Operation control method and device based on user behaviors and storage medium

Abstract

The invention relates to an operation control method and device based on user behaviors and a storage medium. The operation control method based on the user behavior comprises the following steps: determining a current event triggered by the user behavior; in response to a target condition matched with the current event in the user behavior template, determining a target execution operation corresponding to the target condition according to the user behavior template; wherein the user behavior template comprises a corresponding relation between a condition and an execution operation, the user behavior template is obtained by screening candidate user behavior templates based on a large language model, and the candidate behavior template comprises a candidate corresponding relation between the condition and the execution operation, which is obtained by mining based on an event data set; the event data set comprises events associated with user behaviors, and the execution operation is executed when corresponding conditions are met; and executing the target execution operation. According to the invention, reliable user habit inference is provided, and the user experience is improved.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and in particular to an operation control method, apparatus and storage medium based on user behavior. Background Technology

[0002] With the continuous development of computer technology, people's pursuit of quality of life is also constantly improving.

[0003] In related technologies, operation control is usually based on the correlation between different devices, which is cumbersome and not conducive to improving the user experience. Summary of the Invention

[0004] To overcome the problems existing in related technologies, this disclosure provides an operation control method, device and storage medium based on user behavior.

[0005] According to a first aspect of the present disclosure, a user behavior-based operation control method is provided, comprising: determining a current event triggered by a user behavior; responding to the existence of a target condition in a user behavior template that matches the current event, determining a target execution operation corresponding to the target condition according to the user behavior template; wherein the user behavior template includes a correspondence between conditions and execution operations, and the user behavior template is obtained by filtering candidate user behavior templates based on a large language model, the candidate behavior template includes candidate correspondences between conditions and execution operations obtained by mining an event dataset, the event dataset includes events associated with user behavior, wherein the execution operation is performed when the corresponding condition is met; and executing the target execution operation.

[0006] In one implementation, the candidate user behavior template is obtained by mining an event dataset in the following manner: collecting multiple user behavior-related events to obtain an event dataset; determining the event data of the i-th user in the event dataset, and classifying the event data of the i-th user to obtain M. i Class conditions and N i The class performs the operation; for the M i Class conditions and N i The operations of each class are combined in pairs to form M. i *N i Class combination; determine the M i *N i The correlation metric between the condition and the operation in each combination is determined; combinations with correlation metrics greater than the correlation metric threshold are used as candidate behavior templates for the i-th user; the above process is repeated until candidate behavior templates for the multiple users are determined.

[0007] In one embodiment, the correlation metric between conditions and execution operations in each combination is determined as follows: the total number of execution operations and the total number of conditions appearing in each combination are counted; the ratio between the total number of execution operations and the total number of conditions is determined; the ratio is multiplied by a first coefficient to obtain a first value, and the total number of execution operations is multiplied by a second coefficient to obtain a second value; the sum of the first value and the second value is determined as the correlation metric between conditions and execution operations.

[0008] In one implementation, the user behavior template is obtained by filtering candidate user behavior templates based on a large language model in the following manner: setting prompt words, which are used to prompt the large language model to judge the rationality of the candidate user behavior templates; filling the prompt words with the conditions and execution operations in the candidate user behavior templates according to the text format of the prompt words to obtain prompt words corresponding to the candidate user behavior templates; inputting the prompt words corresponding to the candidate user behavior templates into the large language model to obtain the rationality judgment result of the candidate user behavior templates; filtering out candidate user behavior templates with reasonable rationality judgment results, and obtaining user behavior templates based on reasonable candidate user behavior templates.

[0009] In one embodiment, obtaining a user behavior template based on a reasonable candidate user behavior template includes: acquiring event data corresponding to each user; matching events as conditions from the event data corresponding to the user in the reasonable candidate user behavior template; statistically analyzing the execution operations corresponding to the events as conditions in the reasonable candidate user behavior template, and determining the probability of the corresponding execution operation occurring when the event as condition is satisfied; and using the execution operations with a probability greater than a threshold, and the conditions corresponding to the execution operations, as the user behavior template.

[0010] In one embodiment, the step of collecting multiple user behavior-related events to obtain an event dataset includes: collecting user behavior data within a preset time period and collecting environmental state data at the time the user behavior data occurs; wherein, the user behavior data includes user operation events on objects collected by sensors and user activity events obtained by perception algorithms; the environmental state data includes environmental states collected by sensors and environmental state change events identified by perception algorithms; and timestamps corresponding to the time of collection are added to the user behavior data and the environmental state data respectively to generate an event dataset.

[0011] According to a second aspect of the present disclosure, an operation control device based on user behavior is provided, comprising: a determining unit, configured to determine a current event triggered by user behavior; a processing unit, configured to, in response to the existence of a target condition in a user behavior template that matches the current event, determine a target execution operation corresponding to the target condition according to the user behavior template; wherein the user behavior template includes a correspondence between conditions and execution operations, and the user behavior template is obtained by filtering candidate user behavior templates based on a large language model, the candidate behavior template includes candidate correspondences between conditions and execution operations obtained by mining an event dataset, the event dataset includes events associated with user behavior, and the execution operation is performed when the corresponding condition is met; and an execution unit, configured to execute the target execution operation.

[0012] In one implementation, the processing unit obtains candidate user behavior templates based on an event dataset by mining the dataset as follows: collecting multiple user behavior-related events to obtain an event dataset; determining the event data of the i-th user in the event dataset, and classifying the event data of the i-th user to obtain M. i Class conditions and N i The class performs the operation; for the M i Class conditions and N i The operations of each class are combined in pairs to form M. i *N i Class combination; determine the M i *N i The correlation metric between the condition and the operation in each combination is determined; combinations with correlation metrics greater than the correlation metric threshold are used as candidate behavior templates for the i-th user; the above process is repeated until candidate behavior templates for the multiple users are determined.

[0013] In one embodiment, the processing unit determines the correlation metric between conditions and execution operations in each combination as follows: it counts the total number of execution operations and the total number of conditions in each combination; it determines the ratio between the total number of execution operations and the total number of conditions; it multiplies the ratio by a first coefficient to obtain a first value, and multiplies the total number of execution operations by a second coefficient to obtain a second value; and it determines the sum of the first value and the second value as the correlation metric between conditions and execution operations.

[0014] In one embodiment, the processing unit obtains user behavior templates by filtering candidate user behavior templates based on a large language model in the following manner: setting prompt words, which are used to prompt the large language model to judge the rationality of the candidate user behavior templates; filling the prompt words with the conditions and execution operations in the candidate user behavior templates according to the text format of the prompt words to obtain prompt words corresponding to the candidate user behavior templates; inputting the prompt words corresponding to the candidate user behavior templates into the large language model to obtain the rationality judgment result of the candidate user behavior templates; filtering out candidate user behavior templates with reasonable rationality judgment results, and obtaining user behavior templates based on reasonable candidate user behavior templates.

[0015] In one embodiment, the processing unit obtains a user behavior template based on a reasonable candidate user behavior template in the following manner: for each user, it acquires the event data corresponding to the user; it matches the events that serve as conditions in the event data corresponding to the user in the reasonable candidate user behavior template; it counts the execution operations corresponding to the events that serve as conditions in the reasonable candidate user behavior template, and determines the probability of the corresponding execution operation occurring when the event that serves as the condition is satisfied; it uses the execution operations with a probability greater than a threshold, and the conditions corresponding to the execution operations, as the user behavior template.

[0016] In one embodiment, the processing unit collects multiple user behavior-related events in the following manner to obtain an event dataset: user behavior data is collected within a preset time period, and environmental state data is collected when the user behavior data occurs; wherein, the user behavior data includes user operation events on objects collected by sensors, and user activity events obtained based on a perception algorithm; the environmental state data includes environmental states collected by sensors, and environmental state change events identified based on a perception algorithm; and timestamps corresponding to the collection time are added to the user behavior data and the environmental state data respectively to generate the event dataset.

[0017] According to a third aspect of the present disclosure, an electronic device is provided, comprising: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to execute the executable instructions to perform the method of the first aspect or any one of the embodiments of the first aspect.

[0018] According to a fourth aspect of the present disclosure, a storage medium is provided that stores instructions which, when executed by a processor, enable the execution of the method in the first aspect or any of the embodiments of the first aspect.

[0019] The technical solutions provided by the embodiments of this disclosure can include the following beneficial effects: by combining large language models with data mining techniques, the accuracy and efficiency of user habit mining are improved. Furthermore, by utilizing current events and conducting multiple filtering processes, the effectiveness and reliability of user behavior templates are ensured, thereby improving user convenience and quality of life.

[0020] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0021] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0022] Figure 1 This is a flowchart illustrating an operation control method based on user behavior according to an exemplary embodiment.

[0023] Figure 2 This is a flowchart illustrating an example of obtaining an event dataset according to an exemplary embodiment.

[0024] Figure 3 This is a flowchart illustrating a method for mining candidate user behavior templates according to an exemplary embodiment.

[0025] Figure 4 This is a flowchart illustrating a method for determining a correlation metric, according to an exemplary embodiment.

[0026] Figure 5 This is a flowchart illustrating a method for determining a user behavior template according to an exemplary embodiment.

[0027] Figure 6 This is a flowchart illustrating a method for determining a user behavior template according to an exemplary embodiment.

[0028] Figure 7 This is a block diagram of an operation control device based on user behavior, according to an exemplary embodiment.

[0029] Figure 8 This is a block diagram of an apparatus for operation control based on user behavior, according to an exemplary embodiment.

[0030] Figure 9 This is a block diagram of an apparatus for operation control based on user behavior, according to an exemplary embodiment. Detailed Implementation

[0031] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure.

[0032] The user behavior-based operation control method provided in this disclosure is applied to scenarios where user habits are determined. The user behavior-based operation control method involved in this disclosure is mainly applied to scenarios where user habits are mined.

[0033] Habits refer to automated behavioral patterns formed by an individual in specific situations. These patterns reflect an individual's tendency to repeat behaviors in similar situations. These behavioral patterns can be influenced by various factors, including personal preferences, environmental factors, and socio-cultural factors. For example, someone enthusiastic about smart homes might perform a series of actions in a certain order after get off work: first, unlock the front door; then, turn on the living room lights and stereo one by one; draw the curtains; and finally, stream music from their phone to the stereo. When the indoor temperature exceeds 25 degrees Celsius, they might also habitually turn on the air conditioner and set it to vertical airflow mode. Before going to bed, they might habitually set their phone to Do Not Disturb mode to ensure sleep quality.

[0034] By analyzing user habits and patterns within massive amounts of data, automated device control can be achieved, providing users with a more convenient living experience and thus improving their quality of life. For example, based on user habits, the system can automatically adjust light brightness, speaker volume, air conditioning temperature, and mobile phone screen brightness under specific conditions. Such intelligent device control systems not only save users time and effort but also create a more comfortable and convenient living environment, thereby improving their quality of life.

[0035] In recent years, with the rapid development of artificial intelligence, large language models have become an indispensable force leading technological trends. These large language models have made significant progress in natural language processing, not only generating fluent and accurate text but also possessing profound semantic understanding capabilities. By combining large language models with traditional data mining techniques, it is possible to delve deeper into the potential information and patterns hidden within data, thereby significantly improving the efficiency and accuracy of data analysis. However, despite the promising prospects of this field, in-depth exploration has yet to be conducted.

[0036] In the related art, it usually involves using the association relationship between devices to perform linkage control on devices, which cannot be directly applied to user habit mining. Habit mining focuses on the causal relationship between events, that is, after event 1 occurs, the user executes event 2. However, the related art mainly focuses on correlation rather than causality. In addition, in the related art, it is also impossible to focus on the characteristics of the events themselves, and the characteristics of the events themselves may have an important impact on the user's preferences and behaviors. In addition, the sparsity and noise of the data also bring certain uncertainties to the calculation of the association degree, thus affecting the accuracy of the recommendation.

[0037] In view of this, the present disclosure provides an operation control method based on user behavior, which determines an execution action having a causal relationship with the user's habit according to the user's habit, thereby improving the accuracy of the event execution by the device to be executed.

[0038] The target habit template involved in the embodiments of the present disclosure can be understood as a habit template between devices as conditions and executions, or can be understood as a habit template between user behaviors and devices as conditions and executions.

[0039] Figure 1 is a flowchart of an operation control method based on user behavior shown according to an exemplary embodiment. As Figure 1 shown, it includes the following steps.

[0040] In step S11, determine the current event triggered by the user behavior.

[0041] In the embodiments of the present disclosure, the action currently executed by the user or the change in the environmental state can be determined according to the user behavior data and the environmental state data, and the event that has changed is determined as the current event.

[0042] In step S12, in response to the existence of a target condition in the user behavior template that matches the current event, determine the target execution operation corresponding to the target condition according to the user behavior template.

[0043] Among them, the user behavior template includes the corresponding relationship between the condition and the execution operation, and the user behavior template is obtained by screening the candidate user behavior templates based on a large language model. The candidate behavior templates include the candidate corresponding relationships between the conditions and the execution operations mined from the event dataset. The event dataset includes events associated with the user behavior, and the execution operation is executed under the condition that the corresponding condition is satisfied.

[0044] In step S13, execute the target execution operation.

[0045] In the embodiments of the present disclosure, the target execution operation is executed according to the current event triggered by the user behavior.

[0046] According to exemplary embodiments of this disclosure, a target operation is determined by matching a user behavior template with the current event. This approach combines large language models with data mining techniques to improve the accuracy and efficiency of user habit mining. Through multi-stage filtering, the effectiveness and reliability of the user behavior template are ensured, thereby enhancing user convenience and quality of life.

[0047] In exemplary embodiments of this disclosure, the following are employed: Figure 2 The method shown collects multiple user behavior-related events to obtain an event dataset. Figure 2 This is a flowchart illustrating an example of obtaining an event dataset. For example... Figure 2 As shown, it includes the following steps.

[0048] In step S21, user behavior data is collected within a preset time period, and environmental state data is collected when user behavior data occurs.

[0049] The user behavior data includes user operation events on objects collected by sensors, as well as user activity events obtained by perception algorithms; the environmental state data includes environmental state collected by sensors, as well as environmental state change events identified by perception algorithms.

[0050] In step S22, timestamps corresponding to the time of collection are added to the user behavior data and environmental status data respectively to generate an event dataset.

[0051] According to exemplary embodiments of this disclosure, user behavior data and environmental perception data are collected to construct an event dataset, which improves the accuracy of subsequent analysis using the time dataset and reduces erroneous correlation combinations introduced by data sparsity and noise.

[0052] In this embodiment of the disclosure, real-time data from multiple users over a period of time is collected and timestamped to form an event dataset. The event dataset includes user behavior data and environmental state data. The user behavior data includes directly recorded user operation data on the device, as well as user activity events indirectly captured by one or more sensors and identified by a perception algorithm. The environmental state data includes the environmental state output by sensors and environmental state change events identified by the perception algorithm.

[0053] The event dataset stores real-time data from different users over a given period of time. This period can be set according to actual data collection needs; for example, it could be set to one week, or, to make the resulting event dataset more universally applicable, it could be set to one month. An event dataset can be understood as the events that occurred at each of the multiple points in time within a given period.

[0054] The data collection source can be a device that performs corresponding operations according to user habits. This device can collect data by directly obtaining data from the cloud, or by reading data stored locally, or by linking with other devices to read and combine data from other devices. This disclosure does not limit the collection method.

[0055] In one exemplary embodiment, for ease of understanding of the above content, the event dataset involved in this disclosure may include the following content in the following form:

[0056] Sunrise at XX:XX:XX on XX:XX in XX month XX year.

[0057] On XX month XX day at XX hour XX minute XX second, the temperature in room XX was XX degrees Celsius.

[0058] On XX month XX day at XX hour XX minute XX second, the door lock of room XX was opened by XX.

[0059] On XX month XX day at XX hour XX minute XX second, the air conditioner in room XX was turned on.

[0060] In exemplary embodiments of this disclosure, the following are employed: Figure 3 The method shown is to mine candidate user behavior templates based on the event dataset. Figure 3 This is a flowchart illustrating a method for mining candidate user behavior templates according to an exemplary embodiment. For example... Figure 3 As shown, it includes the following steps.

[0061] In step S31, multiple user behavior-related events are collected to obtain an event dataset.

[0062] In this embodiment of the disclosure, user behavior-related events of multiple users over a period of time are collected to form an event dataset.

[0063] User behavior-related events can be understood as events with added timestamps, and these events include conditions and actions performed. The event dataset can be understood as primarily comprising two aspects: user behavior data and environmental state data.

[0064] In step S32, the event data of the i-th user in the event dataset is determined, and the event data of the i-th user is classified to obtain M. i Class conditions and N i The class performs the operation.

[0065] In step S33, for M i Class conditions and N i The operations are combined in pairs to form M. i *Ni Class composition.

[0066] In step S34, M is determined. i *N o The correlation measure between conditions and execution operations in each category of the combination.

[0067] In step S35, combinations with relevance measure values ​​greater than the relevance measure threshold are used as candidate behavior templates for the i-th user.

[0068] In step S36, the above process is repeated until candidate behavior templates for multiple users are determined.

[0069] In one exemplary embodiment, the events in the obtained event dataset are divided into events that serve as conditions and events that serve as operations to be performed.

[0070] In this embodiment of the disclosure, for real-time data of multiple users within a time period existing in the event dataset, the real-time data of one user within a time period is randomly selected. Data that can be used as conditions and data that can be used as execution operations are extracted from the obtained real-time data. The extracted data used as conditions and the extracted data used as execution operations are respectively classified to obtain M... i The conditions of the class, and having N i The set of execution operations for the class. The resulting M... i The conditions of the class, and N i Multiplying the execution operations of a class yields multiple combinations between the condition and the execution operation. These combinations are considered as the habits of the i-th type of user.

[0071] In this context, the data used as conditions can be all data in the event dataset, and the data used for execution operations can be user operation data recorded in the event dataset. In this embodiment, the data used as conditions can also be used as data for execution operations; conditions can include execution operations, and the M involved in this disclosure... i Greater than N i M i This can be understood as extracting M types of conditions from the real-time data recorded by the i-th user within a certain period of time in the event dataset. i This can be understood as extracting N types of operations from the real-time data recorded by the i-th user within a certain period of time in the event dataset.

[0072] The execution operation in the event dataset can be the user's operation data on the device extracted from the event dataset after a preset time when the conditions occur.

[0073] According to an exemplary embodiment of this disclosure, conditions and operations are classified from an event dataset containing multiple users to improve the coverage of user habits. Furthermore, the classified conditions and operations are combined in pairs, and combinations with a correlation metric value greater than a correlation metric threshold are identified as candidate templates for users, thereby improving the universality of user candidate templates.

[0074] In the exemplary embodiments of this disclosure, the correlation metric between conditions and execution operations in each type of combination is adopted as follows: Figure 4 The method shown is used to determine this. Figure 4 This is a flowchart illustrating a method for determining a correlation metric, according to an exemplary embodiment. For example... Figure 4 As shown, it includes the following steps.

[0075] In step S41, the total number of execution operations and the total number of conditions in each combination are counted.

[0076] In this embodiment of the disclosure, for the obtained M i *N i The number of execution operations and the number of conditions appearing in each category of the combination are counted separately to obtain the total number of execution operations and the total number of conditions appearing in each category of the combination.

[0077] In step S42, the ratio between the total number of operations performed and the total number of conditions is determined.

[0078] In step S43, the ratio is multiplied by the first coefficient to obtain the first value, and the total number of operations is multiplied by the second coefficient to obtain the second value.

[0079] The first coefficient can be understood as a value set based on experience, and the second coefficient can be understood as a value set based on experience; this disclosure does not limit this.

[0080] In step S44, the sum of the first value and the second value is determined as the correlation measure between the condition and the execution operation.

[0081] In this embodiment of the disclosure, M i Class conditions and N i For each of the multiple combinations resulting from the execution of a class's operations, an event correlation analysis is performed. This yields a correspondence between the conditions and the executed operations that demonstrates event correlation.

[0082] The implementation method can be to select M. i Class conditions and N i The j-th combination among the multiple combinations obtained from the execution operation of the class is counted, and the number of times y appears as a candidate condition in the j-th combination is counted. j And the number of times x appears as a candidate for executionj The cumulative number of times y appears as a candidate condition in the combinations of class j. j And the sum of the candidate conditions is recorded as Y. i The cumulative number of times x appears as a candidate for execution in the j-th combination is calculated. j And record the cumulative result of the candidate execution as X. j Their mathematical relationship can be expressed as:

[0083] Y j =∑y i

[0084] X j =∑x i

[0085] The relevance measure of the j-th type of combination is calculated based on the cumulative results of candidate conditions and candidate executions, and M is calculated in the same way. i Class conditions and N i The correlation measure of each combination in the set of multiple combinations obtained by the execution operation of the class is sorted from high to low, the correlation measures below the set threshold are filtered out, and the remaining combinations are regarded as candidate behavior templates.

[0086] The correlation measure of the j-th type of combination can be calculated using the following mathematical relationship based on the conditional summation result and the execution summation result:

[0087]

[0088] Where k1 and k2 are both given constants, Z j This indicates a correlation measure.

[0089] In exemplary embodiments of this disclosure, the following are employed: Figure 5 The method shown is based on a large language model to filter candidate user behavior templates to obtain user behavior templates. Figure 5 This is a flowchart illustrating a method for determining a user behavior template according to an exemplary embodiment. For example... Figure 5 As shown, it includes the following steps.

[0090] In step S51, a prompt word is set.

[0091] Among them, prompt words are used to prompt the large language model to judge the rationality of candidate user behavior templates.

[0092] In step S52, the conditions and operations in the candidate user behavior template are filled into the prompt words according to the text format of the prompt words to obtain the prompt words of the corresponding candidate user behavior template.

[0093] The text format can be a fixed text form, such as when (candidate condition), the user may (candidate execution).

[0094] In step S53, the prompt words corresponding to the candidate user behavior template are input into the large language model to obtain the rationality judgment result of the candidate user behavior template.

[0095] In step S54, candidate user behavior templates with reasonable rationality judgment results are selected, and user behavior templates are obtained based on reasonable candidate user behavior templates.

[0096] According to exemplary embodiments of this disclosure, by leveraging the deep semantic understanding capabilities of a large language model, the characteristics and attributes of events themselves are accurately considered, thereby evaluating the rationality between two events serving as conditions and events serving as execution operations. This reduces erroneous correlations introduced by data sparsity and noise. Consequently, it accurately determines whether a clear causal relationship exists between the events serving as conditions and the events serving as execution operations, providing reliable inferences about user habits and obtaining reasonable candidate user behavior templates, thus improving the accuracy of user behavior template determination.

[0097] In this embodiment of the disclosure, prompt words are designed for the obtained candidate user behavior templates. The prompt words are used to determine whether there is a causal relationship between the candidate conditions and the candidate execution in the candidate user behavior templates. Candidate user behavior templates with causal relationship are filled into a specified text form to obtain user behavior templates with causal relationship.

[0098] For example, the specified text format could be: When (candidate condition), the user may (candidate execution).

[0099] In one exemplary embodiment, a customary template with a causal relationship can be determined in the following manner:

[0100] Define relevant instructions, such as: We are performing user behavior sequence prediction to automatically control smart home devices, improving convenience and comfort. Now we need to determine the reasonableness of the predictions through step-by-step consideration. Define input specifications, such as: the input contains multiple lines, each line representing a prediction, and we need to determine its reasonableness. Define output specifications, such as: the output must adhere to all of the following specifications: determine the reasonableness of each prediction line by line. For each line, output the following four fields in JSON format: "Prediction", "Judgment Result", "Judgment Reason", and "Confidence Level" (ranging from 0 to 1). Encapsulate the results of all lines in JSON format.

[0101] In one example, the input is: When the temperature detected by the environmental sensor changes, the user may operate the air conditioner, altering its on / off state. The thought process is as follows: 1. A change in the temperature detected by the environmental sensor means that the temperature monitored by the sensor changes, for example, rising or falling. 2. Operating the air conditioner and altering its on / off state means controlling the on / off state of the air conditioning device, for example, turning it on or off. 3. Is the change in the temperature detected by the environmental sensor a possible reason for operating the air conditioner and altering its on / off state? If so, it means that a rise in temperature might prompt the user to turn on the air conditioner to lower the indoor temperature, while a drop in temperature might prompt the user to turn off the air conditioner to maintain a comfortable temperature. Based on the above thought process, the judgment result is deemed reasonable.

[0102] In another example, the input is: When the air quality index of the air purifier changes, the user might adjust the brightness of a light. The thought process is as follows: 1. A change in the air quality index of the air purifier means that the air quality index monitored by the air purifier changes, for example, worsening or improving. 2. Adjusting the brightness of a light means that the brightness of the light is changed, for example, brightening or dimming it. 3. Is a change in the air quality index monitored by the air purifier a possible reason for adjusting the brightness of the light? If not, it means that under normal circumstances, the user's adjustment of the light brightness has no direct relationship with air quality. Based on the above thought process, the judgment result is deemed unreasonable.

[0103] In the exemplary embodiments of this disclosure, obtaining a user behavior template based on a reasonable candidate user behavior template typically employs methods such as... Figure 6 As shown in the diagram. Figure 6 This is a flowchart illustrating a method for determining a user behavior template according to an exemplary embodiment. For example... Figure 6 As shown, it includes the following steps.

[0104] In step S61, for each user, the event data corresponding to the user is obtained separately.

[0105] In this embodiment of the disclosure, event data for each user is acquired, including user behavior data and environmental state data.

[0106] In step S62, the events that are used as conditions in the event data corresponding to the user are matched from the reasonable candidate user behavior templates.

[0107] In this embodiment of the disclosure, events in the user data that match the conditions in the template are searched based on the determined reasonable candidate user behavior template.

[0108] In step S63, among reasonable candidate user behavior templates, the execution operations corresponding to the events used as conditions are counted, and the probability of the corresponding execution operation occurring when the events used as conditions are met is determined.

[0109] In this embodiment of the disclosure, for each condition in a reasonable candidate user behavior template, the number of times the associated execution operation occurs is counted. The probability of the execution operation occurring when the condition is met is calculated.

[0110] In step S64, the execution operation with a probability greater than the threshold, and the corresponding conditions of the execution operation, are used as user behavior templates.

[0111] In this embodiment of the disclosure, a probability threshold is set. If the probability of a certain execution operation occurring when the condition is met is greater than the threshold, then the execution operation and its corresponding condition are included as part of the user behavior template.

[0112] The threshold can be understood as the value with the highest probability among multiple execution operations obtained by matching the same condition.

[0113] In this embodiment of the disclosure, for a target user, the events of the target user within a certain time period are analyzed and matched with a target habit template. If the match fails, it indicates that the user does not have such a habit, or the user's habit does not belong to the habit in the target habit template. If the match is successful, specific parameters are further filled in, such as what specific device the user is operating and what the specific value of the environmental state is.

[0114] In this embodiment of the disclosure, for each type of target habit template that matches the target user, data related to that type of habit template is retrieved. For example, for the habit template "unlock the door - turn on the light", all "unlock the door" and "turn on the light" event data of the user are retrieved.

[0115] In this embodiment of the disclosure, the matching user target habit is determined according to different matching methods for each type of target habit template.

[0116] In one example, when the "condition" is a recognized user activity event, there are Q devices that "execute". For each "execute" device, the probability of "execute" on the q-th device when the "condition" is met is calculated, where q = 1, 2, ..., Q. If the probability is greater than a given threshold, the user habit of "execution" on the q-th device is output; otherwise, the match fails.

[0117] In another example, when the "condition" is a user's device operation event, let there be P devices with the "condition" and Q devices with the "execution". Calculate the probability that the q-th device will "execute" when the p-th "condition" is met, where p = 1, 2, ..., P, q = 1, 2, ..., Q. If the probability is greater than a given threshold, output the user habit of "execution" by the p-th "condition" and the q-th device; otherwise, the match fails. Taking "unlocking a door lock - turning on a light" as an example, assuming the user has 2 door locks and 3 lights, calculate the probability of "turning on the j-th light" when "unlocking the i-th door lock".

[0118] In another example, when the "condition" is an environment state, there are P possible values ​​for the environment state, and Q devices in that environment state can be "executed". The probability of "executing" the q-th device when the p-th "condition" is met is calculated, where p = 1, 2, ..., P, and q = 1, 2, ..., Q. If the probability is greater than a given threshold, the user habit of "executing" the p-th "condition" - the q-th device is output; otherwise, the match fails.

[0119] According to an exemplary embodiment of this disclosure, user behavior data and environmental perception data are collected to construct an event dataset. Then, the correlation between pairs of events in the event dataset is calculated, and events with low correlation are filtered out, resulting in a series of candidate habit templates in the form of "condition-execution". These templates are then filtered using a large language model and carefully designed prompts to retain target habit templates with causal relationships. For each target habit template, the event data of each user is used to match the target habit template and analyze its probability to obtain the final habit. This disclosure utilizes the semantic understanding capabilities of a large language model to mine user habits, improving the accuracy of determining the execution device based on user habits.

[0120] In this embodiment of the disclosure, the correlation between two events can be understood as analyzing the collected event dataset to determine the relationship between events occurring within a certain time period. For example, the event dataset records that at a certain time point, a user is detected opening a door lock, and at a second time point some time later, the user is detected turning on the entrance light. In this case, the time period between the first and second time points is considered a certain time period, and the relationship between opening the door lock and turning on the entrance light during this time period is considered a correlation between the events.

[0121] In this embodiment, by analyzing user habit patterns from massive amounts of user data, we can achieve automated control of devices, providing users with a more convenient living experience and thus improving their quality of life. For example, based on user habits, lights and air conditioners can be automatically switched on and off under specific conditions, and speaker volume and mobile phone screen brightness can be adjusted. Such an intelligent device control system not only saves users time and energy but also creates a more comfortable and convenient living environment for them, thereby improving their quality of life.

[0122] In this exemplary embodiment, the correlation between pairs of events in user data is processed, and items with low correlation are filtered out to obtain a series of candidate templates in the form of condition-execution. These templates are then filtered using a large oracle model and prompt words to determine whether a causal relationship exists, thereby retaining reasonable habit templates. The conditional and execution parameters of the habit templates are learned for each user's data to obtain the final habit. By filtering out items with low correlation between pairs of events in the data for preprocessing, the length of the text input to the large language model can be significantly reduced, thereby reducing generation costs and increasing speed, making it more practical. The deep semantic understanding capabilities of the large oracle model are utilized to accurately consider the characteristics and attribute information of the events themselves, thereby evaluating the correlation between two events.

[0123] Based on the same concept, embodiments of this disclosure also provide an operation control device based on user behavior.

[0124] It is understood that the user behavior-based operation control device provided in this disclosure includes hardware structures and / or software modules corresponding to each function in order to achieve the above-mentioned functions. In conjunction with the units and algorithm steps of the various examples disclosed in this disclosure, this disclosure can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed by hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the technical solutions of this disclosure.

[0125] Figure 7 This is a block diagram illustrating an operation control device based on user behavior, according to an exemplary embodiment. (Refer to...) Figure 7 The device 100 includes a determining unit 101, a processing unit 102, and an execution unit 103.

[0126] The determining unit 101 is used to determine the current event triggered by the user's behavior.

[0127] The processing unit 102 is used to respond to the existence of a target condition in the user behavior template that matches the current event, and to determine the target execution operation corresponding to the target condition according to the user behavior template; wherein, the user behavior template includes the correspondence between conditions and execution operations, and the user behavior template is obtained by filtering candidate user behavior templates based on a large language model, the candidate behavior template includes candidate correspondence between conditions and execution operations obtained by mining based on the event dataset, the event dataset includes events associated with user behavior, and the execution operation is performed when the corresponding condition is met.

[0128] The execution unit 103 is used to perform the target execution operation.

[0129] In one implementation, the processing unit 102 obtains candidate user behavior templates based on an event dataset by mining the event dataset as follows: collecting multiple user behavior-related events to obtain an event dataset; determining the event data of the i-th user in the event dataset, and classifying the event data of the i-th user to obtain M. i Class conditions and N i Class performs operations; for M i Class conditions and N i The operations of each class are combined in pairs to form M. i *N i Class composition; determine M i *N i The correlation metric between the condition and the action to be performed in each combination is determined; combinations with correlation metrics greater than the correlation metric threshold are used as candidate behavior templates for the i-th user; the above process is repeated until candidate behavior templates for multiple users are determined.

[0130] In one embodiment, the processing unit 102 determines the correlation metric between conditions and execution operations in each combination as follows: count the total number of execution operations and the total number of conditions in each combination; determine the ratio between the total number of execution operations and the total number of conditions; multiply the ratio by a first coefficient to obtain a first value, and multiply the total number of execution operations by a second coefficient to obtain a second value; and determine the sum of the first value and the second value as the correlation metric between conditions and execution operations.

[0131] In one embodiment, the processing unit 102 uses a large language model to filter candidate user behavior templates to obtain user behavior templates in the following manner: setting prompt words to prompt the large language model to judge the rationality of the candidate user behavior templates; filling the prompt words with the conditions and execution operations in the candidate user behavior templates according to the text format of the prompt words to obtain the prompt words for the corresponding candidate user behavior templates; inputting the prompt words for the corresponding candidate user behavior templates into the large language model to obtain the rationality judgment result of the candidate user behavior templates; filtering out candidate user behavior templates with reasonable rationality judgment results, and obtaining user behavior templates based on reasonable candidate user behavior templates.

[0132] In one embodiment, the processing unit 102 obtains a user behavior template based on a reasonable candidate user behavior template in the following manner: for each user, the event data corresponding to the user is obtained; the event used as a condition in the event data corresponding to the user is matched in the reasonable candidate user behavior template; in the reasonable candidate user behavior template, the execution operation corresponding to the event used as the condition is counted, and the probability of the corresponding execution operation occurring when the event used as the condition is satisfied is determined; the execution operation with a probability greater than a threshold, and the condition corresponding to the execution operation, are used as the user behavior template.

[0133] In one embodiment, the processing unit 102 collects multiple user behavior-related events in the following manner to obtain an event dataset: user behavior data is collected within a preset time period, and environmental state data is collected when the user behavior data occurs; wherein, the user behavior data includes user operation events on objects collected by sensors, and user activity events obtained based on perception algorithms; the environmental state data includes environmental states collected by sensors, and environmental state change events identified based on perception algorithms; and timestamps corresponding to the collection time are added to the user behavior data and environmental state data respectively to generate the event dataset.

[0134] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0135] Figure 8 This is a block diagram illustrating an apparatus 200 for user behavior-based operation control according to an exemplary embodiment. For example, apparatus 200 may be a mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, medical device, fitness equipment, personal digital assistant, etc.

[0136] Reference Figure 8The device 200 may include one or more of the following components: processing component 202, memory 204, power component 206, multimedia component 208, audio component 210, input / output (I / O) interface 212, sensor component 214, and communication component 216.

[0137] Processing component 202 typically controls the overall operation of device 200, such as operations associated with display, telephone calls, data communication, camera operation, and recording. Processing component 202 may include one or more processors 220 to execute instructions to perform all or part of the steps of the methods described above. Furthermore, processing component 202 may include one or more modules to facilitate interaction between processing component 202 and other components. For example, processing component 202 may include a multimedia module to facilitate interaction between multimedia component 208 and processing component 202.

[0138] Memory 204 is configured to store various types of data to support the operation of device 200. Examples of such data include instructions for any application or method operating on device 200, contact data, phonebook data, messages, pictures, videos, etc. Memory 204 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0139] The power supply component 206 provides power to the various components of the device 200. The power supply component 206 may include a power management system, one or more power sources, and other components associated with generating, managing, and distributing power to the device 200.

[0140] Multimedia component 208 includes a screen that provides an output interface between the device 200 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touchscreen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors may sense not only the boundaries of the touch or swipe action but also the duration and pressure associated with the touch or swipe operation. In some embodiments, multimedia component 208 includes a front-facing camera and / or a rear-facing camera. When the device 200 is in an operating mode, such as a shooting mode or a video mode, the front-facing camera and / or the rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0141] Audio component 210 is configured to output and / or input audio signals. For example, audio component 210 includes a microphone (MIC) configured to receive external audio signals when device 200 is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 204 or transmitted via communication component 216. In some embodiments, audio component 210 also includes a speaker for outputting audio signals.

[0142] I / O interface 212 provides an interface between processing component 202 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, power buttons, and lock buttons.

[0143] Sensor assembly 214 includes one or more sensors for providing status assessments of various aspects of device 200. For example, sensor assembly 214 may detect the on / off state of device 200, the relative positioning of components such as the display and keypad of device 200, changes in the position of device 200 or a component of device 200, the presence or absence of user contact with device 200, the orientation or acceleration / deceleration of device 200, and temperature changes of device 200. Sensor assembly 214 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 214 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, sensor assembly 214 may also include an accelerometer, a gyroscope, a magnetometer, a pressure sensor, or a temperature sensor.

[0144] Communication component 216 is configured to facilitate wired or wireless communication between device 200 and other devices. Device 200 can access wireless networks based on communication standards, such as WiFi, 2G, or 3G, or combinations thereof. In one exemplary embodiment, communication component 216 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 216 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0145] In an exemplary embodiment, the apparatus 200 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the methods described above.

[0146] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 204 including instructions, which can be executed by a processor 220 of the device 200 to perform the above-described method. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0147] Figure 9 This is a block diagram illustrating an apparatus 300 for user behavior-based operation control according to an exemplary embodiment. For example, apparatus 300 may be provided as a server. (Refer to...) Figure 9 The device 300 includes a processing component 322, which further includes one or more processors, and memory resources represented by memory 332 for storing instructions, such as application programs, that can be executed by the processing component 322. The application programs stored in memory 332 may include one or more modules, each corresponding to a set of instructions. Furthermore, the processing component 322 is configured to execute instructions to perform the methods described above.

[0148] Device 300 may also include a power supply component 326 configured to perform power management of device 300, a wired or wireless network interface 350 configured to connect device 300 to a network, and an input / output (I / O) interface 358. Device 300 may operate on an operating system stored in memory 332, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, or similar.

[0149] It is understood that in this disclosure, "multiple" refers to two or more, and other quantifiers are similar. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. The singular forms "a," "the," and "the" are also intended to include the plural forms unless the context clearly indicates otherwise.

[0150] It is further understood that the terms "first," "second," etc., are used to describe various types of information, but this information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another, and do not indicate a specific order or degree of importance. In fact, the expressions "first," "second," etc., are completely interchangeable. For example, without departing from the scope of this disclosure, first information can also be referred to as second information, and similarly, second information can also be referred to as first information.

[0151] It is further understood that the terms “center,” “longitudinal,” “lateral,” “front,” “rear,” “up,” “down,” “left,” “right,” “vertical,” “horizontal,” “top,” “bottom,” “inner,” and “outer,” etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this embodiment and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation.

[0152] It can be further understood that, unless otherwise specified, "connection" includes both direct connections where no other components exist between the two parties and indirect connections where other components exist between them.

[0153] It is further understood that although operations are described in a specific order in the accompanying drawings in the embodiments of this disclosure, this should not be construed as requiring these operations to be performed in the specific order or serial order shown, or requiring all of the shown operations to be performed to obtain the desired result. In certain environments, multitasking and parallel processing may be advantageous.

[0154] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein.

[0155] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

Claims

1. A user behavior-based operation control method, characterized in that, The method includes: Determine the current event triggered by the user's action; In response to the existence of a target condition in the user behavior template that matches the current event, the target execution operation corresponding to the target condition is determined according to the user behavior template; The user behavior template includes a correspondence between conditions and execution operations. The user behavior template is obtained by filtering candidate user behavior templates based on a large language model. The candidate behavior template includes candidate correspondences between conditions and execution operations obtained by mining event datasets. The event dataset includes events associated with user behavior. The execution operation is performed when the corresponding conditions are met. Perform the target execution operation.

2. The method according to claim 1, characterized in that, The candidate user behavior templates are obtained by mining from the event dataset in the following way: Collect multiple user behavior-related events to obtain an event dataset; Identify the event data of the i-th user in the event dataset, and classify the event data of the i-th user to obtain M. i Class conditions and N i The class performs the operation; For the M i Class conditions and N i The operations are combined in pairs to form M. i *N i Class composition; Determine the M i *N i The correlation measure between conditions and execution operations in each category of the combination; Combinations whose relevance metric values ​​are greater than the relevance metric threshold are used as candidate behavior templates for the i-th user. Repeat the above process until candidate behavior templates for the multiple users are determined.

3. The method according to claim 2, characterized in that, The correlation metric between the conditions and the execution operation in each combination is determined in the following manner: Count the total number of execution operations and the total number of conditions in each combination; Determine the ratio of the total number of operations performed to the total number of conditions met; Multiply the ratio by the first coefficient to obtain the first value, and multiply the total number of operations by the second coefficient to obtain the second value; The sum of the first value and the second value is determined as a correlation metric between the condition and the execution operation.

4. The method according to any one of claims 1 to 3, characterized in that, The user behavior templates are obtained by filtering candidate user behavior templates based on a large language model in the following way: Set prompt words, which are used to prompt the large language model to judge the rationality of candidate user behavior templates; The conditions and operations in the candidate user behavior template are filled into the prompt words according to the text format of the prompt words to obtain the prompt words corresponding to the candidate user behavior template; The prompt words corresponding to the candidate user behavior template are input into the large language model to obtain the rationality judgment result of the candidate user behavior template; Select candidate user behavior templates that are deemed reasonable based on the reasonableness judgment, and obtain user behavior templates based on these reasonable candidate user behavior templates.

5. The method according to claim 4, characterized in that, The process of obtaining user behavior templates based on reasonable candidate user behavior templates includes: For each user, obtain the event data corresponding to that user separately; Match the events that are used as conditions in the event data corresponding to the user from a reasonable candidate user behavior template; In a reasonable candidate user behavior template, the execution operations corresponding to the events used as conditions are statistically analyzed, and the probability of the corresponding execution operation occurring when the events used as conditions are satisfied is determined. The operation with a probability greater than the threshold, along with the corresponding conditions, is used as a user behavior template.

6. The method according to claim 2, characterized in that, The process involves collecting multiple user behavior-related events to obtain an event dataset, including: Collect user behavior data within a preset time period, and collect environmental state data when the user behavior data occurs; The user behavior data includes user operation events on objects collected by sensors and user activity events obtained by perception algorithms; the environmental state data includes environmental state collected by sensors and environmental state change events identified by perception algorithms. Add the timestamps corresponding to the time of collection to the user behavior data and the environmental state data respectively to generate an event dataset.

7. An operation control device based on user behavior, characterized in that, include: The determination unit is used to determine the current event triggered by the user's behavior; The processing unit is configured to, in response to the existence of a target condition in the user behavior template that matches the current event, determine the target execution operation corresponding to the target condition according to the user behavior template; The user behavior template includes a correspondence between conditions and execution operations. The user behavior template is obtained by filtering candidate user behavior templates based on a large language model. The candidate behavior template includes candidate correspondences between conditions and execution operations obtained by mining event datasets. The event dataset includes events associated with user behavior. The execution operation is performed when the corresponding conditions are met. An execution unit is used to perform the target execution operation.

8. The apparatus according to claim 7, characterized in that, The processing unit obtains candidate user behavior templates based on the event dataset using the following method: Collect multiple user behavior-related events to obtain an event dataset; Identify the event data of the i-th user in the event dataset, and classify the event data of the i-th user to obtain M. i Class conditions and N i The class performs the operation; For the M i Class conditions and N i The operations of each class are combined in pairs to form M. i *N i Class composition; Determine the M i *N i The correlation measure between conditions and execution operations in each category of the combination; Combinations whose relevance metric values ​​are greater than the relevance metric threshold are used as candidate behavior templates for the i-th user. Repeat the above process until candidate behavior templates for the multiple users are determined.

9. The apparatus according to claim 8, characterized in that, The processing unit determines the correlation metric between conditions and execution operations in each combination using the following method: Count the total number of execution operations and the total number of conditions in each combination; Determine the ratio of the total number of operations performed to the total number of conditions met; Multiply the ratio by the first coefficient to obtain the first value, and multiply the total number of operations by the second coefficient to obtain the second value; The sum of the first value and the second value is determined as a correlation metric between the condition and the execution operation.

10. The apparatus according to any one of claims 7 to 9, characterized in that, The processing unit uses the following method to filter candidate user behavior templates based on a large language model to obtain user behavior templates: Set prompt words, which are used to prompt the large language model to judge the rationality of candidate user behavior templates; The conditions and operations in the candidate user behavior template are filled into the prompt words according to the text format of the prompt words to obtain the prompt words corresponding to the candidate user behavior template; The prompt words corresponding to the candidate user behavior template are input into the large language model to obtain the rationality judgment result of the candidate user behavior template; Select candidate user behavior templates that are deemed reasonable based on the reasonableness judgment, and obtain user behavior templates based on these reasonable candidate user behavior templates.

11. The apparatus according to claim 10, characterized in that, The processing unit obtains user behavior templates based on reasonable candidate user behavior templates in the following manner: For each user, obtain the event data corresponding to that user separately; Match the events that are used as conditions in the event data corresponding to the user from a reasonable candidate user behavior template; In a reasonable candidate user behavior template, the execution operations corresponding to the events used as conditions are statistically analyzed, and the probability of the corresponding execution operation occurring when the events used as conditions are satisfied is determined. The operation with a probability greater than the threshold, along with the corresponding conditions, is used as a user behavior template.

12. The apparatus according to claim 8, characterized in that, The processing unit collects multiple user behavior-related events in the following manner to obtain an event dataset: Collect user behavior data within a preset time period, and collect environmental state data when the user behavior data occurs; The user behavior data includes user operation events on objects collected by sensors and user activity events obtained by perception algorithms; the environmental state data includes environmental state collected by sensors and environmental state change events identified by perception algorithms. Add the timestamps corresponding to the time of collection to the user behavior data and the environmental state data respectively to generate an event dataset.

13. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to perform the method according to any one of claims 1-6.

14. A storage medium, characterized in that, The storage medium stores instructions that, when executed by a processor, enable the execution of the method described in any one of claims 1-6.

Images