GPT-based equipment control method and device, storage medium and electronic device

Abstract

The invention discloses an equipment control method and device based on GPT, a storage medium and an electronic device, and relates to the technical field of smart home / smart home, and the equipment control method comprises the steps: obtaining first equipment indication information triggered by a user account at a first moment; performing feature extraction on the first equipment indication information to obtain a user portrait feature associated with the first equipment indication information and an equipment attribute feature associated with the first equipment indication information; determining historical operation of the first intelligent device on second device indication information of the user account at a second moment and historical operation information corresponding to the historical operation from the historical interaction information; at least two intelligent devices including the first intelligent device and the second intelligent device are obtained, the first intelligent device is controlled to execute the first operation, and the second intelligent device is controlled to execute the second operation. The technical problem of low accuracy of equipment control is solved.

Description

Technical Field

[0001] The present application relates to the technical field of smart home / smart family. Specifically, it relates to a device control method, device, storage medium and electronic device based on GPT. Background Art

[0002] In the process of human-computer interaction, users often use voice commands to control corresponding devices. However, the method of simply inputting voice commands to output the control operation of a certain device is too simple, and there are defects in that it cannot fully meet the user's device control requirements, and it is difficult to achieve the accuracy of device control, resulting in the technical problem of low accuracy of device control in related technologies.

[0003] In view of the above technical problems, no effective solution has been proposed yet. Summary of the Invention

[0004] Embodiments of the present application provide a device control method, device, storage medium and electronic device based on GPT to at least solve the technical problem of low accuracy of device control.

[0005] According to one aspect of the embodiments of the present application, a device control method based on GPT is provided, including: obtaining first device indication information triggered by a user account at a first moment, where the first device indication information is used to indicate controlling a first smart device to perform a first operation; extracting features from the first device indication information to obtain a user portrait feature associated with the first device indication information and a device attribute feature associated with the first device indication information, where the user portrait feature is used to represent the user portrait when the user account triggers the first device indication information, and the device attribute feature is used to represent the first device attribute of the first smart device when the user account triggers the first device indication information; when historical interaction information at a second moment is obtained and the historical portrait feature corresponding to the historical interaction information conforms to the user portrait feature, determining, from the historical interaction information, a historical operation of the first smart device on a second device indication information of the user account at the second moment and historical operation information corresponding to the historical operation, where the historical interaction information is information of interaction between the user account and the first smart device, and the second moment is a historical moment of the first moment; through a GPT model, performing prediction processing on the device attribute feature and the historical operation information to obtain at least two smart devices including the first smart device and a second smart device, and controlling the first smart device to perform the first operation and controlling the second smart device to perform a second operation, where the second operation is associated with the first operation, and the second device attribute of the second smart device is associated with the first device attribute.

[0006] According to another aspect of the embodiments of the present application, there is also provided a GPT-based device control device, including: an acquisition unit, configured to acquire first device indication information triggered by a user account at a first moment, where the first device indication information is used to indicate controlling a first intelligent device to perform a first operation; an extraction unit, configured to perform feature extraction on the first device indication information to obtain a user portrait feature associated with the first device indication information and a device attribute feature associated with the first device indication information, where the user portrait feature is used to represent the user portrait when the user account triggers the first device indication information, and the device attribute feature is used to represent the first device attribute of the first intelligent device when the user account triggers the first device indication information; a determination unit, configured to, when historical interaction information at a second moment is acquired and the historical portrait feature corresponding to the historical interaction information conforms to the user portrait feature, determine, from the historical interaction information, a historical operation of the first intelligent device on a second device indication information of the user account at the second moment and historical operation information corresponding to the historical operation, where the historical interaction information is information of an interaction between the user account and the first intelligent device, and the second moment is a historical moment of the first moment; a control unit, configured to perform prediction processing on the device attribute feature and the historical operation information through a GPT model to obtain at least two intelligent devices including the first intelligent device and a second intelligent device, and control the first intelligent device to perform the first operation and control the second intelligent device to perform a second operation, where the second operation is correlated with the first operation, and a second device attribute of the second intelligent device is correlated with the first device attribute. According to yet another aspect of the embodiments of the present application, there is provided a computer program product or a computer program, where the computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the above device control method.

[0007] According to yet another aspect of the embodiments of the present application, there is also provided an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, where the above processor executes the above GPT-based device control method through the computer program.

[0008] In the embodiments of the present application, by using the above-mentioned GPT-based device control method, based on the first intelligent device that responds to the first device instruction information to determine the execution of the first operation, through processing such as feature extraction and portrait feature verification of the first device instruction information, and in the case where the portrait feature verification is passed, the second intelligent device that executes the second operation associated with the first operation is determined by combining historical interaction information and device attribute features. Furthermore, the purpose of responding to multiple intelligent devices and multiple associated operations of a user request is achieved through more comprehensive feature mining and verification, thereby achieving the technical effect of improving the accuracy of device control and solving the technical problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The accompanying drawings herein are incorporated into the specification and form a part of the specification, showing embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application.

[0010] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the accompanying drawings required for use in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0011] Figure 1 is a schematic diagram of the hardware environment of a GPT-based device control method according to an embodiment of the present application;

[0012] Figure 2 is a schematic diagram of the flow of an optional GPT-based device control method according to an embodiment of the present application;

[0013] Figure 3 is a schematic diagram of an optional GPT-based device control method according to an embodiment of the present application;

[0014] Figure 4 is a schematic diagram of another optional GPT-based device control method according to an embodiment of the present application;

[0015] Figure 5 is a schematic diagram of an optional information processing device according to an embodiment of the present invention;

[0016] Figure 6 is a schematic diagram of the structure of an optional electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0018] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0019] According to one aspect of the embodiment of the present application, a device control method based on GPT is provided. The device control method based on GPT is widely used in smart home (Smart Home), smart home, smart home device ecology, smart residential (IntelligenceHouse) ecology and other whole-house intelligent digital control application scenarios. Optionally, in this embodiment, the above-mentioned smart home device interaction method can be applied to Figure 1 In the hardware environment composed of the terminal device 102 and the server 104 shown in FIG. Figure 1 As shown, the server 104 is connected to the terminal device 102 via a network, and can be used to provide services (such as application services, etc.) for the terminal or a client installed on the terminal. A database can be set on the server or independently of the server to provide data storage services for the server 104. Cloud computing and / or edge computing services can be configured on the server or independently of the server to provide data computing services for the server 104.

[0020] The above network may include, but is not limited to, at least one of the following: a wired network, a wireless network. The above wired network may include, but is not limited to, at least one of the following: a wide area network, a metropolitan area network, a local area network. The above wireless network may include, but is not limited to, at least one of the following: WIFI (Wireless Fidelity), Bluetooth. The terminal device 102 is not limited to a PC, mobile phone, tablet computer, smart air conditioner, smart range hood, smart refrigerator, smart oven, smart stove, smart washing machine, smart water heater, smart washing equipment, smart dishwasher, smart projection device, smart TV, smart clothes hanger, smart curtain, smart audio and video, smart socket, smart speaker, smart sound box, smart fresh air device, smart kitchen and bathroom equipment, smart bathroom equipment, smart floor sweeping robot, smart window cleaning robot, smart mopping robot, smart air purification device, smart steam box, smart microwave oven, smart kitchen water heater, smart purifier, smart water dispenser, smart door lock, etc.

[0021] Optionally, as an alternative implementation, as Figure 2 shown, the GPT-based device control method includes:

[0022] S202, obtaining first device indication information triggered by a user account at a first moment, where the first device indication information is used to indicate controlling a first smart device to perform a first operation;

[0023] S204, extracting features from the first device indication information to obtain user portrait features associated with the first device indication information and device attribute features associated with the first device indication information, where the user portrait features are used to represent the user portrait when the user account triggers the first device indication information, and the device attribute features are used to represent the first device attributes of the first smart device when the user account triggers the first device indication information;

[0024] S206, when historical interaction information at a second moment is obtained and the historical portrait features corresponding to the historical interaction information match the user portrait features, determining, from the historical interaction information, the historical operations of the first smart device on a second device indication information of the user account at the second moment and the historical operation information corresponding to the historical operations, where the historical interaction information is information of interactions between the user account and the first smart device, and the second moment is a historical moment of the first moment;

[0025] S208, through a GPT model, performing prediction processing on the device attribute features and the historical operation information to obtain at least two smart devices including the first smart device and a second smart device, and controlling the first smart device to perform the first operation and controlling the second smart device to perform a second operation, where the second operation is correlated with the first operation, and the second device attributes of the second smart device are correlated with the first device attributes.

[0026] Optionally, in this embodiment, the device control method may be, but is not limited to, applied to the human-computer interaction scenario of intelligent devices. In this scenario, the user may, but is not limited to, input relevant control voice commands to the intelligent device to control the intelligent device to perform control operations associated with the control voice commands.

[0027] It should be noted that the method of simply outputting the control operation of a certain device by inputting a voice command is too simple, and there is a defect that the user's device control requirements cannot be fully realized, and it is difficult to achieve the accuracy of device control, which further leads to the technical problem of low accuracy of device control in related technologies.

[0028] In response to the above problems, using the above-mentioned GPT-based device control method, when the control voice command triggered by the user account is obtained, not only the first operation directly responding to the control voice command and the first intelligent device for executing the first operation are determined, but also through processing such as feature extraction and portrait feature verification of the control voice command, and when the portrait feature verification passes, the second intelligent device for executing the second operation associated with the first operation is determined by combining historical interaction information and device attribute features. Furthermore, the purpose of responding to multiple intelligent devices and multiple associated operations of a user request is achieved through more comprehensive feature mining and verification, thus achieving the technical effect of improving the accuracy of device control and solving the technical problem.

[0029] Optionally, in this embodiment, the first device indication information may be, but is not limited to, used to indicate the first intelligent device in at least one intelligent device to execute the first operation, where the at least one intelligent device may be, but is not limited to, all intelligent devices within a certain range when the first device indication information is triggered.

[0030] Optionally, in this embodiment, the first device indication information may be, but is not limited to, the voice information carrying the device control intention triggered by the user account at the first moment, where the device control intention may be, but is not limited to, used to indicate the first intelligent device to execute the first operation.

[0031] Optionally, in this embodiment, based on a pre-trained GPT model, feature extraction is performed on the first device indication information to obtain the user portrait features associated with the first device indication information, where the user portrait features are the user portraits when the user account triggers the first device indication information at the first moment, and may include, but are not limited to, user attribute features and user behavior features. The user attribute features may be, but are not limited to, the static features of the user associated with the user account at the first moment, such as the user's age information, interest information, location information, etc. The user behavior features may be, but are not limited to, the dynamic features of the user associated with the user account at the first moment, such as the user's habit of performing operation A, and the user's habit of performing operation B after performing operation A.

[0032] Optionally, in this embodiment, based on a pre-trained model, feature extraction is performed on the first device indication information to obtain device attribute features associated with the first device indication information, where the device attribute features may, but are not limited to, be used to represent the first device attributes of the first intelligent device when the user account triggers the first device indication information.

[0033] Optionally, in this embodiment, historical interaction information at a second moment is obtained, where the second moment may, but is not limited to, be a historical moment of the first moment, may, but is not limited to, be the previous moment of the first moment, and the historical interaction information may, but is not limited to, be information on the interaction between the user account and the first interaction device, and may, but is not limited to, include historical portrait feature information and historical operation information, where the historical operation information may, but is not limited to, include, at the second moment, a second device indication information triggered by the user account and a third operation performed by the first intelligent device in response to the second device indication information.

[0034] Optionally, in this embodiment, historical portrait features corresponding to the historical interaction information are obtained, and it is determined whether the historical portrait features at the second moment conform to the user portrait features at the first moment.

[0035] It should be noted that in the case where the historical portrait features conform to the user portrait features, it may, but is not limited to, indicate that the habit features of the user associated with the user account have not changed significantly between the second moment and the first moment, and the second intelligent device for performing the second operation may, but is not limited to, be determined from at least one intelligent device based on the device attribute features and the historical interaction information.

[0036] It should be noted that in the case where the historical portrait features do not conform to the user portrait features, it may, but is not limited to, indicate that the habit features of the user associated with the user account have changed significantly between the second moment and the first moment, and the second intelligent device for performing the second operation may, but is not limited to, be determined from at least one intelligent device based on the device attribute features and the user portrait features.

[0037] Optionally, in this embodiment, the device attribute features may, but are not limited to, be used to indicate the first device attributes of the first intelligent device for performing the first operation, and may, but is not limited to, be used to determine the second intelligent device having second device attributes that are mutually associated with the first device attributes.

[0038] For example, the first intelligent device may, but is not limited to, be an intelligent air conditioning device, and the first device attribute may, but is not limited to, be used to indicate the power switch state of the intelligent air conditioning device; the second intelligent device may, but is not limited to, be an intelligent window device, and the second device attribute may, but is not limited to, be used to indicate the opening state of the intelligent window device.

[0039] Optionally, in this embodiment, the historical interaction information may include, but is not limited to, historical portrait features and historical operation information. Among them, the historical operation information may include, but is not limited to, indicating a historical operation, which is a response operation of the first intelligent device to the second device indication information of the user account at the second moment.

[0040] It should be noted that predictive processing is performed on the device attribute features and historical operation information to determine a second intelligent device for performing a second operation from at least one intelligent device.

[0041] It can be understood that after determining the first intelligent device and the first operation, further prediction is performed based on the device attribute features associated with the first intelligent device and the historical response operation information of the first intelligent device, to obtain a second intelligent device associated with the first intelligent device and a second operation associated with the first operation. Thus, through more comprehensive feature mining and verification, the purpose of responding to multiple intelligent devices and multiple associated operations for a user request is achieved, thereby realizing the technical effect of improving the accuracy of device control.

[0042] Further, for example, when the first device attribute indicates that the intelligent air conditioner device is in the power-on state, that is, the first intelligent device is an intelligent air conditioner device, and the first operation is to turn on the power of the intelligent air conditioner device, a second intelligent window device may be determined from at least one intelligent device, and the opening state of the second intelligent window device may be controlled and adjusted to the fully closed state.

[0043] It should be noted that using the above device control method, after determining a first intelligent device from at least one intelligent device to perform a first operation in response to the first device indication information triggered by the user account at the first moment, a second intelligent device associated with the first intelligent device is further determined to perform a second operation associated with the first operation. Among them, the first operation is a direct response to the first device indication information, and the second operation is an indirect response to the first device indication information, and is based on the comparison result of the historical portrait features corresponding to the historical interaction information and the user portrait features at the current moment, and is determined from the device attribute features and historical interaction information. Therefore, it reflects the personalized habit features of the user in the recent period. Through more comprehensive feature mining and verification, the purpose of responding to multiple intelligent devices and multiple associated operations for a user request is achieved, thereby realizing the technical effect of improving the accuracy of device control.

[0044] Further, for example, as Figure 3 shown, an optional device control method specifically includes the following steps:

[0045] Step S302: Obtain the first device indication information triggered by the user account at the first moment;

[0046] Step S304: Extract features from the first device indication information to obtain the first intelligent device and the first operation indicated by the first device indication information;

[0047] Step S306: Obtain the device attribute features associated with the first intelligent device;

[0048] Step S308: Extract features from the first device indication information to obtain the user portrait features associated with the first device indication information;

[0049] Step S110: Obtain the historical interaction information at the second moment, where the second moment is before the first moment;

[0050] Step S112: Obtain the historical portrait features corresponding to the historical interaction information;

[0051] Step S114: When the historical portrait features match the user portrait features, determine the second intelligent device and the second operation based on the device attribute features and the historical interaction information.

[0052] Through the embodiments provided in the present application, when the control voice command triggered by the user account is obtained, not only the first operation directly responding to the control voice command and the first intelligent device for executing the first operation are determined, but also through the processing such as feature extraction and portrait feature verification of the control voice command, and when the portrait feature verification is passed, the second intelligent device for executing the second operation associated with the first operation is determined in combination with the historical interaction information and the device attribute features, thereby achieving the purpose of responding to multiple intelligent devices and multiple associated operations of a user request through more comprehensive feature mining and verification, thus achieving the technical effect of improving the accuracy of device control and solving the technical problems.

[0053] As an optional solution, through the GPT model, perform prediction processing on the device attribute features and the historical operation information to obtain at least two intelligent devices including the first intelligent device and the second intelligent device, including:

[0054] S1: Input the device attribute features and the historical operation information into the GPT model to perform the prediction processing to obtain a target operation set, where the target operation set includes at least one response operation, and at least one response operation is mutually associated with the first operation;

[0055] S2: Obtain the first response operation with the highest association confidence degree with the first operation from at least one response operation, where the association confidence degree is used to indicate the association degree between at least one response operation and the first operation;

[0056] S3. Obtain the third device attribute associated with the first response device, where the first response device is used to perform a first response operation;

[0057] S4. When the third device attribute meets the device confidence condition, determine the first response operation as the second operation and the first response device as the second intelligent device, where at least two intelligent devices include the second intelligent device.

[0058] Optionally, in this embodiment, predictive processing is performed on the device attribute features and historical operation information, and the target operation set including at least one response operation can be obtained, but is not limited to this, where at least one response operation is associated with the first operation.

[0059] Optionally, in this embodiment, after determining at least one response operation included in the target operation set, obtain the association confidence between at least one response operation and the first operation, and determine the first response operation with the highest association confidence therefrom, where the association confidence is used to indicate the degree of association between the response operation and the first operation.

[0060] Optionally, in this embodiment, after determining the first response operation, obtain the third device attribute associated with the first response device corresponding to the first response operation, and determine whether the third device attribute meets the confidence condition.

[0061] Optionally, in this embodiment, the confidence condition can be used to indicate, but is not limited to, whether the response device is in an available state.

[0062] Optionally, when the third device associated with the first response device meets the device confidence condition, determine the first response operation as the second operation and the first response device as the second intelligent device.

[0063] It should be noted that when the third device associated with the first response device does not meet the device confidence condition, determine the second response operation with the highest association confidence from the target operation set that does not include the above first response operation; obtain the fourth device attribute associated with the second response device, where the second response device is used to perform the above second response operation; determine whether the fourth device attribute meets the device confidence condition; when the fourth device attribute meets the device confidence condition, determine the second response operation as the above second operation and the second response device as the above second intelligent device.

[0064] It should be noted that when the fourth device attribute does not meet the device confidence condition, determine the third response operation with the highest association confidence from the target operation set that does not include the first response operation and the second response operation, and repeat the above operations until the second operation and the second intelligent device are determined, or all response operations in the target operation set are traversed.

[0065] Through the embodiments provided in this application, device attribute features and historical operation information are input into a GPT model for the prediction process to obtain a set of target operations, where the set of target operations includes at least one response operation, and at least one response operation is correlated with a first operation; a first response operation with the highest correlation confidence with the first operation is obtained from at least one response operation, where the correlation confidence is used to indicate the degree of correlation between at least one response operation and the first operation; a third device attribute associated with the first response device is obtained, where the first response device is used to execute the first response operation; when the third device attribute meets the device confidence condition, the first response operation is determined as a second operation, and the first response device is determined as a second intelligent device, where at least two intelligent devices include the second intelligent device. The response operation with the highest degree of correlation whose device attribute meets the device confidence condition is determined from the multiple predicted response operations, and then the optimal second operation and second intelligent device are determined based on the user's habits (based on user behavior characteristics), thereby achieving the technical effect of improving the accuracy of device control.

[0066] As an alternative solution, after feature extraction is performed on the first device indication information to obtain user portrait features associated with the first device indication information and device attribute features associated with the first device indication information, the method further includes:

[0067] S1. When historical interaction information is obtained and the historical portrait features corresponding to the historical interaction information do not match the user portrait features, a third intelligent device is determined from at least one intelligent device based on the device attribute features and the user portrait features, and the first intelligent device is controlled to execute a first operation, and the third intelligent device is controlled to execute a third operation, where the third device attribute of the third intelligent device is correlated with the first device attribute, the first operation is correlated with the third operation, the second moment is the historical moment of the first moment, and the historical interaction information is the information of the user account interacting with the first intelligent device.

[0068] It should be noted that when the historical portrait features corresponding to the historical interaction information do not match the user portrait features, it indicates that the user habit features associated with the user account have changed between the second moment and the first moment. Then, based on the user's current habit features, that is, a third intelligent device is determined from at least one intelligent device based on the device attribute features and the user portrait features.

[0069] It should be noted that, after determining the first smart device from at least one smart device to perform the first operation in response to the first device instruction information triggered at the first moment of the user account, the third smart device associated with the first smart device is further determined to perform the third operation associated with the first operation. The first operation is a direct response to the first device instruction information, and the third operation is an indirect response to the first device instruction information. It is based on the comparison result of the historical portrait features corresponding to the historical interaction information and the user portrait features at the current moment, and is determined from the device attribute features and the user portrait features. Therefore, it reflects the personalized habit features of the user at the current time, achieving the purpose of responding to multiple smart devices and multiple associated operations for a user request through more comprehensive feature mining and verification, thereby achieving the technical effect of improving the accuracy of device control.

[0070] Through the embodiments provided in the present application, in the case where the historical interaction information is obtained and the historical portrait features corresponding to the historical interaction information do not conform to the user portrait features, the third smart device is determined from at least one smart device based on the device attribute features and the user portrait features, and the first smart device is controlled to perform the first operation, and the third smart device is controlled to perform the third operation. The third device attribute of the third smart device is associated with the first device attribute, the first operation is associated with the third operation, the second moment is the historical moment of the first moment, and the historical interaction information is the information of the interaction between the user account and the first smart device.

[0071] As an optional solution, determining the third smart device from at least one smart device based on the device attribute features and the user portrait features includes:

[0072] S1, determining the user attribute features and the user behavior features from the user portrait features. The user attribute features are used to indicate the static features of the user associated with the user account at the first moment, and the user behavior features are used to indicate the dynamic features of the user associated with the user account at the first moment.

[0073] S2, performing a prediction process on the device attribute features and the user behavior features, and determining the third smart device for performing the third operation from at least one smart device.

[0074] Optionally, in this embodiment, the user portrait features may include, but are not limited to, the user attribute features and the user behavior features. The user attribute features may include, but are not limited to, the static features of the user associated with the user account at the first moment, such as the age information, interest information, location information, etc. of the user, and the user behavior features may include, but are not limited to, the dynamic features of the user associated with the user account at the first moment, such as the user's habit of performing operation A, and the user's habit of performing operation B after performing operation A.

[0075] It should be noted that predictive processing is performed on device attribute features and user behavior features, and a third intelligent device for performing a third operation is determined from at least one intelligent device.

[0076] Through the embodiments provided in this application, user attribute features and user behavior features are determined from user portrait features, where the user attribute features are used to indicate the static features of the user associated with the user account at the first moment, and the user behavior features are used to indicate the dynamic features of the user associated with the user account at the first moment; predictive processing is performed on device attribute features and user behavior features, and a third intelligent device for performing a third operation is determined from at least one intelligent device. After determining the first intelligent device and the first operation, further prediction is performed based on the device attribute features associated with the first intelligent device and the user behavior features indicated by the user portrait features, to obtain a third intelligent device associated with the first intelligent device and a third operation associated with the first operation, thereby achieving the purpose of responding to multiple intelligent devices and multiple associated operations of a user request through more comprehensive feature mining and verification, and thus achieving the technical effect of improving the accuracy of device control.

[0077] As an optional solution, after performing feature extraction on the first device indication information to obtain user portrait features associated with the first device indication information and device attribute features associated with the first device indication information, the method further includes:

[0078] S1. Obtain historical user attribute features and historical user behavior features associated with historical interaction information, and determine the historical user attribute features and historical user behavior features as historical portrait features, where the historical user attribute features are used to indicate the static features of the user associated with the historical interaction information at the second moment, and the historical user behavior features are used to indicate the dynamic features of the user associated with the historical interaction information at the second moment;

[0079] S2. When the historical user attribute features match the user attribute features and the historical user behavior features match the user behavior features, determine that the historical portrait features match the user portrait features.

[0080] Optionally, in this embodiment, the historical interaction information may include, but is not limited to, historical operation information and historical portrait features, where the historical portrait features may include, but are not limited to, historical user attribute features and historical user behavior features, and the historical user attribute features may be used to indicate the static features of the user associated with the user account at the second moment, and the historical user behavior features may be used to indicate the dynamic features of the user associated with the user account at the second moment.

[0081] It should be noted that when the historical user attribute features conform to the user attribute features and the historical user behavior features conform to the user behavior features, it is determined that the historical portrait features conform to the user portrait features.

[0082] It should be noted that when the historical user attribute features do not conform to the user attribute features or the historical user behavior features do not conform to the user behavior features, it is determined that the historical portrait features do not conform to the user portrait features.

[0083] Through the embodiments provided in the present application, the historical user attribute features and historical user behavior features associated with the historical interaction information are obtained, and the historical user attribute features and historical user behavior features are determined as the historical portrait features, where the historical user attribute features are used to indicate the static features of the user associated with the historical interaction information at the second moment, and the historical user behavior features are used to indicate the dynamic features of the user associated with the historical interaction information at the second moment; when the historical user attribute features conform to the user attribute features and the historical user behavior features conform to the user behavior features, it is determined that the historical portrait features conform to the user portrait features. As an alternative solution, after obtaining the first device indication information triggered by the user account at the first moment, the method further includes:

[0084] S1. Input the first device indication information into the GPT model for feature extraction processing to obtain the first interaction corpus information, user attribute information, user behavior information associated with the first device indication information, and device service information associated with the first area, where the first area is the geographical area where the user account is located at the first moment;

[0085] S2. When the second interaction corpus information associated with the user account at the second moment is obtained, perform the first filling process based on the first interaction corpus information, user attribute information, user behavior information, device service information, and the second interaction corpus information to obtain the filled template information, where the historical interaction information includes the second interaction corpus information;

[0086] S3. Use the GPT model to perform response prediction processing on the template information to obtain a prediction output result, where the prediction output result includes: the second intelligent device and the second operation.

[0087] Optionally, in this embodiment, the first interaction corpus information may, but is not limited to, be used to indicate the voice output information of the user associated with the user account at the first moment; the user attribute information may, but is not limited to, be used to indicate the static characteristics of the user associated with the user account at the first moment, such as the user's age information, interest information, location information, etc.; the user behavior information may, but is not limited to, be the dynamic characteristics of the user associated with the user account at the first moment, such as the user is accustomed to performing operation A, and the user is accustomed to performing operation B after performing operation A; the device service information may, but is not limited to, be used to indicate the device hardware information (whether it is available) and device service priority information, etc. within the first area range at the first moment.

[0088] Optionally, in this embodiment, the second interaction corpus information may, but is not limited to, be used to indicate the voice output information and system response operation of the user associated with the user account at the second moment, where the system response operation may, but is not limited to, be used to respond to the voice output information to control a certain device to perform the corresponding operation.

[0089] It should be noted that in the case of obtaining the second interaction corpus information associated with the user account at the second moment, based on the first interaction corpus information, user attribute information, user behavior information, device service information, and second interaction corpus information, a first filling process is performed to obtain the filled template information.

[0090] It should be noted that the first filling process may, but is not limited to, be used to indicate filling the first interaction corpus information, user attribute information, user behavior information, device service information, and second interaction corpus information into a preset prompt template to obtain the filled template information, where the template information may, but is not limited to, be input into the GPT model for response prediction processing to obtain the predicted output result.

[0091] For example, write the prompt template: "Given that the user attribute information of the user is Value i , the user behavior information is often Action i , the device service information is that there is a device Hardware i , the previous interaction corpus is C (including the previous interaction corpus information q i and the previous system response r i ), and now the user says q n , the system should [MASK]".

[0092] Further, for example, after filling the first interaction corpus information, user attribute information, user behavior information, device service information, and second interaction corpus information into the prompt template to obtain the template information, the template information is input into the GPT model to obtain the result of the [MASK] part (i.e., the second intelligent device and the corresponding second operation).

[0093] Through the embodiments provided in this application, the first device indication information is input into the GPT model for feature extraction processing to obtain the first interaction corpus information, user attribute information, user behavior information associated with the first device indication information, and device service information associated with the first area, where the first area is the geographical area where the user account is located at the first moment; in the case of obtaining the second interaction corpus information associated with the user account at the second moment, based on the first interaction corpus information, user attribute information, user behavior information, device service information, and second interaction corpus information, a first filling process is performed to obtain the filled template information, where the historical interaction information includes the second interaction corpus information; using the GPT model, a response prediction process is performed on the template information to obtain a prediction output result, where the prediction output result includes: a second intelligent device, and a second operation.

[0094] As an optional solution, after obtaining the first device indication information triggered by the user account at the first moment, the method further includes:

[0095] S1, input the first device indication information into the GPT model for feature acquisition processing to obtain a first output result output by the GPT model, where the first output result includes user portrait features and device attribute features;

[0096] S2, use the GPT model to perform information verification processing on the historical interaction information and user portrait features to obtain a second output result, where the second output result is used to indicate whether the historical portrait features corresponding to the historical interaction information conform to the user portrait features;

[0097] S3, in the case of determining the second output result, input the device attribute features and prediction reference information into the GPT model for response prediction processing to obtain a third output result, where the third output result includes a second operation and a second intelligent device, and the prediction reference information includes historical interaction information or user portrait features.

[0098] Optionally, in this embodiment, the GPT model can be but is not limited to being used to perform feature acquisition processing on the first device indication information to obtain a first output result, where the first output result includes user portrait features and device attribute features.

[0099] Optionally, in this embodiment, the GPT model can be but is not limited to being used to perform information verification processing on the historical interaction information and user portrait features to obtain a second output result, where the second output result is used to indicate whether the historical portrait features corresponding to the historical interaction information conform to the user portrait features.

[0100] Optionally, in this embodiment, the GPT model can also be used, but not limited to, for response prediction processing based on device attribute features and preset reference information to obtain a third output result, where the third output result includes a second operation and a second intelligent device.

[0101] It should be noted that in the case where the second output result indicates that the historical portrait features conform to the user portrait features, the prediction reference information is the historical interaction information; in the case where the second output result indicates that the historical portrait features do not conform to the user portrait features, the prediction reference information is the user portrait features.

[0102] For further illustration, as Figure 4 shown, a device control method based on the GPT model includes:

[0103] Step S402, the prediction model 402 performs feature acquisition processing on the input first device indication information 404 to obtain a first output result 406, where the first output result 406 includes user portrait features and device attribute features;

[0104] Step S404, the prediction model 402 performs information verification processing on the historical interaction information and the user portrait features based on the first output result 406 to obtain a second output result 408, where the second output result is used to indicate whether the historical portrait features corresponding to the historical interaction information conform to the user portrait features;

[0105] Step S406, the prediction model 402 performs response prediction processing on the device attribute features and the prediction reference information based on the second output result 408 to obtain a third output result 410, where the third output result includes a predicted second intelligent device and a second operation.

[0106] Through the embodiment provided by this application, the first device indication information is input into the GPT model for feature acquisition processing to obtain a first output result output by the GPT model, where the first output result includes user portrait features and device attribute features; using the GPT model, information verification processing is performed on the historical interaction information and the user portrait features to obtain a second output result, where the second output result is used to indicate whether the historical portrait features corresponding to the historical interaction information conform to the user portrait features; in the case where the second output result is determined, the device attribute features and the prediction reference information are input into the GPT model for response prediction processing to obtain a third output result, where the third output result includes a second operation and a second intelligent device, and the prediction reference information includes historical interaction information or user portrait features.

[0107] As an optional solution, before inputting the first device indication information into the GPT model for feature extraction processing, the method further includes:

[0108] S1. Obtain an initialized first model;

[0109] S2. Use the user portrait sample data and user context sample data to train the first model to obtain a trained second model, where the user portrait sample data includes user attribute sample data and user behavior sample data, and the user context sample data includes device attribute sample data and historical interaction sample data;

[0110] S3. When the second model meets the preset convergence condition, determine the second model as the GPT model.

[0111] Optionally, in this embodiment, the initialized first model can be, but is not limited to, a GPT model constructed using open-source text data, and can be, but is not limited to, initialized using the public interface CHATGPT, and can also be, but is not limited to, initialized according to business requirements.

[0112] Optionally, in this embodiment, the first model can be, but is not limited to, trained using the user portrait sample data and user context sample data to obtain a trained second model, and when the second model meets the preset convergence condition, it is determined that the current training is completed, and the second model is determined as the GPT model.

[0113] Optionally, in this embodiment, the user portrait sample data includes user attribute sample data and user behavior sample data, and the user context sample data includes device attribute sample data and historical interaction sample data.

[0114] Through the embodiment provided by this application, an initialized first model is obtained; the first model is trained using the user portrait sample data and user context sample data to obtain a trained second model, where the user portrait sample data includes user attribute sample data and user behavior sample data, and the user context sample data includes device attribute sample data and historical interaction sample data; when the second model meets the preset convergence condition, the second model is determined as the GPT model.

[0115] As an optional solution, before using the user portrait sample data and user context sample data to train the first model to obtain a trained second model, the method further includes:

[0116] S1. Obtain initial sample data within a preset time period;

[0117] S2. Perform a first reading process on the initial sample data to obtain a set of user attribute data, and perform a second reading process on the initial sample data to obtain a set of user behavior data;

[0118] S3. Perform a second filling process and a first integration process on the user attribute data set to obtain a user behavior data set, and perform a third filling process and a second integration process on the user behavior data set to obtain user attribute sample data;

[0119] S4. Obtain user portrait sample data based on the user attribute sample data and the user behavior sample data.

[0120] Optionally, in this embodiment, the initial sample data may but is not limited to include user interaction logs, where the user interaction logs may but is not limited to include corpus information of N rounds of interactions between the user associated with the user account and at least one intelligent device. Among them, the corpus information of each round of interaction may but is not limited to include a device instruction information triggered by the user account and at least one intelligent device and at least one response operation in response.

[0121] Optionally, in this embodiment, the user attribute data set may but is not limited to include multiple static features of the user associated with the user account, such as the user's age information, location information, interest information, etc.

[0122] Optionally, in this embodiment, the user behavior data set may but is not limited to include multiple dynamic features of the user associated with the user account, such as the user often likes to perform operation A, the user is used to performing operation B after performing operation A, etc.

[0123] It should be noted that user portrait sample data is obtained based on the user attribute sample data and the user behavior sample data.

[0124] Through the embodiments provided in this application, obtain the initial sample data within a preset time period; perform a first reading process on the initial sample data to obtain a user attribute data set, and perform a second reading process on the initial sample data to obtain a user behavior data set; perform a second filling process and a first integration process on the user attribute data set to obtain a user behavior data set, and perform a third filling process and a second integration process on the user behavior data set to obtain user attribute sample data; obtain user portrait sample data based on the user attribute sample data and the user behavior sample data.

[0125] As an alternative solution, apply the above GPT-based device control method in a GPT-based personalized interaction self-learning scenario. In this scenario, there are many applications that require human-computer interaction, such as voice device control in smart homes, artificial customer service, etc. During the human-computer interaction process, each user has their own usage characteristics and personalized needs. The interaction system needs to generate corresponding different decision-making schemes according to the different usage habits of each user to achieve user-friendly interaction and make the interaction more convenient and accurate.

[0126] Optionally, in this embodiment, it is possible but not limited to use the technology of constructing a user profile, and then use the tags in the user profile to assist the dialogue system in making decisions to achieve personalized services.

[0127] Optionally, in this embodiment, the information included in the user profile generally has two types: user attributes and user behaviors. Among them, the user attributes mainly include: the user's gender, age, education level, income level, city where they are located, etc.; the behavior information mainly includes: user browsing information, click situation (product website), interaction corpus, behavior habits (voice interaction system), etc.

[0128] Optionally, in this embodiment, the system uses the information in the constructed user profile to provide better services for users. For example, the information of the user profile is used for product screening conditions, search result sorting score adjustment, conversation decision-making judgment, etc.

[0129] Optionally, in this embodiment, a personalized interactive self-learning system based on GPT includes a model construction module, a user profile construction module, and a user profile application module, specifically including:

[0130] (1) Model construction module

[0131] Using open-source text data, construct a GPT (Generative Pretrained Model) model. It is possible but not limited to use the public interface CHATGPT for construction, and it is possible but not limited to train the GPT model according to business needs.

[0132] (2) User profile construction module

[0133] (1) User attribute generation sub-module

[0134] Read the user interaction log from the log, denoted as C = {q0, r0, q1, r1, …, q n , r n}, where q i is the user corpus of the i-th round of interaction, and r i is the system response of the i-th round of interaction.

[0135] Determine the set of user attributes to be obtained {Property0, Property1, …, Property n}.

[0136] Write a prompt template, for example, "According to the chat record C, it can be known that the speaker is [MASK]Property i ".

[0137] Fill C into the prompt template, input it into the GPT model to obtain the result of the [MASK] part, and normalize the information of [MASK] using simple rules or statistical models.

[0138] (2) User Behavior Generation Sub-module

[0139] Read the user interaction log from the log, denoted as C = {q0, r0, q1, r1, …, q n , r n}, where q i is the user corpus of the i-th round of interaction, and r i is the system response of the i-th round of interaction.

[0140] Write prompt templates, such as "According to the chat record C, the speaker often [MASK]", "According to the chat record C, what the speaker likes to operate is [MASK]", "According to the chat record C, the speaker often performs the action of [MASK2] after operating [MASK1]".

[0141] Fill C into the prompt template, input it into the GPT model to obtain the result of the [MASK] part, and normalize the information of [MASK] using simple rules or statistical models.

[0142] (3) User Portrait Application Module

[0143] The user portrait information includes two parts: namely, user information {Property i , Value i} and user behavior {Action i}.

[0144] The user context information includes: user home hardware information {Hardware i}, the previous voice interaction information C = {q i , r i}.

[0145] The current user's utterance information is q n .

[0146] Write a prompt template for predicting the current interaction result: "Given that the user's Property i is Value i , the user often Action i , the user has Hardware i at home, the previous interaction corpus is C, and now the user says q n , the system should [MASK]".

[0147] Fill the user profile information, user context information, and the current round of user utterance information into the prompt template, input it into the GPT model to obtain the result of the [MASK] part, and normalize the information of [MASK] using simple rules or statistical models.

[0148] Through the embodiments provided in this application, using user logs and prompt templates, automatically construct user attributes in the user profile and automatically summarize user behaviors in the user profile, and use the user profile + context + prompt template to achieve personalized user interaction, improving interaction efficiency and accuracy.

[0149] It can be understood that in the specific implementation of this application, it involves data related to user information, etc. When the above embodiments of this application are applied to specific products or technologies, user permission or consent needs to be obtained, and the collection, use, and processing of relevant data need to comply with relevant laws, regulations, and standards of relevant countries and regions.

[0150] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of operation combinations. However, those skilled in the art should know that this application is not limited by the described operation sequence, because according to this application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the operations and modules involved are not necessarily essential to this application.

[0151] According to another aspect of the embodiments of this application, there is also provided a GPT-based device control device for implementing the above GPT-based device control method. As Figure 5 shown, the device includes:

[0152] An acquisition unit 502, configured to acquire first device indication information triggered by a user account at a first moment, where the first device indication information is used to instruct a first intelligent device to perform a first operation;

[0153] An extraction unit 504, configured to extract features from the first device indication information to obtain user profile features associated with the first device indication information and device attribute features associated with the first device indication information, where the user profile features are used to represent the user profile when the user account triggers the first device indication information, and the device attribute features are used to represent the first device attribute of the first intelligent device when the user account triggers the first device indication information;

[0154] A determination unit 506, configured to, when the historical interaction information at the second moment is obtained and the historical portrait features corresponding to the historical interaction information conform to the user portrait features, determine, from the historical interaction information, the historical operation of the first intelligent device on the second device indication information of the user account at the second moment, and the historical operation information corresponding to the historical operation, where the historical interaction information is the information of the interaction between the user account and the first intelligent device, and the second moment is the historical moment of the first moment;

[0155] A control unit 508, configured to perform prediction processing on the device attribute features and the historical operation information through a GPT model to obtain at least two intelligent devices including the first intelligent device and the second intelligent device, and control the first intelligent device to perform the first operation and control the second intelligent device to perform a second operation, where the second operation is associated with the first operation, and the second device attribute of the second intelligent device is associated with the first device attribute.

[0156] As an optional solution, the determination unit 506 includes:

[0157] Input the device attribute features and the historical operation information into the GPT model to perform the prediction processing, and obtain a target operation set output by the GPT model, where the target operation set includes at least one response operation, and the at least one response operation is associated with the first operation;

[0158] Obtain a first response operation with the highest association confidence degree with the first operation from the at least one response operation, where the association confidence degree is used to indicate the association degree between the at least one response operation and the first operation;

[0159] Obtain a third device attribute associated with the first response device, where the first response device is used to perform the first response operation;

[0160] When the third device attribute meets the device confidence condition, determine the first response operation as the second operation and determine the first response device as the second intelligent device, where the at least two intelligent devices include the second intelligent device.

[0161] As an optional solution, the device further includes:

[0162] A control module is configured to extract features from the first device indication information to obtain user portrait features associated with the first device indication information and device attribute features associated with the first device indication information. After obtaining historical interaction information and when the historical portrait features corresponding to the historical interaction information do not match the user portrait features, the control module determines a third intelligent device from at least one intelligent device based on the device attribute features and the user portrait features, controls the first intelligent device to perform a first operation, and controls the third intelligent device to perform a third operation. Herein, the third device attribute of the third intelligent device is mutually associated with the first device attribute, the first operation is mutually associated with the third operation, the second moment is the historical moment of the first moment, and the historical interaction information is the information of the interaction between the user account and the first intelligent device.

[0163] As an alternative solution, the above control module includes:

[0164] A second determination sub-module is configured to determine user attribute features and user behavior features from the user portrait features. The user attribute features are used to indicate the static features of the user associated with the user account at the first moment, and the user behavior features are used to indicate the dynamic features of the user associated with the user account at the first moment.

[0165] A prediction sub-module is configured to perform prediction processing on the device attribute features and the user behavior features, and determine a second intelligent device for performing a second operation from at least one intelligent device.

[0166] As an alternative solution, the above device further includes:

[0167] A first acquisition module is configured to, after extracting features from the first device indication information to obtain user portrait features associated with the first device indication information and device attribute features associated with the first device indication information, acquire historical user attribute features and historical user behavior features associated with the historical interaction information, and determine the historical user attribute features and the historical user behavior features as historical portrait features. The historical user attribute features are used to indicate the static features of the user associated with the historical interaction information at the second moment, and the historical user behavior features are used to indicate the dynamic features of the user associated with the historical interaction information at the second moment;

[0168] A first determination module is configured to, after extracting features from the first device indication information to obtain user portrait features associated with the first device indication information and device attribute features associated with the first device indication information, determine that the historical portrait features match the user portrait features when the historical user attribute features match the user attribute features and the historical user behavior features match the user behavior features.

[0169] As an alternative solution, the above device further includes:

[0170] An extraction module, configured to, after obtaining first device indication information triggered by a user account at a first moment, input the first device indication information into a GPT model for feature extraction processing to obtain first interaction corpus information, user attribute information, user behavior information, and device service information associated with a first region, where the first region is the geographical region where the user account is located at the first moment; GPT model GPT model

[0171] A second determination module, configured to, after obtaining first device indication information triggered by a user account at a first moment, in the case of obtaining second interaction corpus information associated with the user account at a second moment, perform first filling processing based on the first interaction corpus information, user attribute information, user behavior information, device service information, and second interaction corpus information to obtain filled template information, where the historical interaction information includes the second interaction corpus information GPT model;

[0172] A third determination module, configured to, after obtaining first device indication information triggered by a user account at a first moment, use the GPT model to perform response prediction processing on the template information to obtain a prediction output result, where the prediction output result includes: a second intelligent device, and a second operation GPT model.

[0173] As an optional solution, the above device further includes:

[0174] A second acquisition module, configured to obtain an initialized first model before inputting the first device indication information into the GPT model for feature extraction processing;

[0175] A training module, configured to, before inputting the first device indication information into the GPT model for feature extraction processing, use user portrait sample data and user context sample data to train the first model to obtain a trained second model, where the user portrait sample data includes user attribute sample data and user behavior sample data, and the user context sample data includes device attribute sample data and historical interaction sample data;

[0176] A fourth determination module, configured to, before inputting the first device indication information into the GPT model for feature extraction processing, in the case where the second model meets a preset convergence condition, determine the second model as the GPT model.

[0177] As an optional solution, the above device further includes:

[0178] A third acquisition module, configured to obtain initial sample data within a preset time period before using the user portrait sample data and user context sample data to train the first model to obtain a trained second model;

[0179] A reading module, configured to perform a first reading process on initial sample data to obtain a set of user attribute data, and perform a second reading process on the initial sample data to obtain a set of user behavior data, before training a first model using user portrait sample data and user context sample data to obtain a trained second model;

[0180] A fourth determination module, configured to perform a second filling process and a first integration process on the set of user attribute data to obtain a set of user behavior data, and perform a third filling process and a second integration process on the set of user behavior data to obtain user attribute sample data, before training a first model using user portrait sample data and user context sample data to obtain a trained second model;

[0181] A fifth determination module, configured to obtain user portrait sample data based on user attribute sample data and user behavior sample data, before training a first model using user portrait sample data and user context sample data to obtain a trained second model.

[0182] For specific embodiments, reference may be made to the examples shown in the above device control method, and details will not be elaborated herein.

[0183] According to another aspect of the embodiments of the present application, there is also provided an electronic device for implementing the above GPT-based device control method, as Figure 6 shown. The electronic device includes a memory 602 and a processor 604. A computer program is stored in the memory 602, and the processor 604 is configured to execute the steps in any of the above method embodiments through the computer program.

[0184] Optionally, in this embodiment, the above electronic device may be at least one network device among multiple network devices in a computer network.

[0185] Optionally, in this embodiment, the above processor may be configured to execute the steps in the above device control method through the computer program.

[0186] Optionally, those of ordinary skill in the art can understand that Figure 6 the structure shown is only schematic. The electronic device may also be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a palm computer, and a mobile Internet device (MID), a PAD, or other terminal devices. Figure 6 It does not limit the structure of the above electronic device. For example, the electronic device may further include more or fewer components (such as a network interface, etc.) than those shown in Figure 6 , or have a different configuration from that shown in Figure 6 .

[0187] Among them, the memory 602 can be used to store software programs and modules, such as the program instructions / modules corresponding to the device control method and device in the embodiments of the present application. The processor 604 executes various functional applications and data processing by running the software programs and modules stored in the memory 602, that is, implements the above-mentioned device control method. The memory 602 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memories. In some instances, the memory 602 may further include a memory remotely disposed relative to the processor 604, and these remote memories may be connected to the terminal through a network. Examples of the above network include but are not limited to the Internet, enterprise intranet, local area network, mobile communication network, and combinations thereof. Among them, the memory 602 may specifically but not limitedly be used to store information such as first device indication information. As an example, as Figure 6 shown, the above-mentioned memory 602 may include but are not limited to the acquisition unit 502, extraction unit 504, determination unit 506, and control unit 508 in the above-mentioned device control device. In addition, it may also include but are not limited to other module units in the above-mentioned device control device, which will not be elaborated in this example.

[0188] Optionally, the above-mentioned transmission device 606 is used to receive or send data via a network. Specific examples of the above-mentioned network may include a wired network and a wireless network. In one instance, the transmission device 606 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices and routers through a network cable, so as to communicate with the Internet or local area network. In one instance, the transmission device 606 is a radio frequency (Radio Frequency, RF) module, which is used to communicate with the Internet wirelessly.

[0189] In addition, the above-mentioned electronic device further includes: a display 608, which is used to display information such as the above-mentioned first device indication information; and a connection bus 610, which is used to connect each module component in the above-mentioned electronic device.

[0190] According to one aspect of the present application, a computer program product is provided. The computer program product includes computer programs / instructions, and the computer programs / instructions contain program codes for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part, and / or installed from a removable medium. When the computer program is executed by the central processing unit, it executes various functions provided by the embodiments of the present application.

[0191] The serial numbers of the above-mentioned embodiments of the present application are only for description, and do not represent the advantages or disadvantages of the embodiments.

[0192] It should be noted that the computer system of the electronic device is only an example and should not impose any limitations on the functions and scope of use of the embodiments of the present application.

[0193] The computer system includes a Central Processing Unit (CPU), which can perform various appropriate operations and processes according to the program stored in the Read-Only Memory (ROM) or the program loaded from the storage section into the Random Access Memory (RAM). In the random access memory, various programs and data required for system operation are also stored. The central processing unit, the read-only memory, and the random access memory are connected to each other through a bus. An Input / Output interface (I / O interface) is also connected to the bus.

[0194] In particular, according to the embodiments of the present application, the processes described in the various method flowcharts can be implemented as computer software programs. For example, the embodiments of the present application include a computer program product that includes a computer program carried on a computer-readable medium, and the computer program contains program codes for executing the methods shown in the flowcharts. In such an embodiment, the computer program can be downloaded and installed from the network through the communication section and / or installed from a removable medium. When the computer program is executed by the central processing unit, various functions defined in the system of the present application are executed.

[0195] According to an aspect of the present application, a computer-readable storage medium is provided. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the GPT-based device control method provided in the above various alternative implementation manners.

[0196] Optionally, in this embodiment, the above computer-readable storage medium may be set to store a computer program for executing the steps in the above device control method.

[0197] Optionally, in this embodiment, those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by a program instructing the relevant hardware of the terminal device, and the program can be stored in a computer-readable storage medium. The storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc, etc.

[0198] The serial numbers of the embodiments of the present application above are only for description and do not represent the superiority or inferiority of the embodiments.

[0199] If the integrated units in the above embodiments are implemented in the form of software function units and sold or used as independent products, they can be stored in the above computer-readable storage media. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing one or more computer devices (which can be personal computers, servers, or network devices, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application.

[0200] In the above embodiments of the present application, the descriptions of the various embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0201] In the several embodiments provided by the present application, it should be understood that the disclosed client can be implemented in other ways. Among them, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces. The indirect coupling or communication connection of units or modules can be in an electrical or other form.

[0202] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0203] In addition, the functional units in the various embodiments of the present application can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software function units.

[0204] The above is only the preferred embodiment of the present application. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present application, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present application.

Claims

1. A GPT-based device control method, characterized in that, Including: Obtain first device indication information triggered by a user account at a first moment, where the first device indication information is used to indicate controlling a first intelligent device to perform a first operation; Extract features from the first device indication information to obtain user portrait features associated with the first device indication information and device attribute features associated with the first device indication information, where the user portrait features are used to represent the user portrait when the user account triggers the first device indication information, and the device attribute features are used to represent the first device attributes of the first intelligent device when the user account triggers the first device indication information; When historical interaction information at a second moment is obtained and the historical portrait features corresponding to the historical interaction information match the user portrait features, determine from the historical interaction information the historical operation of the first intelligent device on a second device indication information of the user account at the second moment and the historical operation information corresponding to the historical operation, where the historical interaction information is information of interaction between the user account and the first intelligent device, and the second moment is a historical moment of the first moment; through a GPT model, perform prediction processing on the device attribute features and the historical operation information to obtain at least two intelligent devices including the first intelligent device and a second intelligent device, and control the first intelligent device to perform the first operation and control the second intelligent device to perform a second operation, where the second operation is associated with the first operation, and the second device attributes of the second intelligent device are associated with the first device attributes.

2. The method according to claim 1, wherein The performing, through the GPT model, prediction processing on the device attribute features and the historical operation information to obtain at least two intelligent devices including the first intelligent device and a second intelligent device includes: Input the device attribute features and the historical operation information into the GPT model to perform the prediction processing to obtain a target operation set output by the GPT model, where the target operation set includes at least one response operation, and the at least one response operation is associated with the first operation; obtain a first response operation with the highest association confidence degree with the first operation from the at least one response operation, where the association confidence degree is used to indicate the degree of association between the at least one response operation and the first operation; Obtain third device attributes associated with the first response device, where the first response device is used to perform the first response operation; When the third device attributes meet the device confidence condition, determine the first response operation as the second operation and determine the first response device as the second intelligent device, where the at least two intelligent devices include the second intelligent device.

3. The method according to claim 1, characterized in that After the extracting features from the first device indication information to obtain user portrait features associated with the first device indication information and device attribute features associated with the first device indication information, the method further includes: In the case where the historical interaction information is obtained and the historical portrait features corresponding to the historical interaction information do not conform to the user portrait features, a third intelligent device is determined from the at least one intelligent device based on the device attribute features and the user portrait features, and the first intelligent device is controlled to perform the first operation, and the third intelligent device is controlled to perform the third operation, where the third device attribute of the third intelligent device is correlated with the first device attribute, the first operation is correlated with the third operation, the second moment is the historical moment of the first moment, and the historical interaction information is the information of the interaction between the user account and the first intelligent device.

4. The method according to claim 3, wherein The determining of the third intelligent device from the at least one intelligent device based on the device attribute features and the user portrait features includes: Determining user attribute features and user behavior features from the user portrait features, where the user attribute features are used to indicate the static features of the user associated with the user account at the first moment, and the user behavior features are used to indicate the dynamic features of the user associated with the user account at the first moment. Performing a prediction process on the device attribute features and the user behavior features, and determining the third intelligent device for performing the third operation from the at least one intelligent device.

5. The method according to claim 4, characterized in that, After performing feature extraction on the first device indication information to obtain the user portrait features associated with the first device indication information and the device attribute features associated with the first device indication information, the method further includes: Obtaining historical user attribute features and historical user behavior features associated with the historical interaction information, and determining the historical user attribute features and the historical user behavior features as the historical portrait features, where the historical user attribute features are used to indicate the static features of the user associated with the historical interaction information at the second moment, and the historical user behavior features are used to indicate the dynamic features of the user associated with the historical interaction information at the second moment; In the case where the historical user attribute features conform to the user attribute features and the historical user behavior features conform to the user behavior features, determining that the historical portrait features conform to the user portrait features.

6. The method according to claim 1, wherein After obtaining the first device indication information triggered by the user account at the first moment, the method further includes: Inputting the first device indication information into the GPT model for feature extraction processing to obtain the first interaction corpus information, user attribute information, user behavior information associated with the first device indication information, and device service information associated with the first region, where the first region is the geographical region where the user account is located at the first moment; In the case of obtaining the second interactive corpus information associated with the user account at the second moment, perform a first filling process based on the first interactive corpus information, the user attribute information, the user behavior information, the device service information, and the second interactive corpus information to obtain the filled template information, where the historical interactive information includes the second interactive corpus information; Use the GPT model to perform a response prediction process on the template information to obtain a prediction output result, where the prediction output result includes: the second intelligent device and the second operation.

7. The method according to claim 6, wherein Before inputting the first device indication information into the GPT model for feature extraction processing, the method further includes: Obtain an initialized first model; Use user portrait sample data and user context sample data to train the first model to obtain a trained second model, where the user portrait sample data includes user attribute sample data and user behavior sample data, and the user context sample data includes device attribute sample data and historical interaction sample data; In the case where the second model meets the preset convergence condition, determine the second model as the GPT model.

8. The method according to claim 7, wherein Before using the user portrait sample data and user context sample data to train the first model to obtain a trained second model, the method further includes: Obtain initial sample data within a preset time period; Perform a first reading process on the initial sample data to obtain a user attribute data set, and perform a second reading process on the initial sample data to obtain a user behavior data set; Perform a first filling process and a first integration process on the user attribute data set to obtain the user behavior data set, and perform a second filling process and a second integration process on the user behavior data set to obtain the user attribute sample data; Based on the user attribute sample data and the user behavior sample data, obtain the user portrait sample data.

9. A device control device based on GPT, characterized in that, Includes: An acquisition unit, configured to acquire first device indication information triggered by a user account at a first moment, where the first device indication information is used to instruct a first intelligent device to perform a first operation; An extraction unit, configured to perform feature extraction on the first device indication information to obtain a user portrait feature associated with the first device indication information and a device attribute feature associated with the first device indication information, where the user portrait feature is used to represent the user portrait when the user account triggers the first device indication information, and the device attribute feature is used to represent the first device attribute of the first intelligent device when the user account triggers the first device indication information; A determination unit, configured to determine, when historical interaction information at a second moment is obtained and the historical portrait features corresponding to the historical interaction information conform to the user portrait features, a historical operation of the first intelligent device on the second device indication information of the user account at the second moment and historical operation information corresponding to the historical operation from the historical interaction information, where the historical interaction information is information of an interaction between the user account and the first intelligent device, and the second moment is a historical moment of the first moment; A control unit, configured to perform prediction processing on the device attribute features and the historical operation information through a GPT model to obtain at least two intelligent devices including the first intelligent device and a second intelligent device, and control the first intelligent device to perform the first operation and control the second intelligent device to perform a second operation, where the second operation is associated with the first operation, and the second device attribute of the second intelligent device is associated with the first device attribute.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, where the program, when running, executes the method according to any one of claims 1 to 8.

11. An electronic device, comprising a memory and a processor, characterized in that, A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 8 through the computer program.

Images