice interaction results under unstable network conditions.

[0006] In the first aspect, the present application provides a data display method for aging-friendly services, which is applied to a data processing system. The method includes: collecting historical interaction data of target users, and determining the user interaction mode based on the interaction frequency, interaction time and interaction content in the historical interaction data; based on the user interaction mode, dividing the user interaction instructions into an offline instruction set and an online instruction set; the offline instruction set is multiple instructions for the user to query local data; in response to the user's voice input, the voice input is converted into an interaction instruction based on a local language model; when the interaction instruction belongs to the offline instruction set, a local response result is generated based on the corresponding local cache data, and the local response result is displayed; when the interaction instruction belongs to the online instruction set, the network connection status is determined; when the network connection status is normal, the interaction instruction is sent to the server, the online response result returned by the server is received, and the online response result is displayed; when the network connection status is abnormal, the historical response data of the corresponding interaction instruction is determined based on the historical interaction data, a temporary response result is generated based on the historical response data, and the temporary response result is displayed.

[0007] In the above embodiment, the data processing system divides instructions into offline and online instruction sets. When processing offline instructions or the network is normal, the response results can be directly displayed. When the network is abnormal, a temporary response can be generated based on historical data, thereby achieving service continuity; it not only ensures the usage habits of elderly users, but also optimizes the service interruption problem under unstable network conditions, thereby improving the reliability and user experience of aging-friendly services.

[0008] In combination with some embodiments of the first aspect, in some embodiments, the step of collecting historical interaction data of the target user and determining the user interaction pattern based on the interaction frequency, interaction time and interaction content in the historical interaction data specifically includes: collecting historical interaction data of the target user including active interaction data and passive interaction data; the active interaction data is the voice command initiated by the user, and the passive interaction data is the user's feedback operation on the system push information; based on the time sequence and operation type, constructing an association mapping between the active interaction data and the passive interaction data to generate a user interaction record; extracting the time association features, instruction association features and response association features in the user interaction record, constructing a user behavior link, and generating a user interaction pattern based on the user behavior link.

[0009] In the above embodiment, the data processing system forms an interaction pattern by analyzing the user's active interaction data and passive interaction data and constructing a user behavior link; based on the user's interaction characteristics and habits, it can more accurately understand and predict the user's interaction needs, thereby providing a service experience that is more in line with the usage habits of elderly users and lowering the usage threshold for elderly users.

[0010] In combination with some embodiments of the first aspect, in some embodiments, the step of dividing user interaction instructions into offline instruction sets and online instruction sets based on the user interaction mode specifically includes: determining the execution dependencies and data dependencies of the user interaction instructions based on the user interaction mode, and constructing an instruction association graph; determining independently executed instructions and timing-associated instructions based on the instruction association graph; the independently executed instructions are instructions that only depend on local data; the timing-associated instructions are instructions executed in a fixed order; merging the independently executed instructions and timing-associated instructions into an offline instruction set; determining the instructions in the user interaction instructions that do not belong to the offline instruction set, and generating an online instruction set.

[0011] In the above embodiment, the data processing system can identify independently executed instructions and timing-related instructions by analyzing the execution dependencies and data dependencies of instructions, and generate an offline instruction set; thereby ensuring the efficiency and accuracy of local services, reducing unnecessary network requests, and improving system response speed.

[0012] In combination with some embodiments of the first aspect, in some embodiments, after determining the instructions in the user interaction instructions that do not belong to the offline instruction set and generating the online instruction set, the method also includes: determining the execution frequency and time distribution of each instruction in the online instruction set, and generating an instruction execution feature record; the instruction execution feature record includes the instruction identifier, historical execution time and number of executions; according to the instruction execution feature record, calculating the time concentration and frequency ratio of each instruction in the online instruction set, and generating a high-frequency instruction list; based on the historical execution time of each high-frequency instruction in the high-frequency instruction list, determining the data prefetch time window and prefetch data range of the high-frequency instructions, and generating a prefetch strategy list; when detecting that the data prefetch time window has been entered, sending a data prefetch request to the server according to the corresponding prefetch strategy list, receiving the prefetched data returned by the server and storing it in the local cache.

[0013] In the above embodiment, the data processing system can pre-fetch data based on instruction execution feature records. By analyzing the time concentration and frequency ratio of high-frequency instructions, data can be pre-acquired in the appropriate time window, thereby reducing network request delays and improving system response speed and service quality.

[0014] In combination with some embodiments of the first aspect, in some embodiments, before the step of responding to the user's voice input and converting the voice input into interactive instructions based on the local language model, the method also includes: obtaining voice sample data of the target user in different emotional states; extracting the voiceprint features and voice expression features of the voice sample data to construct a user pronunciation feature library; training the voice recognition model based on the user pronunciation feature library to generate a voice recognition model; constructing a local language model based on the voice recognition model and preset semantic understanding constraints; the semantic understanding constraints include the voice acquisition waiting time.

[0015] In the above embodiment, the data processing system collects voice samples under different emotional states to build a personalized voice recognition model and local language model, thereby improving the accuracy of voice recognition, better adapting to the pronunciation characteristics and expression habits of elderly users, and making voice interaction more natural and smooth.

[0016] In combination with some embodiments of the first aspect, in some embodiments, when the interaction instruction belongs to an offline instruction set, the steps of generating a local response result based on the corresponding local cache data and displaying the local response result specifically include: constructing an index structure of local cache data; the index structure includes a data identifier, a timeliness mark and associated instructions; when the interaction instruction belongs to an offline instruction set, retrieving the cache record of the corresponding interaction instruction in the index structure; verifying the timeliness mark of the cache record to determine the valid status of the data; generating a local response result based on the valid status of the data and the associated instructions, and displaying the local response result.

[0017] In the above embodiment, the data processing system establishes an efficient local cache data index structure, manages data through timeliness tags and associated instructions, ensures the timeliness and accuracy of local responses, improves the quality of offline services, and reduces dependence on the network.

[0018] In combination with some embodiments of the first aspect, in some embodiments, the step of constructing an index structure for local cached data specifically includes: collecting query operation records of local cached data to generate a data access sequence; the data access sequence includes an access timestamp, a data item identifier, and an access source; based on the data access sequence, determining the query interval and query trigger method for each data item, and generating a data activity report; according to the average query interval in the data activity report, dividing the data items into high-frequency access data and low-frequency access data, and generating a hierarchical storage scheme; the hierarchical storage scheme includes multiple storage schemes with different data update frequencies; constructing an index structure based on the hierarchical storage scheme, storing the data items in the corresponding storage area, and recording the storage location, timeliness mark, and associated instructions of the data items in the index structure.

[0019] In the above embodiment, the data processing system establishes a hierarchical storage solution by analyzing the data access sequence, thereby realizing an efficient data storage and access mechanism, optimizing system resource utilization, improving data processing efficiency, and ensuring service stability and response speed.

[0020] In a second aspect, an embodiment of the present application provides a data processing system, which includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the data processing system to execute the method described in the first aspect and any possible implementation of the first aspect.

[0021] In a third aspect, an embodiment of the present application provides a computer program product comprising instructions, which, when the computer program product is run on a data processing system, enables the data processing system to execute the method described in the first aspect and any possible implementation of the first aspect.

[0022] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium comprising instructions, which, when executed on a data processing system, causes the data processing system to execute the method described in the first aspect and any possible implementation of the first aspect.

[0023] It is understandable that the data processing system provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the methods provided in the embodiments of the present application. Therefore, the beneficial effects that can be achieved can be referenced to the beneficial effects of the corresponding methods and will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: 1. By adopting a solution that analyzes user interaction patterns based on historical interaction data and intelligently divides instruction sets, as well as a response mechanism that combines local language models and network status adaptation, the system can accurately understand user habits and maintain service continuity under different network conditions. This effectively solves the problem of service interruption when the network is unstable due to complete reliance on cloud processing in related technologies, thereby achieving a more reliable aging-friendly service experience. By analyzing users' historical interaction data, an interaction model that conforms to the usage habits of elderly users is established, and then offline and online instruction sets are divided based on this model, so that the system can still maintain basic functions in the event of network anomalies, reducing dependence on the network and improving response speed.

[0025] 2. Due to the adoption of an instruction set partitioning scheme based on instruction dependency analysis and intelligent classification, combined with an identification mechanism for independently executed instructions and time-related instructions, the system can accurately distinguish between offline execution and online processing of services, effectively solving the problem in related technologies that cannot perform intelligent service scheduling based on instruction characteristics, thereby achieving more efficient local offline services; by analyzing the execution dependencies and data dependencies between instructions, a complete instruction association graph is constructed, thereby achieving more accurate instruction classification and improving the system's response efficiency and service quality.

[0026] 3. Due to the adoption of a local cache data management mechanism based on an index structure, combined with a verification scheme for timeliness marking and associated instructions, the system can efficiently manage and utilize local data, effectively solving the problems of inefficient local data management and inability to guarantee data timeliness in related technologies, thereby achieving more reliable local response services; by building an efficient index structure, the required data can be quickly located and accessed, while ensuring the validity of the data through timeliness marking, thereby improving the response speed and accuracy of local services. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 This is a flow chart of a data display method for an aging-friendly service in an embodiment of the present application; Figure 2 This is another flowchart of the data display method for the aging-friendly service in an embodiment of the present application; Figure 3 It is a schematic diagram of the structure of a physical device of the data processing system in an embodiment of the present application. DETAILED DESCRIPTION

[0028] The terms used in the following examples of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the specification of the present application, the singular expressions "a", "an", "above", "the", and "this" are intended to include plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in the present application refers to any or all possible combinations of one or more of the listed items.

[0029] In the following, the terms "first" and "second" are used for descriptive purposes only and should not be understood to imply or suggest relative importance or implicitly indicate the number of the technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of this application, unless otherwise specified, "plurality" means two or more.

[0030] For ease of understanding, the application scenarios of the embodiments of the present application are introduced below.

[0031] In a retirement community, elderly user A frequently uses a smart voice assistant to check weather, news, and family information. Due to his age, he struggles with mobile phone operation and relies primarily on voice interaction. Every morning at 6:00 AM, he inquires about the weather; after breakfast at 8:00 PM, he checks the news; and at 11:00 AM and 7:00 PM, he checks his family's location. However, the community's network signal is unstable and occasionally disconnects.

[0032] In related technologies, cloud computing-based voice interaction solutions can be used to meet the intelligent service needs of elderly users. All user voice commands are uploaded to a cloud server in real time. The server performs voice recognition, semantic understanding, and task processing, and then returns the results to the client for display. The following describes a scenario using the data display method for aging-friendly services in related technologies.

[0033] Currently, elderly user A uses a voice assistant system based on a traditional cloud computing architecture. When he issues a voice command to inquire about the weather, the system needs to upload the voice data to a cloud server for recognition and processing. One day, when elderly user A was asking about the weather as usual, the network experienced a network failure. Although the system received the voice message, it was unable to connect to the server for recognition and processing. The screen displayed a message stating "Network connection failed, please try again later." Unable to understand, the user continued to repeat the request. The system repeatedly attempted to connect to the server but was unsuccessful, rendering the query completely inoperable. In the event of a network failure, the system cannot provide any effective response, severely impacting the user experience.

[0034] The data display method for the aging-friendly service in the embodiment of this application is adopted. By establishing mechanisms such as user interaction pattern analysis, command classification processing, and local voice recognition, it achieves the ability to continue providing services in the event of network anomalies. This not only ensures the stability of basic functions, but also provides an interactive experience that conforms to the usage habits of the elderly. The system can intelligently identify the user's fixed query pattern, pre-cache commonly used data, and still provide timely responses when the network is interrupted. The following describes the scenario in which the data display method for the aging-friendly service in this application is used.

[0035] After adopting the solution of this application, the smart voice assistant of elderly user A was upgraded. Through long-term observation, the system discovered his fixed query habits, divided frequently used commands such as weather query and news browsing into offline command sets, and cached relevant data locally in advance. When elderly user A inquires about the weather at 6 o'clock in the morning, even if the network is unstable, the system can respond based on the locally cached weather data. For online commands such as family location information, the system will predict the upcoming query time window when the network is normal and obtain data in advance. Even in the event of a network failure, a temporary response can be generated based on the most recently cached location data, generating information such as "Your family was in a nearby supermarket 10 minutes ago", avoiding unresponsive conditions.

[0036] It can be seen that the data display method of the aging-friendly service in the embodiment of the present application can not only realize the intelligent voice interaction function, but also effectively solve the service interruption problem caused by network instability. Through localized processing and data pre-fetching mechanism, the continuity and reliability of the service are achieved, thereby significantly improving the user experience of elderly users.

[0037] For ease of understanding, the following describes the process of the method provided by this implementation in combination with the above scenario. Figure 1 , which is a flow chart of the data display method for aging-friendly services in an embodiment of the present application.

[0038] S101. Collect historical interaction data of a target user, and determine the user interaction pattern based on the interaction frequency, interaction time, and interaction content in the historical interaction data.

[0039] Among them, historical interaction data represents all interaction records generated by users during past usage, including proactive voice commands and feedback operations on system push information; interaction frequency refers to the number of times a specific interaction behavior occurs per unit time; interaction time represents the time point distribution characteristics of users' interaction operations; interaction content refers to the user's specific command content and operation type; user interaction pattern represents the set of users' habitual interaction behavior characteristics obtained through data analysis.

[0040] When launching age-friendly services, the data processing system first needs to establish a user interaction behavior model. Specifically, the data processing system first obtains the target user's historical interaction data from local storage and cloud servers, including voice command records and operation feedback data from the past 30 days. The data processing system then organizes this data in chronological order and calculates the frequency of occurrence, time distribution pattern, and content similarity of each type of interaction behavior. By performing a multi-dimensional analysis of these features, the data processing system constructs an interaction behavior model that reflects user habits for subsequent intelligent service adaptation.

[0041] In some embodiments, the construction of user interaction patterns can be achieved in a variety of ways: Optionally, the data processing system can adopt a method based on time series pattern mining, first converting historical interaction data into a time series, using a sliding time window to calculate local features, identifying frequently occurring interaction patterns through a sequence pattern mining algorithm, and finally generating a user behavior feature model in combination with time association rules; Optionally, the data processing system can also adopt a method based on clustering analysis, quantizing the interaction data in three dimensions of time, frequency, and content, clustering the feature vectors using an improved K-means algorithm, and identifying the user's main interaction pattern by analyzing the feature distribution of the cluster center. It is understandable that other machine learning methods can also be used to achieve the construction of user interaction patterns, which are not limited here. It should also be noted that in the feature extraction process, algorithms such as edit distance or word vector cosine similarity can be used to calculate the similarity of the interaction content.

[0042] In practical applications, user interaction data may be sparse, especially for newly registered users or those with low usage frequency. To address this, the data processing system can adopt a progressive model construction approach: first, a preset interaction pattern template is selected based on basic user information (such as age group and usage scenario), and then new interaction data is continuously collected during user usage. The data processing system sets a feature weight decay factor to give newly collected interaction data a higher weight while maintaining the reference value of historical data. When the amount of newly collected data reaches a threshold, the system recalculates user features and updates the interaction pattern.

[0043] S102: Divide the user interaction instructions into an offline instruction set and an online instruction set based on the user interaction mode.

[0044] Among them, user interaction instructions refer to all user operation commands that the system can receive and process; the offline instruction set refers to the instruction set that can be processed only by relying on local data and computing resources; the online instruction set refers to the instruction set that needs to obtain cloud service support through network connection.

[0045] After constructing the user interaction model, the data processing system needs to categorize and manage possible interaction instructions. Specifically, the data processing system first extracts the instruction sequence in the user interaction model and analyzes the execution characteristics of each instruction, including data source, computational complexity, and response timeliness requirements. Instructions that are highly repetitive and rely on relatively stable data are classified as offline instructions. Instructions that require real-time data updates or complex calculations are classified as online instructions. This categorization optimizes system resource utilization and improves service response speed.

[0046] In some embodiments, the partitioning of instruction sets can be achieved in a variety of ways: Optionally, the data processing system can adopt a partitioning method based on graph theory, first construct an instruction dependency graph, where nodes represent instructions and edges represent dependencies, and use a minimum cut algorithm to split the graph into two subgraphs, corresponding to offline and online instruction sets respectively; Optionally, the data processing system can also adopt a classification method based on a decision tree, using the various features of the instructions as decision attributes, selecting the optimal partitioning features through the information gain rate, and constructing a classification decision tree to complete the instruction set partitioning. It is understandable that other classification algorithms can also be used to achieve the partitioning of instruction sets, which is not limited here. It should also be noted that the system also needs to regularly evaluate the rationality of the partitioning results and make dynamic adjustments based on the actual execution effect.

[0047] In practice, instruction set partitioning can face challenges with complex dependency relationships and fuzzy boundaries. To address this, data processing systems can adopt a multi-level partitioning strategy: first, establish a hierarchical classification system for instructions, organizing functionally similar instructions into instruction families; then, analyze dependencies within and between instruction families to identify the smallest atomic instructions; and finally, perform offline and online partitioning based on the characteristics of atomic instructions. For complex dependency relationships, the system implements a transition buffer to maintain some instructions in both offline and online instruction sets, selecting the optimal processing method based on the actual execution environment.

[0048] S103: Respond to the user's voice input and convert the voice input into interaction instructions based on the local language model.

[0049] Among them, voice input refers to the natural language commands issued by the user through voice; the local language model refers to the voice recognition and semantic understanding model deployed on the terminal device.

[0050] After receiving the user's voice input, the data processing system needs to accurately understand the semantics and convert the instructions. Specifically, the data processing system first acquires the user's voice signal through the voice acquisition module and performs noise reduction and feature extraction. The system then uses a local language model to identify the voice features and obtain preliminary text content. Next, the system combines the user's voice recognition model and semantic understanding constraints to perform personalized semantic analysis of the text content and extract the user's true intent. Finally, the system maps the analysis results into standardized interaction instructions in preparation for subsequent processing.

[0051] In some embodiments, the conversion from speech to command can be achieved in a variety of ways: Optionally, the data processing system can adopt an end-to-end conversion method based on deep learning, use an improved Transformer model to directly map the speech feature sequence to the command sequence, and capture the key information in the speech through the attention mechanism to achieve efficient and accurate conversion; Optionally, the data processing system can also adopt a rule-based step-by-step conversion method, first using MFCC feature extraction and GMM-HMM model for speech recognition, and then performing structured parsing of the text through the semantic analysis tree, and finally generating standard instructions through template matching. It is understandable that other natural language processing methods can also be used to achieve voice command conversion, which is not limited here. It should be noted that the system needs to optimize the model based on the pronunciation characteristics and expression habits of elderly users, including appropriately extending the voice waiting time, adding a repeated confirmation mechanism, etc.

[0052] In practical applications, elderly users' voice input may suffer from issues such as non-standard pronunciation and incomplete expressions. To address this, the data processing system can employ a multimodal fusion strategy: first, a personalized pronunciation feature library is built, collecting voice samples from different emotions and environments. Then, a semantic completion model is built by combining contextual information and historical interaction records. When ambiguous or incomplete voice input is detected, the system automatically performs semantic completion and confirmation. For important commands, the system provides voice feedback to confirm the user's understanding.

[0053] S104: When the interactive instruction belongs to an offline instruction set, a local response result is generated based on the corresponding local cache data, and the local response result is displayed.

[0054] Among them, local cached data refers to various data resources stored on the terminal device; local response results refer to service outputs obtained based on local calculations and data processing; data identifiers are used to uniquely identify cached data items; timeliness tags indicate the validity period and update time of the data; associated instructions refer to the set of interactive instructions related to the data item; and the data validity status indicates whether the cached data can be used to generate response results.

[0055] After confirming that the interaction instruction belongs to the offline instruction set, the data processing system immediately performs local response processing. Specifically, the data processing system first searches the index structure based on the interaction instruction to locate the required cached data. The system then checks the cached data's expiration status to determine whether the data is still valid. For valid data, the system directly generates a response based on the pre-set processing logic. For invalid data, the system attempts to use backup data or a downgraded service solution. Finally, the system displays the generated response to the user through the user interface and records relevant information about the interaction process.

[0056] In some embodiments, the generation of local response results can be achieved in a variety of ways: Optionally, the data processing system can adopt a response method based on pre-compiled templates, pre-compile the commonly used response logic into local executable code, and quickly generate response results through parameter filling and conditional judgment to achieve efficient local processing; Optionally, the data processing system can also adopt a response method based on a lightweight reasoning engine, locally deploy a simplified version of the business rule engine, and generate personalized response results through rule matching and reasoning calculations. It is understandable that other local computing methods can also be used to achieve response result generation, which is not limited here. It should be noted that the system needs to design corresponding degradation service solutions for different types of instructions to ensure that basic services can still be provided when data fails.

[0057] In practice, managing local cached data can face challenges such as limited storage space and data consistency. To address this, data processing systems can adopt a hierarchical caching strategy: first, cached data is classified based on access frequency and importance, using memory caching for frequently accessed data and file storage for less frequently accessed data. The cache space is then managed using the LRU-K algorithm, promptly clearing expired and low-value data. Furthermore, the system maintains data version numbers and update timestamps, automatically synchronizing with the server upon network recovery to ensure data consistency.

[0058] S105: When the interactive instruction belongs to an online instruction set, determine the network connection status.

[0059] Among them, the network connection status indicates the communication quality between the current device and the server; normal connection means that the network can stably transmit data; abnormal connection means that the network is interrupted or the transmission quality does not meet the service requirements; network quality parameters include indicators such as latency, packet loss rate, and bandwidth; connection threshold represents the key indicator value for determining the network status.

[0060] After the data processing system identifies an interactive instruction as belonging to the online instruction set, it needs to evaluate the network status. Specifically, the data processing system first obtains the current network connection parameters, including signal strength, latency, and data transmission rate, through the network interface. The system then compares these parameters with the preset quality of service requirements to comprehensively assess the network status. The system uses different evaluation criteria for different types of online instructions to ensure that the network status meets the minimum requirements for instruction execution. Furthermore, the system continuously monitors changes in network status to provide a basis for subsequent service scheduling.

[0061] In some embodiments, the determination of the network status can be achieved in a variety of ways: Optionally, the data processing system can adopt an evaluation method based on multi-dimensional indicators, by establishing a network quality scoring model, calculating a comprehensive score based on weights for indicators such as delay, packet loss rate, and bandwidth, and determining the network status level based on the score; Optionally, the data processing system can also adopt a prediction method based on historical data, by collecting network status data over a period of time, using time series analysis to predict short-term network quality changes, and adjust service strategies in advance. It is understandable that other network quality assessment methods can also be used to determine the network status, which is not limited here. It should be noted that the system needs to dynamically adjust the criteria for determining the network status according to different service scenarios.

[0062] In real-world applications, network status can fluctuate frequently, leading to unstable service quality. To address this, data processing systems can employ adaptive network monitoring strategies: First, a multi-level network status assessment system is established, categorizing network status into multiple levels. A sliding window algorithm is then used to calculate the stability of network metrics, avoiding frequent state transitions caused by transient fluctuations. When critical network status is detected, the system increases the sampling frequency to track network changes in real time. Furthermore, the system automatically adjusts the weights and thresholds of various metrics based on historical experience, ensuring more accurate and reliable network status assessments.

[0063] S106: When the network connection status is normal, send an interactive instruction to the server, receive an online response result returned by the server, and display the online response result.

[0064] Among them, the server represents the cloud computing resources that provide online services; the online response result refers to the data and service content returned by the server after processing the interactive instructions; the data transmission protocol represents the communication specifications between the client and the server; the service quality protocol refers to the service standards that define service response time, retry strategy, etc.; the session state represents the current connection status between the client and the server.

[0065] After confirming a normal network connection, the data processing system begins executing the online service processing flow. Specifically, the data processing system first standardizes and encapsulates the interactive instructions, adding the necessary authentication and session information. The system then sends the instructions to the server via a secure channel and starts a response monitoring timer. While waiting for the server's response, the system maintains the session state and handles possible timeouts and retries. Upon receiving the server's response, the system verifies the data integrity and then converts the result into a display-friendly format, presenting it to the user through the user interface.

[0066] In practice, online services may face issues such as response delays and incomplete data. To address this, data processing systems can employ a multi-tiered service assurance strategy: first, establish a service quality monitoring system to record the performance metrics of each service call in real time; then, use a dynamic load balancing algorithm to select the optimal service node to reduce network transmission latency; for important service requests, the system uses a parallel request mechanism, sending requests to multiple service nodes simultaneously and using the first valid response returned. When a service anomaly is detected, the system automatically switches to a backup service node or downgrades the service.

[0067] S107: When the network connection status is abnormal, determine historical response data corresponding to the interaction instruction based on the historical interaction data, generate a temporary response result based on the historical response data, and display the temporary response result.

[0068] Historical interaction data refers to all operational records generated by users while using the system, including active and passive interaction data. Active interaction data refers to user-initiated actions such as voice commands and touch operations. Passive interaction data refers to user feedback on information pushed by the system. Interaction frequency indicates the number of times a specific operation occurs per unit time. Interaction time refers to the time and duration of a user's operation. Interaction content refers to the specific command type and parameter information of the user's operation. User interaction pattern refers to the set of habitual user behavior characteristics obtained through data analysis. For example, an elderly user's fixed habit of checking the weather at 8 a.m. and health data at 9 p.m. every day constitutes an interaction pattern.

[0069] When launching aging-friendly services, the data processing system first needs to establish a user interaction behavior model. Specifically, the data processing system obtains historical user interaction data through multiple data collection channels, including locally stored operation logs and cloud-synchronized behavior records. For each interaction record, the data processing system extracts key features such as operation timestamp, instruction type, and parameter information. Then, the data processing system uses a sliding time window method to calculate the distribution of interaction features at different time scales. For the interactive content, the data processing system uses a semantic similarity algorithm to perform cluster analysis to identify the types of operations commonly used by users. Finally, the data processing system uses a time series pattern mining algorithm to discover behavioral patterns with strong repetitiveness and high time correlation from the interaction sequence, and construct a user interaction pattern feature library.

[0070] In practical applications, insufficient user interaction data or low-quality data may be encountered. To address this, the data processing system can adopt a hierarchical and progressive modeling strategy: first, based on basic information such as the user's age group and usage scenario, the most suitable basic model is selected from a preset library of interaction pattern templates. The feature weight decay factor λ(t) = e^(-αt) is then set, where t is the data time interval and α is the decay coefficient. This gives newly collected interaction data a higher weight. Furthermore, the data processing system sets a minimum sample threshold N and a confidence threshold β. Only when the number of samples of a certain type of interaction behavior is greater than N and the pattern confidence exceeds β will it be included in the interaction pattern feature library.

[0071] In the above example, the system established a personalized service model by analyzing historical user interaction data. In practical applications, this intelligent prediction and localized processing mechanism based on user habits can effectively reduce network dependence and provide a more stable and reliable service experience. The following supplements the scenario of this example.

[0072] Over time, the system builds a more in-depth profile of elderly user A's usage habits. Discovering that he checks the weather more frequently on rainy days than on sunny ones, the system automatically adjusts its weather data update strategy, increasing the frequency of data pre-fetching when rainy weather is predicted. At the same time, the system notices that users speak more slowly and their voices tremble when they are unwell, automatically extending the voice collection time and adjusting the recognition parameters. Before important holidays, the system predicts based on historical data that users may increase their frequency of inquiries about family members, expanding the time window and range for data pre-fetching in advance. These optimizations not only improve the accuracy of system responses but also demonstrate meticulous care for elderly users, making intelligent services more user-friendly.

[0073] After combining the above scenarios, the following is a more detailed description of the process of the method provided by this implementation. Figure 2 , which is another flow chart of the data display method for aging-friendly services in an embodiment of the present application.

[0074] S201: Collect historical interaction data of target users, and determine the user interaction pattern based on the interaction frequency, interaction time, and interaction content in the historical interaction data.

[0075] Referring to step S101 , the data processing system first determines the user interaction mode.

[0076] In some embodiments, the data processing system will parse the user's historical interaction behavior and determine the user interaction pattern, that is, the data processing system will collect the target user's historical interaction data, including active interaction data and passive interaction data; the active interaction data is the voice command initiated by the user, and the passive interaction data is the user's feedback operation on the system push information; based on the time sequence and operation type, a correlation mapping of the active interaction data and the passive interaction data is constructed to generate a user interaction record; the time correlation features, instruction correlation features and response correlation features in the user interaction record are extracted to construct a user behavior link, and a user interaction pattern is generated based on the user behavior link.

[0077] Among them, the target user refers to the specific elderly user individual who uses the system; the active interaction data refers to the interactive behavior data initiated by the user, including voice commands, touch operations, etc.; the passive interaction data refers to the user's feedback behavior data after the system pushes information, including operations such as viewing, clicking, and closing; the association mapping refers to the correspondence established between user interaction behaviors according to temporal relationships and operation types; the user behavior link refers to the temporal association path of user interaction behaviors.

[0078] When launching aging-friendly services, the data processing system must first establish a user interaction behavior model. Specifically, the data processing system continuously collects and records all user interaction behavior data, including voice commands initiated by users and feedback operations on system push information. The system sorts this interaction data in chronological order and establishes association mapping relationships based on different operation types (such as query, confirmation, cancellation, etc.) to form a complete user interaction record. The system then analyzes these interaction records, extracting temporal regularity features (such as operations within a fixed time period each day), command association features (such as certain commands are frequently used continuously), and response association features (such as fixed feedback methods for certain types of prompts). Based on this, it constructs a user behavior chain model, ultimately generating an interaction pattern that reflects user habits.

[0079] In some embodiments, the construction of user interaction patterns can be achieved in a variety of ways: optionally, clustering the user interaction data in the time dimension through a clustering analysis method to identify interactive behaviors with significant time regularities, and constructing a time series pattern based on the time regularity characteristics; analyzing the association relationship of instruction sequences based on association rule mining algorithms to identify frequently occurring instruction combination patterns; using sequence pattern mining technology to discover typical sequence patterns in user interaction behaviors. Optionally, a probabilistic graph model is used to model user interaction behaviors, construct a behavior transition probability matrix, and predict the user's possible next action; combining the time decay factor to assign different weights to historical interaction data to highlight recent behavior characteristics; through multi-dimensional feature fusion, comprehensively considering factors such as time, scene, and purpose, a more accurate user interaction pattern is generated. It is understandable that other machine learning methods can also be used to achieve the construction of user interaction patterns, which are not limited here.

[0080] In practical applications, users may experience sudden changes in their interaction behavior, meaning that their actual actions deviate significantly from the established interaction pattern. To address this, the data processing system employs a dynamic adaptation strategy: First, the deviation between the current interaction behavior and the predicted pattern is calculated. When the deviation exceeds a preset threshold, a pattern adjustment mechanism is initiated. Model parameters are adjusted accordingly based on the type of deviation (e.g., time offset, sequence change, etc.). Meanwhile, the system continuously monitors the new interaction pattern and updates the user interaction pattern once stability is confirmed. Furthermore, the system records possible causes of pattern changes (e.g., seasonal changes, health conditions, etc.) for subsequent optimization of pattern predictions.

[0081] S202: Based on the user interaction pattern, determine the execution dependency and data dependency of the user interaction instructions, and construct an instruction association graph.

[0082] Execution dependencies refer to the order in which interactive instructions must be executed; data dependencies represent the read and write relationships between data during instruction execution; the instruction association graph is a directed graph structure that describes the dependencies between instructions; node weights represent the execution frequency and importance of instructions; edge weights represent the strength of dependencies; and the critical path is the instruction sequence that affects overall execution efficiency. For example, the command to view health reports must be executed after the command to collect data, forming an execution dependency.

[0083] After obtaining the user interaction pattern, the data processing system needs to conduct an in-depth analysis of the relationships between instructions. Specifically, the data processing system first traverses all instructions in the user interaction pattern, extracting the input data source and output data items for each instruction. Then, based on data flow analysis, the data processing system identifies the data transfer relationships between instructions and establishes a data dependency matrix. Regarding the execution order, the data processing system analyzes historical interaction sequences to statistically analyze the timing relationships and triggering conditions between instructions. Finally, the data processing system converts these dependencies into a weighted directed graph, where nodes represent instructions, edges represent dependencies, and weights reflect the strength of the dependencies.

[0084] In some embodiments, the construction of the instruction association graph can be achieved in a variety of ways: optionally, the data processing system first constructs an adjacency matrix between instructions, calculates the conditional probability between each pair of instructions as the dependency strength, and then uses the Floyd-Warshall algorithm to calculate the shortest path between instructions, and finally identifies the strongly connected components through the graph segmentation algorithm to determine the instruction block; optionally, the data processing system can also adopt a method based on program dependency analysis, first establish a control flow graph and a data flow graph, and then analyze the dependency relationship between instructions through program slicing technology, and finally merge the control dependency and data dependency to generate a comprehensive association graph. It is understandable that other graph theory or program analysis methods can also be used to achieve the construction of the instruction association graph, which is not limited here. It should be noted that the possibility of concurrent execution of instructions needs to be considered and concurrent dependency constraints need to be established.

[0085] In practice, instruction dependencies may contain circular or conflicting dependencies. To address this, data processing systems can employ a hierarchical dependency analysis strategy: First, circular dependencies are detected using a topological sorting algorithm. For detected loops, the coupling degree of the instructions within the loop is calculated. When the coupling degree exceeds a threshold, the instructions within the loop are merged into atomic instruction blocks. For conflicting dependencies, the system constructs a conflict matrix and uses a graph coloring algorithm to determine an instruction scheduling solution that minimizes conflicts.

[0086] S203: Determine independently executed instructions and timing-related instructions based on the instruction association graph.

[0087] Independently executed instructions are instructions that do not depend on the execution results of other instructions; sequentially dependent instructions are instructions that have an execution order dependency with other instructions; instruction in-degree represents the number of dependency edges pointing to the instruction; instruction out-degree represents the number of dependency edges pointing to other instructions; association strength refers to the closeness of the dependency relationship between instructions; and execution constraints represent the preconditions that must be met for instruction execution. For example, an alarm setting instruction is an independently executed instruction, while a health report generation instruction is a sequentially dependent instruction.

[0088] After constructing the instruction association graph, the data processing system needs to classify and analyze the instructions. Specifically, the data processing system first traverses all nodes in the instruction association graph and calculates the in-degree and out-degree of each node. For nodes with both an in-degree and an out-degree of 0, the data processing system marks them as independently executed instructions. Next, the data processing system analyzes nodes with non-zero in-degree or out-degree and calculates the strength of their association with adjacent nodes. For instruction groups with an association strength exceeding a threshold, the data processing system classifies them as sequentially associated instructions. Finally, the data processing system verifies the execution constraints of each instruction category to ensure the correctness of the classification results.

[0089] In some embodiments, instruction classification can be achieved in a variety of ways: optionally, the data processing system first decomposes the instruction association graph into strongly connected components, uses the Tarjan algorithm to identify strongly connected subgraphs in the graph, and then calculates the connectivity between the subgraphs, determines independent subgraphs and associated subgraphs based on a preset threshold, and finally classifies the instructions contained according to the subgraph attributes; optionally, the data processing system can also adopt a classification method based on community discovery, first using a modularity optimization algorithm to identify the community structure in the instruction association graph, and then analyzing the internal connection density and external connection sparsity of the community, and finally classifying the instructions based on the community characteristics. It is understandable that other graph theory analysis methods can also be used to achieve instruction classification, which is not limited here. It should be noted that the execution frequency and importance of the instructions need to be considered in the classification process, and key instructions need to be specially handled.

[0090] In practical applications, instruction classification can face issues with fuzzy boundaries and unstable classification. To address this, data processing systems can employ a fuzzy clustering analysis strategy: First, a similarity function S(i, j) between instructions is defined. Taking into account the execution dependency strength D(i, j) and the degree of data sharing C(i, j), the overall similarity S(i, j) = αD(i, j) + βC(i, j) is calculated, where α and β are weight coefficients. Then, an improved FCM (Fuzzy C-Means) algorithm is used to cluster instructions, assigning each instruction a membership value indicating its degree of independent execution or sequential correlation. When an instruction's classification membership approaches a threshold, the system further analyzes its execution context and dynamically adjusts the classification result. This strategy can address uncertainty in instruction classification and improve classification accuracy and stability.

[0091] S204: Merge the independent execution instructions and the timing-related instructions into an offline instruction set.

[0092] Among them, the offline instruction set refers to the set of instructions that can be executed in the local environment; the merging rules refer to the strategies and constraints for combining different types of instructions; the instruction priority represents the sorting weight of the instruction in the execution queue; the resource requirement is used to represent the computing and storage resources required for instruction execution; the local execution capability refers to the upper limit of the computing and storage resources that the terminal device can provide; the execution efficiency refers to the processing speed and response time of the instruction in the local environment.

[0093] After completing instruction classification, the data processing system needs to merge instruction sets. Specifically, the data processing system first evaluates the resource requirements for independently executed instructions and time-related instructions, including indicators such as computational complexity, memory usage, and storage space. The data processing system then matches these requirements with the local execution capabilities of the terminal device and selects instructions that can be stably executed in the local environment. For time-related instructions, the data processing system also needs to verify whether the data they depend on can be stored and updated locally. Finally, the data processing system organizes qualified instructions into offline instruction sets in a specific order based on instruction priority and execution efficiency.

[0094] In some embodiments, the instruction set merge can be achieved in a variety of ways: Optionally, the data processing system first establishes a multi-dimensional resource constraint model, sets indicators such as CPU usage, memory occupancy, storage space, etc. as constraints, and then uses the knapsack algorithm to solve the optimal instruction combination, and finally determines the instruction execution order through the scheduling algorithm to generate the resource-optimal offline instruction set; Optionally, the data processing system can also adopt a merging method based on heuristic rules, first defining a scoring function for instruction merging, considering factors such as resource efficiency, execution stability, and user experience, and then gradually selecting high-scoring instructions for merging through a greedy strategy, and finally performing conflict detection and optimization on the merged results. It is understandable that other combinatorial optimization methods can also be used to achieve instruction set merging, which is not limited here. It should be noted that it is necessary to consider device performance differences and formulate differentiated merging strategies for terminal devices with different configurations.

[0095] In practice, instruction set consolidation can lead to resource contention or unstable performance. To address this, data processing systems can employ adaptive resource scheduling strategies: First, establish an instruction resource requirement model R = {r_1, r_2, ..., r_n}, where r_i represents the resource requirements of different types. Then, define a device resource capacity vector C = {c_1, c_2, ..., c_n} and monitor resource usage U(t) in real time. When resource usage exceeds a warning threshold θ (i.e., if there exists a value i such that U(t)_i / c_i>θ), the system triggers a dynamic adjustment mechanism: temporarily lowering the execution priority of some non-critical instructions to free up resources for critical instructions. Simultaneously, the system records performance fluctuations and regularly updates instruction resource requirement estimates and scheduling strategies. This strategy ensures stable execution of offline instruction sets and provides a positive user experience.

[0096] S205: Determine the instructions in the user interaction instructions that do not belong to the offline instruction set, and generate an online instruction set.

[0097] The online instruction set refers to the set of instructions that require a network connection to invoke cloud services; instruction computational complexity refers to the amount of computing resources required to execute the instruction; service dependency indicates the degree to which the instruction relies on cloud services; real-time requirements represent the timeliness constraints imposed on the results of instruction execution; data update frequency refers to the rate of change of instruction-related data; and load balancing indicates the distribution ratio of instructions between local and cloud services. For example, real-time traffic condition query instructions are often classified as online because they require frequent data updates.

[0098] After determining the offline instruction set, the data processing system needs to analyze and organize the remaining instructions. Specifically, the data processing system first filters out instructions not included in the offline instruction set from all interactive instructions. Then, the data processing system evaluates the characteristics of these instructions, including computational complexity, data dependencies, and real-time requirements. For instructions with high computational complexity, requiring real-time data updates, or relying on specific cloud services, the data processing system marks them as online instructions. Finally, based on quality of service requirements and load balancing strategies, the data processing system organizes these instructions into a structured online instruction set and establishes a corresponding service call interface.

[0099] In some embodiments, the generation of an online instruction set can be achieved in various ways: Optionally, the data processing system first constructs an instruction feature vector, including dimensions such as computational complexity, data dependency, and real-time requirements, then uses a support vector machine (SVM) classifier to evaluate the instructions, and finally filters and organizes the instructions based on the classification results and confidence levels; Optionally, the data processing system can also adopt an analysis method based on decision trees. First, establish multi-level decision rules, considering factors such as resource consumption, service dependencies, and data characteristics, then determine the online attributes of the instructions through rule reasoning, and finally organize the instructions based on service quality requirements. It can be understood that other machine learning or rule reasoning methods can also be used to generate the online instruction set, which is not limited here. Additionally, it should be noted that the impact of network condition changes on the execution of online instructions needs to be considered, and a corresponding fault tolerance mechanism should be established.

[0100] In practical applications, the management of the online instruction set may face problems such as unstable service quality and unbalanced resource allocation. In this regard, the data processing system can adopt a dynamic optimization management strategy: First, establish a service quality evaluation model Q(i)=f(t, r, d), where t represents the response time, r represents the resource consumption, and d represents the data real-time nature. Then define the service level target S={s_1, s_2,..., s_n}, and set different service requirements for different types of online instructions. When it is detected that the service quality Q(i)<s_i for a certain type of instruction, the system will start an adaptive adjustment mechanism: Dynamically adjust the local cache policy, optimize the service call frequency, and enable standby service nodes if necessary. At the same time, the system dynamically evaluates the service pressure through a load prediction algorithm P(t) to achieve the optimal allocation of local and cloud resources. This strategy can ensure the service quality of online instructions and improve the overall performance of the system.

[0101] In some embodiments, the data processing system will pre-acquire the response data of high-frequency instructions for backup storage. That is, the data processing system will determine the execution frequency and time distribution of each instruction in the online instruction set, and generate an instruction execution feature record; this instruction execution feature record includes instruction identification, historical execution time, and execution count; According to the instruction execution feature record, calculate the time aggregation degree and frequency ratio of each instruction in the online instruction set, and generate a high-frequency instruction list; Based on the historical execution time of each high-frequency instruction in the high-frequency instruction list, determine the data prefetch time window and prefetch data range of the high-frequency instructions, and generate a prefetch policy list; When it is detected that the data prefetch time window is entered, send a data prefetch request to the server according to the corresponding prefetch policy list, and receive the prefetch data returned by the server and store it in the local cache.

[0102] Among them, the instruction execution feature record represents the statistical description of instruction usage; the time concentration represents the concentration degree of instructions in the time dimension; the frequency ratio represents the usage ratio of a specific instruction in the total number of interactions; the data prefetch time window represents the time interval for the system to obtain data in advance; the prefetch strategy list represents the data preloading plan for different instructions, including specific parameters such as prefetch timing and data range.

[0103] The data processing system needs to analyze the usage characteristics of online instructions to implement intelligent prefetching. Specifically, the system first collects usage statistics for each instruction in the online instruction set, records its historical execution time points and cumulative execution times, and generates feature records containing the instruction identifier, execution time, and frequency. Based on these records, the system calculates the time concentration (the degree of concentration of the instruction's usage time distribution in a day) and frequency ratio (the proportion of the instruction's usage among all instructions) of each instruction, and filters instructions that meet preset conditions as high-frequency instructions. The system then analyzes the historical execution patterns of each high-frequency instruction, determines the optimal data prefetch time window and the data range to be prefetched, and forms a list of prefetch strategies. When the system detects that it has entered the prefetch time window for a certain instruction, it requests data from the server according to the corresponding prefetch strategy and stores the obtained data in the local cache to prepare for subsequent rapid response.

[0104] In some embodiments, instruction feature analysis and data prefetching can be achieved in a variety of ways: optionally, using a time series analysis method to perform periodic analysis on the instruction execution time series to identify the time patterns of instruction usage; using a density clustering algorithm to find the concentrated intervals of instruction execution time; based on statistical analysis of historical data, establishing a time probability distribution model for instruction execution. Optionally, combining user scenario characteristics, a multi-level prefetching strategy is constructed: prefetching is divided into three levels: short-term, medium-term, and long-term according to the time span; different prefetching data ranges and update strategies are adopted for different levels; and prefetching parameters are continuously optimized through feedback on prefetching effects. It is understandable that other predictive analysis methods can also be used to achieve data prefetching optimization, which is not limited here.

[0105] In practical applications, data prefetching can lead to wasted prefetching resources, meaning the system prefetches a large amount of data that is not actually used. To address this issue, the data processing system employs an adaptive prefetching strategy: First, a prefetching effectiveness evaluation mechanism is established to record the actual usage of each prefetched data; the prefetch hit rate is calculated, representing the proportion of prefetched data that is actually used; when the hit rate falls below a threshold, the system adjusts the prefetching strategy, including shortening the prefetching window and reducing the amount of prefetched data. Furthermore, data lifecycle management is introduced to promptly clear unused prefetched data to avoid excessive storage resource usage. Furthermore, the system dynamically adjusts the prefetching strategy based on factors such as device storage capacity and network conditions to achieve optimal resource utilization.

[0106] S206: Respond to the user's voice input and convert the voice input into an interaction instruction based on the local language model.

[0107] Referring to step S103 , the data processing system determines the interaction instruction based on the local language model.

[0108] In some embodiments, the data processing system will construct a specialized model for distinguishing user voice and language features, that is, the data processing system will obtain voice sample data of the target user in different emotional states; extract the voiceprint features and voice expression features of the voice sample data to construct a user pronunciation feature library; train the voice recognition model based on the user pronunciation feature library to generate a voice recognition model; and construct a local language model based on the voice recognition model and preset semantic understanding constraints; the semantic understanding constraints include the voice acquisition waiting time.

[0109] Among them, voice sample data represents the voice recording data of users in various emotional states; voiceprint features represent the unique voice feature parameters of the speaker; voice expression features represent the expression features such as intonation and speaking speed in the voice; voice recognition model represents the machine learning model used to identify user voice features; semantic understanding constraints represent the restrictions of the system when processing voice input.

[0110] The data processing system needs to build a recognition model that adapts to the user's personalized voice characteristics. Specifically, the system first collects voice samples of users in different states such as daily interactions and emotional fluctuations to ensure that the samples cover all possible voice performance situations. The collected voice samples are then subjected to feature extraction, including voiceprint features that can identify the user's identity, as well as voice expression features such as intonation, speaking speed, and pauses that reflect the user's expression habits. Based on these feature data, the system conducts targeted training on the general voice recognition model to generate a voice recognition model specifically adapted for the user. Finally, the system combines the trained voice recognition model with preset semantic understanding rules to build a local language model for accurately understanding the user's voice commands.

[0111] In some embodiments, the construction of a personalized speech model can be achieved in a variety of ways: optionally, using deep learning methods to train the voiceprint recognition model and extract multi-level acoustic features; combining emotional computing technology to identify emotional features in speech; and using transfer learning methods to fine-tune the general speech model in a personalized direction. Optionally, a multimodal speech understanding framework can be established: integrating speech, text, and contextual information to improve the accuracy of semantic understanding; constructing a user vocabulary usage profile to optimize the vocabulary coverage of the language model; and using an incremental learning strategy to continuously optimize model performance. It is understandable that other machine learning methods can also be used to achieve personalized optimization of the speech recognition model, which is not limited here.

[0112] In practice, elderly users' speech performance may be unstable, due to variations in their voice due to physical conditions. To address this, the data processing system employs a dynamic adaptation strategy: First, a voice quality assessment mechanism is established to monitor the clarity and stability of user voice input in real time. When significant changes in voice characteristics are detected, the system initiates a model adjustment process. Through incremental learning, the new voice features are integrated into the existing model. Meanwhile, the system continuously tracks the user's voice status and promptly adjusts parameters such as the voice acquisition wait time. Furthermore, the system creates multiple versions of the voice model backup to account for periodic changes in user voice characteristics.

[0113] S207: Build an index structure for local cache data.

[0114] The index structure represents the organizational form used to quickly locate and access cached data; cached data refers to historical response results and intermediate computational data stored locally; index keys identify different cached data items; data validity period represents the available time range for cached data; storage space limit refers to the maximum space available for caching on the local device; and access frequency refers to the number of times a data item is accessed. For example, weather forecast data might be stored using date as the index key.

[0115] After determining the instruction set, the data processing system needs to establish an efficient data access mechanism. Specifically, the data processing system first analyzes the data requirements of each instruction in the offline instruction set to determine the type and size of data to be cached. Then, based on the data access characteristics, the data processing system selects an appropriate index structure type, such as a hash table, B+ tree, or multi-level index. For each cached data item, the data processing system generates a unique index key value and records metadata such as the data's validity period and update time. Finally, the data processing system establishes a mapping relationship between the index and the data, and implements a fast query and update mechanism.

[0116] In some embodiments, the construction of the index structure can be achieved in a variety of ways: optionally, the data processing system first classifies and aggregates the cached data, establishes a multi-level index tree structure, stores the category index at the root node, stores the time index at the intermediate node, and stores the location information of the specific data item at the leaf node, and achieves rapid positioning by traversing the tree; optionally, the data processing system can also adopt a storage solution based on the LSM tree, first writes the new data into the memory table, merges and writes it to the disk when the data volume reaches the threshold, and maintains the Bloom filter to accelerate the query, and performs compression and cleanup regularly. It is understandable that other data structures or storage solutions can also be used to achieve the construction of the index structure, which is not limited here. It should be noted that the update and maintenance overhead of the index structure needs to be considered to balance query efficiency and storage space.

[0117] In practice, index structure maintenance can lead to excessive memory usage or degraded query performance. To address this, data processing systems can employ an adaptive cache management strategy: First, establish a data value assessment model: V(i) = w1*F(i) + w2*T(i) + w3*S(i), where F(i) represents access frequency, T(i) represents timeliness, S(i) represents data size, and w1, w2, and w3 are weight coefficients. When cache space utilization exceeds a threshold θ, the system triggers an elimination mechanism: The value scores of all data items are calculated and eliminated one by one, ranking from low to high, until space utilization reaches a safe level. Simultaneously, the system monitors query latency L(t). When L(t) exceeds a preset threshold, it automatically triggers an index reorganization to optimize the query path. This strategy enables efficient data access and management within limited resource constraints.

[0118] In some embodiments, the data processing system will classify and store local data based on the nature of the data, that is, the data processing system will collect query operation records of local cached data and generate a data access sequence; the data access sequence includes an access timestamp, a data item identifier, and an access source; based on the data access sequence, the query interval and query trigger method of each data item are determined to generate a data activity report; according to the average query interval in the data activity report, the data items are divided into high-frequency access data and low-frequency access data to generate a hierarchical storage scheme; the hierarchical storage scheme includes multiple storage schemes with different data update frequencies; an index structure is constructed based on the hierarchical storage scheme, the data items are stored in the corresponding storage area, and the storage location, timeliness mark, and associated instructions of the data items are recorded in the index structure.

[0119] Among them, the data access sequence represents the time series of usage records of locally cached data; the query interval represents the time difference between two adjacent accesses to the same data item; the query trigger method represents the type of cause of data access; the data activity represents the frequency with which the data item is accessed; and the hierarchical storage scheme represents a multi-level storage strategy formulated based on data access characteristics.

[0120] The data processing system needs to establish an efficient local data storage and access mechanism. Specifically, the system continuously records the access status of local cached data, including the timestamp of each access, the identifier of the data item being accessed, and the source of the access (such as active user query or automatic system call). Based on these access records, the system analyzes the access patterns of each data item, calculates the statistical characteristics of the query time interval, identifies the triggering pattern of the query behavior, and generates an analysis report reflecting the active level of data usage. Then, based on the average access interval of the data item, the system divides the data into multiple levels with different access frequencies, and formulates corresponding storage update strategies for each level. Finally, based on this hierarchical storage scheme, the system constructs an index structure to realize hierarchical storage of data, and maintains the location information, validity period and associated instruction information of the data item in the index.

[0121] In some embodiments, data storage optimization can be achieved in a variety of ways: optionally, a multi-level cache structure is used to keep high-frequency access data in the fast access layer and low-frequency data is downgraded to the slow storage layer; a prediction model is used to estimate the future access probability of data items and adjust the storage location in advance; and a dynamic migration mechanism for data between different storage layers is implemented. Optionally, an intelligent storage management strategy is constructed: differentiated update strategies are designed based on the timeliness characteristics of the data; a data dependency graph is established to ensure the consistency of related data; and an incremental update method is used to reduce data synchronization overhead. It is understandable that other storage optimization methods can also be used to improve the efficiency of data management, which is not limited here.

[0122] In practical applications, storage space constraints may arise, especially when large amounts of high-frequency data need to be cached. To address this issue, the data processing system employs a dynamic storage optimization strategy: First, a storage space monitoring mechanism is established to track space usage at each storage tier in real time. When the space at a particular tier approaches a threshold, a data cleanup process is triggered. The system calculates the retention value of data items based on multiple dimensions, such as access frequency, last access time, and associated dependencies. Data with low retention value is prioritized for cleanup, and some data may be downgraded to a lower storage tier if necessary. Furthermore, the system records and analyzes cleanup operations to optimize storage strategies and avoid frequent data migration and cleanup operations.

[0123] S208: When the interactive instruction belongs to an offline instruction set, a cache record corresponding to the interactive instruction is retrieved from the index structure.

[0124] A cache record represents a collection of historical data items related to an interaction instruction; a retrieval condition refers to the query parameters used to match cached data; the index hit rate represents the probability of successfully finding matching cached data; the retrieval path describes the search process for locating data; the data consistency marker is a marker that identifies the cached data version; and the query optimizer represents the component used to improve retrieval efficiency. For example, when querying the weather forecast for a specific day, the system needs to locate the corresponding weather data record in the index based on the date.

[0125] After receiving an offline instruction, the data processing system needs to quickly locate the relevant cached data. Specifically, the data processing system first parses the interactive instruction and extracts the key parameters used for retrieval. Then, the data processing system constructs index query conditions based on these parameters and calls the index structure's retrieval interface. During the retrieval process, the data processing system records the index traversal path and counts the number of nodes visited. If a matching cache record is found, the data processing system verifies the data's consistency flag to ensure that the data has not been modified by other operations. Finally, the data processing system returns a list of cached records that meet the conditions for subsequent processing.

[0126] In some embodiments, cache record retrieval can be achieved in a variety of ways: optionally, the data processing system first uses a multi-level filter chain, including a Bloom filter to quickly determine whether the data exists, a range filter to filter the time interval, and a tag filter to match attribute features, and finally performs an exact match in the candidate set to gradually narrow the search scope; optionally, the data processing system can also use a retrieval method based on a probability model, first establishing a query feature vector, using locality sensitive hashing (LSH) to quickly locate similar data blocks, and then determining the best matching result through similarity calculation to improve retrieval efficiency. It is understandable that other retrieval algorithms can also be used to achieve cache record search, which is not limited here. It should be noted that it is necessary to design special retrieval strategies for different types of query requests to optimize query performance.

[0127] In practical applications, cached retrieval may face high query latency or low hit rates. To address this, data processing systems can employ intelligent retrieval optimization strategies. First, a query performance monitoring model P(q) = {t, h, c} is established, where t represents query latency, h represents hit rate, and c represents computational overhead. Access patterns are then analyzed based on historical query logs, and a prediction algorithm A(t) is used to estimate the next access time of a data item. When the query performance metric P(q) falls below a threshold, the system initiates an adaptive optimization mechanism. For high-frequency query patterns, a query result cache is established in memory; highly relevant data items are preloaded into a fast-access area; and for complex query conditions, the index structure is dynamically adjusted, with the addition of auxiliary indexes. Optimization decisions are guided by a query cost estimation function C(q) = αt + βm (where m represents memory usage and α and β are weighting coefficients). This strategy can significantly improve retrieval efficiency and enhance the user experience.

[0128] S209: Verify the timeliness mark of the cache record to determine the valid status of the data.

[0129] The timeliness tag indicates the expiration date and update timestamp of cached data; data validity status indicates whether the cached data is available for the current request; the expiration policy defines the rules for determining data expiration; the update cycle indicates the time interval at which data needs to be refreshed; the data priority is the importance ranking of cached data; and the consistency level indicates the strictness of data updates. For example, weather forecast data may be valid for 24 hours, and after that, it will be marked as invalid.

[0130] After obtaining the cache record, the data processing system needs to ensure the timeliness of the data. Specifically, the data processing system first reads the timeliness mark of the cache record, including the creation time, update time, and validity period. Then, the data processing system calculates the existence time of the data based on the current time and compares it with the preset validity period. For different types of data, the data processing system will apply different expiration strategies, such as absolute time determination or relative time determination. If the data is still within the validity period, it is marked as available; if the data has expired, it is marked as invalid. The data processing system also checks the consistency level of the data to ensure that the data meets current business needs.

[0131] In some embodiments, data validity verification can be achieved in various ways: Optionally, the data processing system first establishes a multi-dimensional verification rule set, including checking whether the data is expired in the time dimension, verifying whether the data is up-to-date in the version dimension, and confirming whether the data is corrupted in the integrity dimension. Then, item-by-item verification is performed through a rule engine, and finally, a comprehensive evaluation is conducted to generate a validity determination result. Optionally, the data processing system can also adopt a probability-based verification method. First, a data reliability model is constructed, considering factors such as data age, update frequency, and usage frequency. Then, the reliability score is calculated, and the data status is determined by setting a threshold. It can be understood that other verification algorithms can also be used to achieve the determination of data validity, which is not limited here. Additionally, the system needs to maintain a data status conversion log for easy tracking and analysis of the data life cycle.

[0132] In practice, data validity verification may encounter problems such as inconsistent determination criteria or excessive verification overhead. In this regard, the data processing system can adopt a hierarchical verification strategy: First, establish a data timeliness evaluation model E(d)=f(A, U, I), where A represents data age, U represents update frequency, and I represents importance. Then, set a hierarchical verification mechanism: For high-priority data (I>threshold_h), adopt a strict verification strategy, including integrity check and version consistency check; for medium-priority data (threshold_l<I≤threshold_h), use timestamp-based fast verification; for low-priority data (I≤threshold_l), adopt a sampling verification method. At the same time, the system uses a verification cost function C(v)=τt + μm (where t is the verification time, m is the memory consumption, and τ and μ are weight coefficients) to balance the verification intensity and system overhead. This strategy can improve the verification efficiency while ensuring data reliability.

[0133] S210. Generate a local response result based on the data valid state and associated instructions, and display the local response result.

[0134] Among them, the local response result represents the processing result generated based on cached data; the associated instructions refer to the set of instructions related to the current data processing; the response template is used to define the organization form of the result; the rendering rule represents the style and layout of the result display; the fallback scheme refers to the alternative processing scheme when the data is invalid; the user preference setting represents personalized display requirements. For example, a weather application may select different weather display styles according to user settings.

[0135] After confirming the data's validity, the data processing system needs to generate a response suitable for display. Specifically, the data processing system first analyzes the data's validity to determine whether a degradation solution should be activated. Then, the data processing system calculates and converts the valid data based on the processing logic of the associated instructions. For complex data processing, the data processing system executes step-by-step according to a predefined processing flow. After generating the results, the data processing system selects appropriate response templates and rendering rules based on user preferences and device characteristics, converting the results into a user-friendly display format. Finally, the data processing system displays the processing results through interface components and records user interaction feedback.

[0136] In some embodiments, the generation and display of response results can be achieved in a variety of ways: optionally, the data processing system first builds a response result generation pipeline, including cleaning and normalizing data in the data preprocessing stage, executing business logic in the calculation processing stage, performing format conversion and optimization in the post-processing stage, and finally generating the final display result through template engine rendering; optionally, the data processing system can also adopt a component-based response solution, first decomposing the complex response logic into independent processing components, then defining the processing flow through component orchestration, and finally integrating the output of each component to generate a complete response. It is understandable that other result generation methods can also be used to implement response processing, which is not limited here. It should be noted that the system needs to formulate corresponding display strategies for different types of response results.

[0137] In practical applications, response generation can face challenges such as complex processing logic or poor presentation quality. To address this, the data processing system can employ an intelligent response processing strategy. First, a response quality assessment model Q(r) = g(p, t, u) is established, where p represents processing accuracy, t represents response time, and u represents user satisfaction. The processing strategy is then dynamically adjusted based on the assessment results. When processing time T(r) exceeds a threshold, simplified processing logic is enabled. When presentation quality E(r) is unsatisfactory, an alternative template is automatically switched. The system also maintains a response optimizer O(r), which continuously optimizes processing parameters using machine learning methods: ω(t+1) = ω(t) + η∇Q(r), where ω represents the processing parameter and η represents the learning rate. This strategy provides a stable and efficient response experience that meets the needs of diverse users.

[0138] S211: When the interactive instruction belongs to an online instruction set, determine the network connection status.

[0139] Referring to step S105 , the data processing system determines the network connection status.

[0140] S212: When the network connection status is normal, send an interactive instruction to the server, receive an online response result returned by the server, and display the online response result.

[0141] Referring to step S106 , the data processing system will display the online response result when there is no abnormality in the network.

[0142] S213: When the network connection status is abnormal, determine historical response data corresponding to the interaction instruction based on the historical interaction data, generate a temporary response result based on the historical response data, and display the temporary response result.

[0143] Referring to step S107 , the data processing system will display a temporary response result when the network is abnormal.

[0144] In the embodiments of the present application, due to the use of an instruction classification mechanism based on user interaction patterns, a local voice recognition model and a multi-level data caching strategy, it is possible to accurately understand the usage habits of elderly users and provide corresponding localized service support, effectively solving the problems of over-reliance on network connections, easy service interruptions, and poor interactive experience in related technologies; by constructing a user interaction pattern, the user's query needs are accurately predicted, and intelligent pre-fetching and local caching of data are realized; through the local voice recognition model, user instructions are quickly responded to and network transmission overhead is reduced; through a hierarchical storage strategy, device storage space is rationally utilized to ensure fast access to high-frequency data; through a dynamic optimization mechanism, the personalization level of services is continuously improved, so that the system can better adapt to the usage characteristics of elderly users; and a stable, reliable, and elderly-friendly intelligent service system is constructed.

[0145] The data processing system in the embodiment of the present invention is described below from the perspective of hardware processing. Figure 3 , is a schematic diagram of a physical device structure of a data processing system in an embodiment of the present application.

[0146] It should be noted that Figure 3 The structure of the data processing system shown is only an example and should not limit the functions and scope of use of the embodiments of the present invention.

[0147] like Figure 3 As shown, the data processing system includes a CPU 301, which can perform various appropriate actions and processes according to the programs stored in the ROM 302 or the programs loaded from the storage unit 308 into the RAM 303, such as executing the methods described in the above embodiments. The RAM 303 also stores various programs and data required for system operation. The CPU 301, ROM 302, and RAM 303 are connected to each other via a bus 304. An I / O interface 305 is also connected to the bus 304.

[0148] The following components are connected to the I / O interface 305: an input section 306 including an audio input device, push button switches, and the like; an output section 307 including a liquid crystal display (LCD), an audio output device, indicator lights, and the like; a storage section 308 including a hard disk and the like; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card or a modem. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. Removable media 311, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 310 as needed, so that computer programs read from the removable media can be installed in the storage section 308 as needed.

[0149] In particular, according to an embodiment of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, an embodiment of the present invention includes a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 309 and / or installed from the removable medium 311. When the computer program is executed by the CPU 301, the various functions defined in the present invention are performed.

[0150] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. Each box in the flowchart or block diagram can represent a module, program segment, or part of the code, and the above-mentioned module, program segment, or part of the code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings.

[0151] Specifically, the data processing system of this embodiment includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, the data display method for the aging-friendly service provided by the above embodiment is implemented.

[0152] As another aspect, the present invention further provides a computer-readable storage medium, which may be included in the data processing system described in the above embodiments, or may exist independently and not incorporated into the data processing system. The storage medium carries one or more computer programs. When executed by a processor of the data processing system, the one or more computer programs enable the data processing system to implement the data display method for aging-friendly services provided in the above embodiments.

[0153] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0154] As used in the above embodiments, the term “when” may be interpreted to mean “if” or “after” or “in response to determining that” or “in response to detecting that”, depending on the context. Similarly, the phrases “upon determining that” or “if (stated condition or event) is detected” may be interpreted to mean “if determining that” or “in response to determining that” or “upon detecting (stated condition or event)” or “in response to detecting (stated condition or event)”, depending on the context.

Claims

1. A data display method for aging-friendly services, characterized in that: Applied to a data processing system, the method comprises: Collect historical interaction data of the target user, and determine the user interaction pattern based on the interaction frequency, interaction time, and interaction content in the historical interaction data; Based on the user interaction mode, the user interaction instructions are divided into an offline instruction set and an online instruction set; the offline instruction set is a plurality of instructions for the user to query local data; In response to a user's voice input, convert the voice input into an interaction instruction based on a local language model; When the interactive instruction belongs to the offline instruction set, generating a local response result based on the corresponding local cache data, and displaying the local response result; When the interactive instruction belongs to the online instruction set, determining a network connection state; When the network connection status is normal, sending the interaction instruction to the server, receiving an online response result returned by the server, and displaying the online response result; When the network connection status is abnormal, historical response data corresponding to the interaction instruction is determined based on the historical interaction data, a temporary response result is generated based on the historical response data, and the temporary response result is displayed.

2. The method according to claim 1, characterized in that The step of collecting historical interaction data of the target user and determining the user interaction pattern based on the interaction frequency, interaction time, and interaction content in the historical interaction data specifically includes: Collect historical interaction data of the target user, including active interaction data and passive interaction data; the active interaction data is the voice command initiated by the user, and the passive interaction data is the user's feedback operation on the system push information; Based on the time sequence and the operation type, a correlation map between the active interaction data and the passive interaction data is constructed to generate a user interaction record; Extracting time-related features, instruction-related features, and response-related features from the user interaction record, constructing a user behavior link, and generating a user interaction pattern based on the user behavior link.

3. The method according to claim 1, characterized in that The step of dividing the user interaction instructions into an offline instruction set and an online instruction set based on the user interaction mode specifically includes: Based on the user interaction pattern, determining the execution dependency and data dependency of the user interaction instructions, and constructing an instruction association graph; Determining independently executed instructions and sequentially associated instructions based on the instruction association graph; the independently executed instructions are instructions that only depend on local data; and the sequentially associated instructions are instructions that are executed in a fixed order; Combining the independent execution instructions and the timing-related instructions into the offline instruction set; Determine instructions in the user interaction instructions that do not belong to the offline instruction set, and generate an online instruction set.

4. The method according to claim 3, characterized in that After the step of determining instructions in the user interaction instructions that do not belong to the offline instruction set and generating an online instruction set, the method further includes: Determine the execution frequency and time distribution of each instruction in the online instruction set, and generate an instruction execution feature record; the instruction execution feature record includes an instruction identifier, a historical execution time, and an execution count; Calculating the time concentration and frequency ratio of each instruction in the online instruction set based on the instruction execution feature records, and generating a high-frequency instruction list; Determining a data prefetch time window and a prefetch data range of the high-frequency instructions based on a historical execution time of each high-frequency instruction in the high-frequency instruction list, and generating a prefetch strategy list; When it is detected that the data prefetch time window has arrived, a data prefetch request is sent to the server according to the corresponding prefetch strategy list, and the prefetched data returned by the server is received and stored in the local cache.

5. The method according to claim 1, wherein Before the step of responding to the user's voice input and converting the voice input into an interaction instruction based on the local language model, the method further includes: Acquire speech sample data of the target user in different emotional states; Extracting voiceprint features and speech expression features of the speech sample data to construct a user pronunciation feature library; Training a speech recognition model based on the user pronunciation feature library to generate a speech discrimination model; The local language model is constructed based on the speech recognition model and preset semantic understanding constraints; the semantic understanding constraints include a speech acquisition waiting time.

6. The method according to claim 1, characterized in that When the interactive instruction belongs to the offline instruction set, generating a local response result based on the corresponding local cache data and displaying the local response result specifically includes: Constructing an index structure for local cached data; the index structure includes data identifiers, timeliness tags, and associated instructions; When the interactive instruction belongs to the offline instruction set, searching the index structure for a cache record corresponding to the interactive instruction; Verifying the timeliness mark of the cache record to determine the validity status of the data; The local response result is generated according to the data validity status and the associated instruction, and the local response result is displayed.

7. The method according to claim 6, characterized in that The step of constructing the index structure of the local cache data specifically includes: Collect query operation records of local cache data and generate a data access sequence; the data access sequence includes an access timestamp, a data item identifier, and an access source; Based on the data access sequence, determining the query interval and query triggering method for each data item, and generating a data activity report; Divide data items into high-frequency access data and low-frequency access data based on the average query interval in the data activity report, and generate a hierarchical storage solution; the hierarchical storage solution includes multiple storage solutions with different data update frequencies; An index structure is constructed based on the hierarchical storage solution, data items are stored in corresponding storage areas, and the storage location, timeliness mark and associated instructions of the data items are recorded in the index structure.

8. A data processing system, characterized in that: The data processing system includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to cause the data processing system to execute the method according to any one of claims 1 to 7.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on a data processing system, the data processing system is caused to execute the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product is run on a data processing system, the computer program product causes the data processing system to execute the method according to any one of claims 1 to 7.

Images