Data processing method and system and related equipment

Abstract

The embodiment of the invention discloses a data processing method and system and related equipment. The method comprises the following steps: receiving first request information input by a user; sending first request information to a second unit; receiving first response information corresponding to the first request information sent by the second unit; sending second request information to a second unit; the second request information is used for confusing the first request information, and receiving second response information corresponding to the second request information sent by the second unit. The false request information is acquired through the trusted unit or the device (the first unit or the third unit), and the true request information is mixed by using the false request information, so that the trustless unit cannot clarify the true request information, and the privacy protection of the user is improved.

Description

[0001] This application claims the priority of the Chinese patent application filed with the State Intellectual Property Office on September 22, 2023, with application number 202311241182.7 and invention name “A data processing method and related equipment”, the entire contents of which are incorporated by reference in this application. Technical Field

[0002] The embodiments of the present application relate to the field of artificial intelligence, and in particular, to a data processing method, system, and related equipment. Background Art

[0003] Artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a branch of computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is also the study of the design principles and implementation methods of various intelligent machines, so that machines have the functions of perception, reasoning and decision-making. Research in the field of artificial intelligence includes robotics, natural language processing, computer vision, decision-making and reasoning, human-computer interaction, recommendation and search, basic AI theory, etc.

[0004] In the end-to-cloud scenario, the end side sends the request information (such as questions) input by the user to the cloud side, which infers the information stored in the cloud side and returns the response information (such as answers) to the end side, which then presents the response information to the user. It can be seen from this process that the request information input by the user is actually transmitted out of the user side, which will bring the risk of privacy leakage.

[0005] Therefore, how to reduce the leakage of user privacy while using cloud-based reasoning capabilities is a technical problem that needs to be solved urgently. Summary of the invention

[0006] The embodiments of the present application provide a data processing method and related equipment, which can obtain false request information based on the real request information input by the user or the real response information corresponding to the real request information, so that the first unit can use the false request information to confuse the real request information, and use the false request information to confuse the model, thereby improving the user's privacy protection.

[0007] The first aspect of the embodiment of the present application provides a data processing method. The method can be executed by a first device / first unit, or by a component of the first device / first unit (such as a processor, a chip, or a chip system, etc.). In the method, a first request message input by a user is sent to a second unit; a first response message corresponding to the first request message sent by the second unit is received; it is characterized in that the method also includes: sending a second request message to the second unit, the second request message is used to confuse the first request message; receiving a second response message corresponding to the second request message sent by the second unit. Among them, the first request message can also be understood as a real request message, the first response message can also be understood as a real response message, the second request message can be understood as a false request message, and the second response message can also be understood as a false response message.

[0008] In an embodiment of the present application, false request information is obtained through a trusted unit or device (first unit or third unit), so that the first unit can use the false request information to confuse the real request information, thereby making it impossible for the untrusted unit to clearly identify the real request information, thereby improving the user's privacy protection.

[0009] Optionally, in a possible implementation manner of the first aspect, the above steps further include: acquiring second request information based on the first request information.

[0010] In this possible implementation, false request information corresponding to the real request information is obtained through a trusted unit or device (first unit or third unit), so that the first unit can use the false request information to prevent the untrusted unit from clearly identifying the real request information, thereby improving user privacy protection.

[0011] Optionally, in a possible implementation manner of the first aspect, the above step: obtaining the second request information based on the first request information includes: determining a first keyword in the first request information; and obtaining the second request information based on the first keyword and the first request information.

[0012] In this possible implementation, false request information is obtained by means of a first keyword, thereby confusing an untrusted unit by blurring / obfuscating the keyword, so that the untrusted unit cannot clearly identify the real request information, thereby improving user privacy protection.

[0013] Optionally, in a possible implementation of the first aspect, the above step of: obtaining the second request information based on the first keyword and the first request information includes: determining a first replacement word corresponding to the first keyword; replacing the first keyword in the first request information with the first replacement word to obtain the second request information.

[0014] In this possible implementation, a related first replacement word is obtained through the first keyword, and the first replacement word is used to replace the first keyword in the first request information, so that the first keyword can be hidden and obfuscated, reducing the risk of the first keyword being identified by an untrusted unit and improving user privacy.

[0015] Optionally, in a possible implementation manner of the first aspect, the above step: obtaining the second request information based on the first keyword and the first request information includes: sending the first keyword and the first request information to a third unit; and receiving the second request information sent by the third unit.

[0016] In this possible implementation, false request information is generated by a trusted unit (ie, the third unit), so that the real request information can be hidden from the untrusted unit, thereby improving user privacy.

[0017] Optionally, in a possible implementation manner of the first aspect, the above step of: determining the first keyword in the first request information includes: determining the first keyword based on a first operation of the user or a first preset rule.

[0018] In this possible implementation, there are multiple ways to determine the keyword, which can be selected according to user operations or preset rules, thereby improving the applicability of the solution.

[0019] Optionally, in a possible implementation manner of the first aspect, the above step: obtaining the second request information based on the first request information includes: sending the first request information to a third unit; and receiving the second request information sent by the third unit.

[0020] In this possible implementation, false request information is generated by a trusted unit (ie, the third unit), so that the real request information can be hidden from the untrusted unit, thereby improving user privacy.

[0021] Optionally, in a possible implementation manner of the first aspect, the above steps also include: generating second request information based on the first response information.

[0022] In this possible implementation, the real response information is first obtained through the real request information, and then multiple false request information is obtained based on the real response information. In general, the second unit retrieves logs and other methods within a certain time period, as long as the second unit infers that the time interval between the real request information and the false request information is less than the above time period. In terms of the log, the first unit can confuse the real request information with multiple false request information, thereby improving the user's privacy protection.

[0023] Optionally, in a possible implementation manner of the first aspect, the above step: generating the second request information based on the first response information includes: obtaining the third response information based on the first response information; and obtaining the second request information based on the third response information.

[0024] In this possible implementation, the real response information is first obtained through the real request information, and then the real response information is obfuscated to obtain multiple false response information, so that the multiple false response information can be reversed to obtain multiple false request information. In general, the second unit retrieves the log and other methods within a certain time period, as long as the second unit infers that the time interval between the real request information and the false request information is less than the above time period. In terms of the log, the first unit can confuse the real request information with multiple false request information, thereby improving the user's privacy protection.

[0025] Optionally, in a possible implementation of the first aspect, the above step of: obtaining the third response information based on the first response information includes: determining the second keyword in the first response information; determining the second replacement character corresponding to the second keyword; replacing the second keyword in the first response information with the second replacement character to obtain the third response information.

[0026] In this possible implementation, multiple false request information is obtained based on the second keyword in the real result. Since the second unit generally retrieves logs and other methods within a certain time period, as long as the time interval between the real request information and the false request information inferred by the second unit is less than the above time period, in the embodiment of the log, the first unit can confuse the real request information with multiple false request information, thereby improving the privacy protection of users.

[0027] Optionally, in a possible implementation manner of the first aspect, the above steps also include: presenting multiple response information or first response information to the user, the multiple response information including the first response information and the second response information.

[0028] In this possible implementation, it is convenient for the user to use the second unit to process data to obtain the response information required by the user, thereby improving the user experience.

[0029] Optionally, in a possible implementation of the first aspect, the above steps also include: receiving a first request message sent by a third unit; sending multiple response messages or first response messages to the third unit, the multiple response messages including the first response message and the second response message.

[0030] In this possible implementation, the third unit used by the user can obtain the response information through the trusted first unit, thereby providing the response information required by the user and improving the user experience.

[0031] Optionally, in a possible implementation manner of the first aspect, the first unit or the first device is a question answering device.

[0032] In this possible implementation, the first unit may be a question answering device that provides requested services to users. Users may ask questions to the question answering device and receive answers from the question answering device, thereby providing the answers required by users and improving user experience.

[0033] Optionally, in a possible implementation manner of the first aspect, the first request information includes at least one of the following: a question, a text processing request, and a translation request.

[0034] In this possible implementation, the first request information can provide multiple requirements for the user, thereby improving the flexibility of the solution.

[0035] The second aspect of the embodiment of the present application provides a data processing method. The method can be executed by a third device / third unit, or by a component of the third device / third unit (such as a processor, a chip, or a chip system, etc.). In the method, a first request message sent by a first unit or a first response message corresponding to the first request message is received; a second request message is sent to the first unit based on the first request message or the first response message, and the second request message is used to confuse the first request message.

[0036] In an embodiment of the present application, false request information is obtained based on real request information or real response information through a trusted unit or device (third unit), so that the false request information can be used to confuse the real request information, making it impossible for the untrusted unit to clearly identify the real request information, thereby improving the user's privacy protection.

[0037] Optionally, in a possible implementation manner of the second aspect, the third device / third unit is deployed on a network server, and the first unit is deployed on a terminal device.

[0038] In this possible implementation, a trusted network server determines the second request information for the terminal device, so that the terminal device can use false request information to confuse the real request information, making it impossible for the untrusted second unit to clearly identify the real request information, thereby improving user privacy protection.

[0039] Optionally, in a possible implementation of the second aspect, the above-mentioned step: sending the second request information to the first unit based on the first request information or the first response information includes: obtaining the second request information based on the first request information or the first response information, and sending the second request information to the first unit.

[0040] In this possible implementation, the trusted unit or device (third unit) obtains false request information based on the real request information or the real response information, and sends the second request information to the first unit. Thus, the first unit can subsequently confuse the real request information with the false request information, reducing the risk of the real request information being identified by the untrusted unit (third unit).

[0041] The third aspect of the embodiment of the present application provides a data processing method. The method can be performed by the first unit and the third unit together, or by a component of the first unit (such as a processor, a chip, or a chip system, etc.) and a component of the third unit (such as a processor, a chip, or a chip system, etc.). In this method, the first unit sends a first request message input by a user to the second unit; the second unit sends a first response message corresponding to the first request message to the first unit; the second unit receives a second request message sent by the first unit, and the second request message is used to confuse the first request message; the first unit receives a second response message corresponding to the second request message sent by the second unit.

[0042] In an embodiment of the present application, real request information and false request information are sent to an untrusted second unit through a trusted first unit. The first unit can use the false request information to confuse the real request information, so that the untrusted second unit cannot clearly identify the real request information, thereby improving user privacy protection.

[0043] Optionally, in a possible implementation manner of the third aspect, the above steps also include: the first unit obtains the second request information based on the first request information or the first response information.

[0044] In this possible implementation, the false request information corresponding to the real request information is obtained through the trusted first unit, so that the first unit can use the false request information to prevent the untrusted unit from clearly identifying the real request information, thereby improving user privacy protection.

[0045] Optionally, in a possible implementation manner of the third aspect, the above steps also include: the first unit obtains the second request information based on the first response information.

[0046] In this possible implementation, the real response information is first obtained through the real request information, and then multiple false request information is obtained based on the real response information. In general, the second unit retrieves logs and other methods within a certain time period, as long as the second unit infers that the time interval between the real request information and the false request information is less than the above time period. In terms of the log, the first unit can confuse the real request information with multiple false request information, thereby improving the user's privacy protection.

[0047] The fourth aspect of an embodiment of the present application provides a first device, which includes: a transceiver unit, used to send first request information input by a user to a second unit; the transceiver unit is also used to receive first response information corresponding to the first request information sent by the second unit; the transceiver unit is also used to send second request information to the second unit, and the second request information is used to confuse the first request information; the transceiver unit is also used to receive second response information corresponding to the second request information sent by the second unit.

[0048] Optionally, in a possible implementation manner of the fourth aspect, the first device further includes: a processing unit, configured to obtain second request information based on the first request information.

[0049] Optionally, in a possible implementation manner of the fourth aspect, the above-mentioned processing unit is specifically used to determine the first keyword in the first request information; the processing unit is specifically used to obtain the second request information based on the first keyword and the first request information.

[0050] Optionally, in a possible implementation of the fourth aspect, the above-mentioned processing unit is specifically used to determine a first replacement word corresponding to the first keyword; the processing unit is specifically used to replace the first keyword in the first request information with the first replacement word to obtain the second request information.

[0051] Optionally, in a possible implementation manner of the fourth aspect, the above-mentioned processing unit is specifically used to send the first keyword and the first request information to the third unit; the processing unit is specifically used to receive the second request information sent by the third unit.

[0052] Optionally, in a possible implementation manner of the fourth aspect, the above-mentioned processing unit is specifically used to determine the first keyword based on a first operation of the user or a first preset rule.

[0053] Optionally, in a possible implementation manner of the fourth aspect, the above-mentioned processing unit is specifically used to send first request information to the third unit; the processing unit is specifically used to receive second request information sent by the third unit.

[0054] Optionally, in a possible implementation manner of the fourth aspect, the first device further includes: a processing unit, configured to generate second request information based on the first response information.

[0055] Optionally, in a possible implementation of the fourth aspect, the above-mentioned processing unit is specifically used to obtain third response information based on the first response information; the processing unit is specifically used to obtain second request information based on the third response information.

[0056] Optionally, in a possible implementation of the fourth aspect, the above-mentioned processing unit is specifically used to determine the second keyword in the first response information; the processing unit is specifically used to determine the second replacement word corresponding to the second keyword; the processing unit is specifically used to replace the second keyword in the first response information with the second replacement word to obtain a third response information.

[0057] Optionally, in a possible implementation manner of the fourth aspect, the above-mentioned first device also includes: a presentation unit, used to present multiple response information or first response information to the user, and the multiple response information includes the first response information and the second response information.

[0058] Optionally, in a possible implementation of the fourth aspect, the above-mentioned transceiver unit is also used to receive a first request message sent by a third unit; the transceiver unit is also used to send multiple response messages or first response messages to the third unit, and the multiple response messages include first response messages and second response messages.

[0059] Optionally, in a possible implementation manner of the fourth aspect, the first unit or the first device is a question answering device.

[0060] Optionally, in a possible implementation manner of the fourth aspect, the first request information includes at least one of the following: a question, a text processing request, and a translation request.

[0061] The fifth aspect of an embodiment of the present application provides a third device, which includes: a transceiver unit, used to receive a first request message sent by a first unit or a first response message corresponding to the first request message; the transceiver unit is also used to send a second request message to the first unit based on the first request message or the first response message, and the second request message is used to confuse the first request message.

[0062] Optionally, in a possible implementation manner of the fifth aspect, the third device is deployed on a network server, and the first unit is deployed on a terminal device.

[0063] Optionally, in a possible implementation manner of the fifth aspect, the above-mentioned transceiver unit is specifically used to obtain the second request information based on the first request information or the first response information, and the transceiver unit is specifically used to send the second request information to the first unit.

[0064] A sixth aspect of an embodiment of the present application provides a first device, comprising: a processor, the processor being coupled to a memory, the memory being used to store programs or instructions, and when the programs or instructions are executed by the processor, the first device implements the method in the above-mentioned first aspect or any possible implementation manner of the first aspect.

[0065] The seventh aspect of an embodiment of the present application provides a third device, including: a processor, the processor is coupled to a memory, the memory is used to store programs or instructions, when the program or instructions are executed by the processor, the third device implements the method in any possible implementation manner of the above-mentioned second aspect.

[0066] An eighth aspect of an embodiment of the present application provides a communication system, which includes the first unit in the above-mentioned first aspect and the second unit in the first aspect, or the communication system includes the first unit in the above-mentioned first aspect, the second unit in the first aspect and the third unit in the second aspect, or the communication system includes the first device in the above-mentioned fourth aspect and the third device in the fifth aspect.

[0067] A ninth aspect of the embodiments of the present application provides a communication system, the communication system comprising a first unit and a second unit;

[0068] The first unit is used to send first request information input by a user to the second unit;

[0069] The second unit is used to send first response information corresponding to the first request information to the first unit;

[0070] The second unit is further used to receive second request information sent by the first unit, where the second request information is used to confuse the first request information;

[0071] The first unit is further configured to receive second response information corresponding to the second request information sent by the second unit.

[0072] The tenth aspect of the embodiments of the present application provides a computer-readable medium on which a computer program or instruction is stored. When the computer program or instruction is run on a computer, the computer executes a method of any possible implementation method of any one of the first to third aspects above.

[0073] An eleventh aspect of the embodiments of the present application provides a computer program product, which, when executed on a computer, enables the computer to execute a method of any possible implementation of any one of the first to third aspects above.

[0074] A twelfth aspect of an embodiment of the present application provides a chip system, which includes at least one processor for supporting a data processing device to implement the functions involved in any possible implementation method of any one of the first to third aspects above.

[0075] In one possible design, the chip system may also include a memory for storing program instructions and data necessary for the data processing device. The chip system may be composed of a chip, or may include a chip and other discrete devices. Optionally, the chip system also includes an interface circuit, which provides program instructions and / or data for at least one processor.

[0076] Among them, the technical effects brought about by any design method in the fourth to twelfth aspects can refer to the technical effects brought about by different design methods in any one of the first to third aspects mentioned above, and will not be repeated here.

[0077] It can be seen from the above technical solution that the embodiments of the present application have the following advantages: by obtaining false request information through a trusted unit or device (first unit or third unit), the first unit can use the false request information to confuse the real request information, so that the untrusted unit cannot clearly identify the real request information, thereby improving the user's privacy protection. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1A A schematic diagram of a system architecture provided for an embodiment of the present application;

[0079] Figure 1B Another schematic diagram of the system architecture provided in the embodiment of the present application;

[0080] Figure 1C Another schematic diagram of the system architecture provided in the embodiment of the present application;

[0081] Figure 2 A flowchart of a data processing method provided in an embodiment of the present application;

[0082] Figure 3A An example diagram of a user interface displayed by the first unit provided in an embodiment of the present application;

[0083] Figure 3B Another example diagram of a user interface displayed by the first unit provided in an embodiment of the present application;

[0084] Figure 3C Another example diagram of a user interface displayed by the first unit provided in an embodiment of the present application;

[0085] Figure 4A Another example diagram of a user interface displayed by the first unit provided in an embodiment of the present application;

[0086] Figure 4B Another example diagram of a user interface displayed by the first unit provided in an embodiment of the present application;

[0087] Figure 5Another flowchart of the data processing method provided in the embodiment of the present application;

[0088] Figure 6 Another flowchart of the data processing method provided in the embodiment of the present application;

[0089] Figure 7 Another flowchart of the data processing method provided in the embodiment of the present application;

[0090] Figure 8 Another flowchart of the data processing method provided in the embodiment of the present application;

[0091] Fig. 9 Another flowchart of the data processing method provided in the embodiment of the present application;

[0092] Fig.10 Another flowchart of the data processing method provided in the embodiment of the present application;

[0093] Fig.11 Another flowchart of the data processing method provided in the embodiment of the present application;

[0094] Fig.12 A structural schematic diagram of a data processing device provided in an embodiment of the present application;

[0095] Fig.13 A structural schematic diagram of a first device provided in an embodiment of the present application;

[0096] Fig.14 Another structural schematic diagram of a data processing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0097] The embodiments of the present application provide a data processing method and related equipment, which can obtain false request information based on the real request information input by the user or the real response information corresponding to the real request information, use the false request information to confuse the real request information, and then use the false request information to confuse the model, thereby improving the user's privacy protection.

[0098] To facilitate understanding, the relevant terms and concepts mainly involved in the embodiments of the present application are first introduced below.

[0099] 1. Neural network (also called model)

[0100] A neural network can be composed of neural units. A neural unit can be a s is the weight, b is the bias operation unit, and the output of the operation unit can be:

[0101]

[0102] Where, s = 1, 2, ... n, n is a natural number greater than 1, W s For X s The weight of the neural unit, b is the bias of the neural unit. f is the activation function of the neural unit, which is used to introduce nonlinear characteristics into the neural network to convert the input signal in the neural unit into the output signal. The output signal of the activation function can be used as the input of the next convolutional layer. The activation function can be a sigmoid function. A neural network is a network formed by connecting many of the above-mentioned single neural units together, that is, the output of one neural unit can be the input of another neural unit. The input of each neural unit can be connected to the local receptive field of the previous layer to extract the features of the local receptive field. The local receptive field can be an area composed of several neural units.

[0103] 2. Loss Function

[0104] In the process of training a deep neural network, because we hope that the output of the deep neural network is as close as possible to the value we really want to predict, we can compare the predicted value of the current network with the target value we really want, and then update the weight vector of each layer of the neural network according to the difference between the two (of course, there is usually an initialization process before the first update, that is, pre-configuring parameters for each layer in the deep neural network). For example, if the predicted value of the network is high, adjust the weight vector to make it predict lower, and keep adjusting until the neural network can predict the target value we really want. Therefore, it is necessary to pre-define "how to compare the difference between the predicted value and the target value", which is the loss function or objective function, which are important equations used to measure the difference between the predicted value and the target value. Among them, taking the loss function as an example, the higher the output value (loss) of the loss function, the greater the difference, so the training of the deep neural network becomes a process of minimizing this loss as much as possible.

[0105] 3. Trusted units / devices and untrusted units / devices

[0106] In this application, "trusted" and "untrusted" are relative concepts. "Trusted unit / device" means a unit / device that the user trusts. This application assumes that a "trusted unit / device" will not leak the user's privacy or cause privacy risks to the user after learning the user's privacy. Among them, "trusted" can be set by the user, or it can be determined through third-party inspections, etc., or it can be a unit / device that the user has used for a long time. In contrast, "untrusted unit / device" means a unit / device that the user does not trust. This application assumes that an "untrusted unit / device" may leak the user's privacy or there is a risk of privacy being leaked after learning the user's privacy. Among them, "untrusted" can be set by the user, or it can be determined through third-party inspections, etc., or it can be a unit / device that the user has not used.

[0107] See attached Figure 1A The embodiment of the present invention provides a system architecture, which includes: a user 101, a first unit 102 and a second unit 103.

[0108] The user 101 can be understood as the user of the first unit 102, and the user 101 can operate the first unit 102 by means of voice / touch gestures and other operations. When the user 101 uses the first unit 102, the first unit 102 supports independently providing services for the user 101, and can also support joint processing with other devices (such as the third unit 103) to provide services for the user 101.

[0109] The first unit 102 is used to provide services for the user 101. The first unit 102 may be equipped with an operating system and loaded with applications according to the operation of the user 101. The first unit 101 may support at least one of the following functions: installing applications, mobilizing cloud services (such as public clouds, private clouds, etc.), built-in services, etc.

[0110] The second unit 103 stores a processing module, which can process data to obtain a processing result. Alternatively, it can be understood that the second unit 103 has a data processing function, and the data processing function can also be understood as processing user requests (such as question requests, classification requests, segmentation requests, detection requests, or image generation requests, etc.). The processing module can use machine learning and other methods to implement the above-mentioned data processing function. Among them, machine learning includes neural networks, expert systems, symbolic reasoning, decision trees, support vector machines, etc.

[0111] For the convenience of description, the following description is given by taking the example of the processing module using a neural network to perform data processing requests. It can be understood that in actual applications, the processing module in the second unit can also use other machine learning methods to implement data processing functions, which are not specifically limited here.

[0112] Optionally, the user 101 inputs request information to the first unit 102, and the first unit 102 sends the request information to the second unit 103. The second unit 103 uses a neural network to process the request information to obtain response information, and returns the response information to the first unit 102. Thus, the first unit 102 can display the response information to the user 101.

[0113] In the case where the neural network stored in the second unit 103 is a large language model (LLM), since the model itself has very powerful reasoning capabilities, but is limited by factors such as training data, in order to better obtain effective reasoning results, we need to perform a series of processing on the input request information (i.e., the input data of the neural network), and these processing methods include but are not limited to:

[0114] 1) Introducing external knowledge (search, knowledge base): Retrieve and ask questions with the help of search engines, vector knowledge bases, etc., and form a prompt word input model with the question content to obtain response information, thereby improving the reliability of professional knowledge in the answer;

[0115] 2) Request information decomposition / thinking chain: For request information that is too long and too complex, considering that large models cannot directly process such request information that requires strong reasoning capabilities, we can reasonably decompose the request information and break down complex request information into a series of simpler request information combinations and let the model solve them step by step;

[0116] 3) Apply new templates to control the response information method: For some request information, we can use the characteristics of the large model to embed the request information into a specific template and require the model output to meet our specific needs.

[0117] We collectively call the above processing methods prompt engineering, which is one of the best ways to use the language big model to solve request information and obtain effective response information.

[0118] Since large language models have very high requirements for inference computing power, most users currently use models mounted on cloud services (i.e. Figure 1A The neural network in the second unit 103 shown in FIG. 1 is used to complete the use of the large model, which results in the user's request information and the corresponding response information actually being transmitted out of the user's local computer (i.e. Figure 1A This will bring about a series of problems and hidden dangers, including but not limited to:

[0119] 1) User privacy data is used for model training. The training of large models depends on high-quality corpus, and many commercial models currently use user request and response information as training corpus, which is internalized into the model capabilities. This causes user data to be stored on the service provider side for a long time, bringing unknown risks;

[0120] 2) User privacy data is leaked in the model. As mentioned in 1), user data is used for training. Therefore, during the inference process, there is a risk of user privacy data being leaked due to training at other users. Attackers can steal or reconstruct LLM training data through model extraction, model inversion, or member inference techniques.

[0121] 3) User portraits based on user request information. Based on information such as user historical requests, service providers can build user portraits to provide a series of services such as advertising push.

[0122] The above behaviors will have a great impact on the privacy and security of user data. Due to the flourishing of large models, it is difficult for users to distinguish which model services are safe and which are dangerous. Therefore, we need some technologies to enable us to use the model capabilities normally while minimizing the leakage of user privacy.

[0123] In the process of using large model inference services, untrusted service providers (i.e. Figure 1A The second unit 103 shown in the figure can fully obtain our questions and answers. Using these answers, whether for training its own model or for other purposes, will seriously infringe the user's privacy. Based on this assumption, we need a method to prevent untrusted service providers from obtaining and using the user's real request information and response information, while not affecting the user's acquisition of the response information they want on the user side.

[0124] Existing technologies mainly focus on how to protect the server-side model to avoid being contaminated by poisoned data from malicious attacks; or to avoid privacy leakage to a certain extent by replacing the content of user request information at the expense of the accuracy of the request information. These are either based on the trustworthiness of the service provider or have a significant impact on the performance of the response information of the request information. Therefore, a method to prevent data from being used for other purposes and to ensure that the original response information is obtained is necessary.

[0125] The following examples illustrate the existing problems of several prior arts.

[0126] Example 1: The paper "Privacy-Preserving Prompt Tuning for Large Language Model Services" proposes a method called RAPT (privacy-preserving prompt tuning), which uses local differential privacy and user-specific tuning on the server to prevent intruders from obtaining private information locally and in the cloud through embedding invertion attacks and attribute inference attacks.

[0127] The disadvantages of this method are that removing information may reduce the accuracy on the differential privacy side, and the framework assumes that the service provider is reliable, which cannot play a role in privacy protection in our setting.

[0128] Example 2: Replace the user's privacy and other sensitive information with common entity information (such as name, birthday, etc.) before asking questions to avoid privacy leakage.

[0129] The advantage of this method is that it can indeed avoid privacy leakage, but too much entity masking will greatly reduce the accuracy (for example, when providing birthdays to calculate certain dates, masking the birthday will make it completely impossible to process).

[0130] In order to solve the above technical problems, a trusted unit or device obtains false request information based on the real request information input by the user or the real response information corresponding to the real request information, and uses the false request information to confuse the real request information, thereby improving the user's privacy protection.

[0131] First, the system architecture applicable to the method provided in the embodiment of the present application is described.

[0132] The first one, Figure 1A The system architecture is shown.

[0133] exist Figure 1A In the system architecture shown, if the first unit 102 is a trusted unit, the real request information input by the user 101 can be processed by the first unit 102 to obtain multiple false request information, and the real request information and the multiple false request information are sent to the second unit 103. The second unit 103 infers the real request information and the multiple false request information through a large model to obtain multiple response information.

[0134] In this case, for the second unit 103, the request information received by the second unit 103 includes multiple false request information in addition to the real request information, so that the second unit 103 cannot clearly determine which request information received is the real request information, thereby improving user privacy protection.

[0135] The second type, Figure 1B The system architecture is shown.

[0136] exist Figure 1B In the system architecture shown, the system architecture includes: user 101, first unit 104, third unit 105 and second unit 106. Third unit 105 cannot communicate directly with second unit 106, and needs to use first unit 104 as an intermediary. Figure 1A The system architecture is increased with a trusted unit: the third unit 105. The first unit 104 obtains the real request information input by the user 101, and sends the real request information to the third unit 105. The third unit 105 processes the real request information to obtain multiple false request information, and returns the multiple false request information to the first unit 104. Thus, the first unit 104 sends the real request information and the multiple false request information to the second unit 106, and the second unit 106 infers the real request information and the multiple false request information through the large model to obtain multiple response information, and returns the multiple response information to the first unit 104. The first unit 104 can directly display multiple response information or real response information to the user 101, or perform certain processing (such as screening, rendering, format conversion, etc.) on the multiple response information or the real response information before presenting it to the user 101.

[0137] In this case, the trusted third unit 105 obfuscates the real request information. For the second unit 106, the request information received by the second unit 106 includes not only the real request information but also multiple false request information. Therefore, the second unit 106 cannot clearly determine which request information received is the real request information, thereby improving the user's privacy protection.

[0138] The third type, Figure 1C The system architecture is shown.

[0139] exist Figure 1C In the system architecture shown, the system architecture includes: user 101, third unit 107, first unit 108 and second unit 109. And the first unit 108 and the second unit 109 are in the same cloud (such as Figure 1CAs shown). In this case, the third unit 107 obtains the real request information input by the user, and sends the real request information to the first unit 108. After the first unit 108 processes the real request information to obtain multiple false request information, it can directly send the real request information and the multiple false request information to the second unit 109. The second unit 109 infers the real request information and the multiple false request information through the large model to obtain multiple response information, and then returns the multiple response information to the first unit 108. The first unit 108 can directly return the multiple response information to the third unit 107. The third unit 107 can directly present multiple response information to the user 101, or filter out the real response information from the multiple response information and present it to the user 101. Of course, the first unit 108 can also filter out the real response information from the multiple response information, and send the real response information to the third unit 107, so that the third unit 107 presents the real response information to the user 101. That is, the step of filtering the real response information can be performed by the third unit 107, or by the first unit 108, and the specific details are not limited here.

[0140] It is understandable that the above-mentioned request information or data may refer to text, images, audio, video, etc. Figures 1A to 1C The number of users and devices in the system architecture shown is not limited, that is, the number of users, first units and third units can be greater. Figures 1A to 1C The system architecture is only an example. In practical applications, the method provided in the embodiment of the present application can also be applied to other architectures, which are not specifically limited here.

[0141] Understandably, Figures 1A to 1C The units in the system architecture shown are only examples of units that do not belong to different physical devices. In actual applications, the above Figures 1A to 1C Each unit in the system architecture shown may belong to an independent physical device, or two units may belong to the same physical device, etc., which is not limited here. For example, the first unit is deployed on the first device, the second unit is deployed on the second device, and the third unit is deployed on the third device. For another example, the first unit is deployed on the first device, and the second unit and the third unit are deployed on the same physical device or platform. For another example, the second unit is a plug-in of the first device (such as a pluggable component such as a U shield or a U disk).

[0142] In the present application embodiment Figure 1A and Figure 1B The first unit in the system architecture shown or Figure 1CThe third unit in the system architecture shown can be a terminal device or a question answering device that provides services for the user 101. The terminal device can be deployed on land, including indoors or outdoors, handheld, wearable or vehicle-mounted; it can also be deployed on the water surface (such as a ship, etc.); it can also be deployed in the air (such as an airplane, a balloon, and a satellite, etc.). The terminal device can be a mobile phone (such as a folding mobile phone, a non-folding mobile phone), a tablet computer (Pad), a computer with wireless transceiver function, a virtual reality (VR) terminal device, an augmented reality (AR) terminal device, a wireless terminal in industrial control, a vehicle-mounted terminal device, a wireless terminal in self-driving, a wireless terminal in remote medical, a wireless terminal in a smart grid, a wireless terminal in transportation safety, a wireless terminal in a smart city, a wireless terminal in a smart home, a wearable terminal device, etc. The embodiments of the present application do not limit the application scenarios. Terminal equipment may also be sometimes referred to as terminal, user equipment (UE), access terminal equipment, vehicle-mounted terminal, industrial control terminal, UE unit, UE station, mobile station, mobile station, remote station, remote terminal equipment, mobile equipment, UE agent or UE device, etc. Terminal equipment may also be fixed or mobile.

[0143] The second unit in the embodiment of the present application is an untrusted unit, and can be a device or server that processes data through a neural network. For example, the second unit can specifically be a cloud server, a network server, an application server, a management server, a wireless controller, or other device or server with data processing capabilities.

[0144] The first unit and the third unit in the embodiment of the present application are trusted units, which can be devices or servers that confuse the real request information. Figure 1B The third unit in the system architecture shown may specifically be an agent, a cloud server, a network server, an application server, a management server, a wireless controller, or other device or server that obfuscates real request information.

[0145] In combination with the above Figures 1A to 1C The data processing method provided in the embodiment of the present application is described with reference to the system architecture.

[0146] See also Figure 2, a flow chart of a data processing method provided by an embodiment of the present application, the method may include steps 201 to 209. Steps 201 to 209 are described in detail below. Figure 2 The method shown can be applied to Figures 1A to 1C In any of the system architectures shown.

[0147] Step 201: The first unit receives first request information input by a user.

[0148] The first unit receives the first request information input by the user. The first request information in the embodiment of the present application can also be understood as the request information input by the user, which can be specifically a question (also referred to as a question) input / selected by the user, a text processing request (for example, a summary analysis request) input / selected by the user, a translation request input / selected by the user, etc., which is not specifically limited here. In addition, the carrier mode of the request information (for example, the first request information, the second request information, etc.) in the embodiment of the present application can be text, image, voice, etc., which is not specifically limited here.

[0149] For ease of understanding, the first request information in the embodiment of the present application can be understood as real request information, the second request information can also be understood as false request information, the first response information can also be understood as real response information, and the second response information can also be understood as false response information.

[0150] In the embodiment of the present application, there are multiple ways for the first unit to obtain the first request information, which may be a way for the user to directly input the first request information to the first unit (such as keyboard input, voice input, etc.), or a way for the user to select in the interface displayed by the first unit (such as selection, typing input, etc.), or a way to select from a database, or a way to receive forwarding from other devices (such as Figure 1C In the process, the first unit may receive the first request information forwarded by the third unit), etc., and the specific details are not limited here.

[0151] In a possible implementation manner, the user may directly input the first request information to the first unit through an external device (such as a USB flash drive, a keyboard, a mouse, etc.).

[0152] In another possible implementation, the first unit may determine the first request information through user operation. For example, the first unit displays an interface to the user, the interface includes multiple options, and then determines the first request information based on the user's operation on the multiple options (such as a click selection operation, a voice operation, etc.).

[0153] Example 1: For example, the first request information is a question, and the carrier is text. The first request information is: What are the symptoms of stomach disease? In this example, the first unit obtains the first request information as follows: FIG. 3A to FIG. 3C The first unit first displays to the user Figure 3A The first interface shown in FIG. 1 is a first interface, and the 301 area in the first interface is for the user to input the request information. After the user enters the request information in the 301 area, as shown in FIG. Figure 3B As shown, the user can click Send 302 to transmit the request information to the first unit. After receiving the request information input by the user, the first unit can display the following Figure 3C The second interface shown. The second interface includes the request information input by the user window and the prompts in the assistant window during processing. It can be understood that Figure 3C The assistant window may not include prompts during processing, and there is no specific limitation here.

[0154] Example 2: The first request information is: How should I lose weight if I weigh 200 pounds?

[0155] Step 202: The first unit sends first request information to the second unit.

[0156] After the first unit obtains the first request information, it may send the first request information to the second unit. Correspondingly, the second unit receives the first request information sent by the first unit.

[0157] Step 203: The second unit obtains first response information through the neural network.

[0158] After the second unit obtains the first request information, it obtains the first response information through the neural network. For example, if the first request information is a question, the first response information can also be understood as an answer.

[0159] For ease of understanding, the above-mentioned first response information can be understood as true response information.

[0160] The neural network in the embodiments of the present application can be a convolutional neural network (CNN), a feedforward neural network (FNN), a recursive neural network (RNN) (such as a long short-term memory network, a gate-controlled recurrent unit, an attention network, etc.), Transformers, a generative adversarial network (GAN), LLM, etc. The embodiments of the present application do not limit the specific structure, type, and training method of the neural network.

[0161] Optionally, the second unit inputs the first request information into the neural network to obtain first response information.

[0162] Step 204: The second unit sends a first response message to the first unit.

[0163] After obtaining the first response information, the second unit sends the first response information to the first unit. Correspondingly, the first unit receives the first response information sent by the second unit.

[0164] Step 205: The first unit obtains second request information based on the first request information or the first response information.

[0165] In the embodiment of the present application, the number of the second request information may be one or more. It is understandable that in order to improve the obfuscation effect, the number of the second request information is usually multiple, and the following description will also be taken as an example of multiple second request information.

[0166] The second request information in the embodiment of the present application is used to confuse the first request information, or it can be understood that the first request information is the real request information input by the user, and the second request information is a false request information constructed based on the real request information or the real response information corresponding to the real request information.

[0167] Optionally, the specific situation of the second request information is similar to the first request information. The second request information can specifically be a question (also called a problem), a text processing request (for example, a summary and analysis request), a translation request, etc., which is not specifically limited here.

[0168] Generally speaking, the task of the second request information is the same or similar to that of the first request information, or there is an overlap between the second request information and the first request information. For example, the first request information is: What are the symptoms of stomach disease? The second request information includes what are the symptoms of liver disease? The tasks of the first request information in this example are similar to those of the second request information (for example, the task is to ask about symptoms). For another example, the first request information is: What are the symptoms of stomach disease? The second request information includes which hospitals are suitable for treating stomach disease? The first request information in this example is similar to the overlapping parts of the second request information (for example, both are questions about stomach disease).

[0169] There are multiple situations in this step, which are described below.

[0170] In a first method, the first unit obtains a plurality of second request information based on the first request information.

[0171] This situation can be understood as, in this embodiment, the first request information and multiple second request information are carried in the same message, and the first response information and multiple second response information are carried in the same message. That is, it can be understood that the first unit can send the first request information and multiple second request information to the second unit at the same time. The second unit can send the first response information and multiple second response information to the first unit at the same time.

[0172] In an embodiment of the present application, the first unit may obtain multiple second request information based on the first request information and the keyword. The first unit may also obtain multiple second request information based on the first request information and a preset rule. The first unit may also send the first request information to other units and receive multiple second request information fed back by other units. The first unit may also obtain multiple second request information based on the first request information and user operations, etc., which are not limited here. The following is an exemplary description taking the example of the first unit obtaining multiple second request information based on the first request information and the keyword.

[0173] In the embodiment of the present application, there are multiple ways for the first unit to determine the keyword, which may be determined by a selection operation of the user, or determined by a first preset rule, or by other devices forwarding the keyword determined by the user, or by sending the first request information to other devices (such as the third unit, etc.), and the other devices determining the keyword and returning it to the first unit (such as Figure 1C In the example, the first unit may receive keywords forwarded by the third unit), etc., and the specific details are not limited here.

[0174] Among them, the above-mentioned first preset rule may be related to at least one of the following: name, address, place of origin, job, company name, number, date, sentence components (such as subject, predicate, object, attributive, etc.), units of magnitude, nouns, verbs, adjectives, adverbs indicating degree, color, sports, disease names, etc., which involve user privacy, and the specific words are not limited here.

[0175] Optionally, the keyword is a word in the first request information (in this case, it may also be called the first keyword). After the first unit obtains the first request information, it may first determine the first keyword in the first request information, and then obtain multiple second request information based on the first keyword and the first request information.

[0176] For example, continuing the above FIG. 3A to FIG. 3C For example, the first keyword is a word in the first request information, such as Figure 4A After the user inputs the request information, the first unit displays the third interface, the third page includes a user window and an assistant window, and the assistant window is used to determine the first keyword in the first request information through interaction with the user. Figure 4A The assistant window shows "Please indicate the first keyword in the request information by clicking or entering a sequence number, etc.: 1. stomach disease 2. symptoms...". Further, Figure 4B As shown, the first unit responds to the user's click operation 401 to determine that the first keyword in the first request information includes "stomach disease" and "symptoms" (not shown in the figure).

[0177] In addition, in the embodiment of the present application, there is no limitation on the number of first keywords, which can be one or more.

[0178] In the embodiment of the present application, there are multiple ways for the first unit to obtain the multiple second request information based on the first request information and the first keyword. The method may be to obtain the multiple second request information through user operation, or to obtain the multiple second request information through a second preset rule, or to obtain the multiple second request information by sending the first request information and the first keyword to other devices (such as the third unit, etc.), and the other devices determine the multiple second request information based on the first request information and the first keyword and return the multiple second request information to the first unit (for example, Figure 1C In the embodiment, the first unit sends a first request message and a first keyword to the third unit, the third unit may determine a plurality of second request messages based on the first request message and the first keyword, and send the plurality of second request messages to the first unit), etc., and the specific details are not limited here.

[0179] In a possible implementation, the first unit first determines a plurality of first replacement characters based on a second preset rule and a first keyword, and then replaces the first keyword in the first request information with the plurality of first replacement characters to obtain a plurality of second request information. The second preset rule is related to at least one of the following: synonyms, similar words, scrambling, multiplying by a random number (for numbers), replacement characters related to the first keyword, antonyms, etc.

[0180] Since the purpose of the embodiment of the present application is to obfuscate the real request information, in addition to the above-mentioned second preset rule, the obfuscation method can also be table lookup, model generation, manual input, analogy reasoning, mathematical solution, etc.

[0181] For example, if the first request information involves a digital sequence such as an identification number, a bank card number, or a postal code, and the first keyword includes a digital sequence, determining multiple first replacement characters based on the second preset rule and the first keyword may include: randomly shuffling the numbers in the digital sequence or multiplying them by a random number to obtain the first replacement characters. For another example, if the first keyword includes a user's name, address, or place of origin, determining multiple first replacement characters based on the second preset rule and the first keyword may include: randomly replacing characters in the first keyword (modifying the surname, address, or place of origin, etc.).

[0182] Exemplarily, continuing with Example 1 above, the first request information is: What are the symptoms of stomach disease? The first keywords include: "stomach disease" and "symptoms". The first unit determines that the replacement words for "stomach disease" include "liver disease" and "kidney disease" through similar words. The first unit determines that the replacement words for "symptoms" include "treatment methods" and "hospital for treatment" through replacement words related to the first keyword. The multiple second request information obtained by bringing the replacement words into the first request information to replace the corresponding first keywords include: "What are the symptoms of kidney disease?", "What are the symptoms of liver disease?", "What are the treatment methods for liver disease?" and "What are the hospitals for stomach disease?".

[0183] In another possible implementation, the first unit sends the first request information and the first keyword to the third unit, and the third unit generates multiple second request information based on the first request information and the first keyword; the first unit receives the multiple second request information sent by the third unit. Of course, the first unit can also directly send the first request information to the third unit, and the third unit determines the first keyword of the first request information, obtains multiple second request information based on the first keyword, and returns the multiple second request information to the first unit.

[0184] Second, the first unit obtains multiple second request information based on the first response information.

[0185] This situation can be understood as, in this embodiment, the first request information and the multiple second request information are carried in different messages, and the first response information and the multiple second response information are carried in different messages. That is, it can be understood that the first unit can first send the first request information to the second unit, receive the first response information, and then send multiple second request information to the second unit, and receive multiple second response information.

[0186] After the first unit obtains the first request information, it sends the first request information to the second unit. The second unit obtains the first response information through the neural network and sends the first response information to the first unit. The first unit then obtains multiple third response information based on the first response information, and then reversely infers multiple second request information based on the multiple third response information.

[0187] Optionally, the first unit may determine the second keyword in the first response information, and obtain multiple third response information through the second keyword in the first response information. The first unit may also send the first response information to the third unit, and the third unit generates multiple third response information based on the first response information, and returns the third response information to the first unit. The method of obtaining multiple third response information based on the second keyword is similar to the aforementioned method of obtaining multiple second request information based on the first keyword, and will not be repeated here.

[0188] Of course, the first unit can also directly infer multiple third response information to obtain multiple second request information, or send multiple third response information to the third unit, and the third unit infers multiple third response information to obtain multiple second request information, and returns the multiple second request information to the first unit. Alternatively, the first unit sends the first response information to the third unit, and the third unit obtains multiple second request information based on the first response information, and then returns the multiple second request information to the first unit.

[0189] The above-mentioned reverse deduction process can be carried out by users, text generation networks, preset rules, etc., which are not limited here. For example, take the example that the first unit obtains the second request information corresponding to the third response information through the text generation network. The text generation network can be trained with the training data as input and the value of the loss function being less than a threshold as the goal. The loss function is used to represent the difference between the output of the text generation network and the label of the training data.

[0190] Example 2, the first request information is: How should I lose weight if I weigh 200 pounds? The first response information fed back by the second unit based on the first request information is: Eat less and exercise more. The first unit or the third unit can determine that the second keywords in "eat less and exercise more" include "drink more" and "rest more" by means of homonyms and antonyms. Determine that the replacement word for "eat less" includes "drink more". And determine that the replacement word for "exercise more" includes "rest more", so that the replacement word is brought into the first response information to replace the corresponding second keyword to obtain multiple second response information including: "drink more and exercise more", "drink more and rest more" and "drink less and rest more", etc. Thus, the second response information "drink more and exercise more" is reversed to obtain the second request information "What should I pay attention to for kidney stones?". The first unit reverses the second response information "drink more and rest more" to obtain the second request information "How should I adjust after surgery?". The first unit reverses the second response information "drink less and rest more" to obtain the second request information "What should I do if I often drink coffee and can't sleep at night?".

[0191] Step 206: The first unit sends a plurality of second request information to the second unit.

[0192] After the first unit obtains multiple second request information, it can send multiple second request information to the second unit. Correspondingly, the second unit receives multiple second request information sent by the first unit. Multiple request information is used to confuse the first request information, so that the second unit cannot clearly identify the real request information, thereby improving user privacy.

[0193] In the embodiments of the present application, the use of fake request information to confuse the real request information can also be understood as using fake request information to obscure the real request information. Alternatively, it can be understood that by mixing the fake request information with the real request information, the untrusted unit cannot clearly identify the real request information in the mixed multiple request information. In this way, the real request information is protected by the fake request information, thereby reducing the privacy leakage of the user in the real request information.

[0194] Step 207: The second unit obtains a plurality of second response information through the neural network.

[0195] After the second unit obtains the multiple second request information, it obtains the multiple second response information through the neural network, and the multiple second response information corresponds to the multiple second request information.

[0196] For ease of understanding, the above-mentioned multiple second response messages can be understood as multiple false response messages.

[0197] The neural network in the embodiments of the present application can be a convolutional neural network (CNN), a feedforward neural network (FNN), a recursive neural network (RNN) (such as a long short-term memory network, a gate-controlled recurrent unit, an attention network, etc.), Transformers, a generative adversarial network (GAN), LLM, etc. The embodiments of the present application do not limit the specific structure, type, and training method of the neural network.

[0198] Optionally, the second unit inputs multiple second request information into the neural network to obtain multiple second request information. It is understandable that the second request information can be input multiple times or once, and the specific details are not limited here.

[0199] Step 208: The second unit sends a plurality of second response messages to the first unit.

[0200] After acquiring the plurality of second response information, the second unit sends the plurality of second response information to the first unit. Correspondingly, the first unit receives the plurality of second response information sent by the second unit.

[0201] Step 209: The first unit presents response information.

[0202] After receiving the first response information and multiple second response information sent by the second unit, the first unit may present response information to the user, where the response information includes the first response information, or includes multiple second response information and the first response information.

[0203] In a possible implementation, the first unit directly presents a plurality of response information to the user, where the plurality of response information includes a first response information and a plurality of second response information.

[0204] Optionally, the first unit may also present indication information to the user, where the indication information is used to indicate the position of the first response information in the plurality of response information, so that the user can clearly identify the real response information.

[0205] It is understandable that there are multiple ways for the first unit to determine the position of the correct response information in multiple response information. It can be that when the first unit sends the first request information and multiple second request information to the second unit, the position of the first request information is recorded, and then the position of the first response information in the multiple response information can be determined. It can also be that when the first unit sends the first request information and multiple second request information to the second unit, each request information can be attached with an identifier to facilitate the subsequent determination of the position of the correct response information based on the identifier, etc., which is not limited here.

[0206] In another possible implementation, the first unit performs some processing (such as screening, rendering, format conversion, etc.) on the multiple response information before presenting them to the user.

[0207] In addition, the presentation in the embodiment of the present application may include at least one of the following: display, voice playback, etc. For example, the first unit may display the response information to the user, or may present the response information to the user by voice broadcast, etc. The present application does not limit the specific presentation method.

[0208] It should be noted that there is no time sequence restriction for each step in the embodiments of the present application, which is illustrated below with examples.

[0209] In one possible implementation, step 202 is the same step as step 206 (or understood as the first request information in step 202 and the multiple second request information in step 206 are carried in the same message), step 203 is the same step as step 207, and step 204 is the same step as step 208 (or understood as the first response information in step 204 and the multiple second response information in step 208 are carried in the same message). That is, it can be understood that the first unit can send the first request information and the multiple second request information to the second unit at the same time. The second unit can send the first response information and the multiple second response information to the first unit at the same time. In this case Figure 2 The illustrated embodiment can be described as Figure 5 The embodiment shown. Figure 2The method of the illustrated embodiment may be equivalent to the following steps 501 to 506. Step 501, the first unit receives the first request information input by the user. Step 502, the first unit obtains multiple second request information based on the first request information. Step 503, the first unit sends the first request information and multiple second request information to the second unit. Step 504, the second unit obtains multiple response information through the neural network. Step 505, the second unit sends multiple response information to the first unit. Step 506, the first unit presents the response information. Figure 5 The detailed description can be found above and will not be repeated here.

[0210] In another possible implementation, step 202 and step 206 are different steps (or understood as the first request information in step 202 and the multiple second request information in step 206 are carried in different messages), step 203 and step 207 are different steps, and step 204 and step 208 are different steps (or understood as the first response information in step 204 and the multiple second response information in step 208 are carried in different messages). That is, it can be understood that the first unit can first send a first request message to the second unit and receive a first response message. Then send multiple second request messages to the second unit and receive multiple second response messages. In this case, Figure 2 The illustrated embodiment can be described as Figure 6 or Figure 7 The embodiment shown. Figure 2 The method of the illustrated embodiment may be equivalent to the following Figure 6 or Figure 7 The embodiment shown.

[0211] In an embodiment of the present application, on the one hand, multiple false request information corresponding to the real request information is obtained through a trusted unit (the first unit or the third unit), and the first unit can use multiple false request information to confuse the real request information, so that the untrusted unit cannot clearly identify the real request information, thereby improving the privacy protection of the user. On the other hand, the real answer is first obtained through the real request information, and then the real answer is confused to obtain multiple false answers, so that the multiple false answers can be reversed to obtain multiple false request information. Because in general, the second unit retrieves the log and other methods within a certain time period, as long as the second unit infers that the time interval between the real request information and the false request information is less than the above time period. In the embodiment of the log, the first unit can confuse the real request information with multiple false request information, thereby improving the privacy protection of the user.

[0212] It is understood that in one possible implementation, Figure 2 The embodiment shown is applicable to the aforementioned Figure 1A In the scenario shown, Figure 2The first unit in the embodiment shown refers to the aforementioned Figure 1A The first unit in Figure 2 The second unit in the embodiment shown refers to the aforementioned Figure 1A The second unit in.

[0213] In another possible implementation, Figure 2 The embodiment shown is applicable to the aforementioned Figure 1B In the scenario shown, Figure 2 The first unit in the embodiment shown refers to the aforementioned Figure 1B The first unit in Figure 2 The second unit in the embodiment shown refers to the aforementioned Figure 1B The second unit in.

[0214] In another possible implementation, Figure 2 The embodiment shown is applicable to the aforementioned Figure 1C In the scenario shown, Figure 2 The first unit in the embodiment shown refers to the aforementioned Figure 1C The first unit in Figure 2 The second unit in the embodiment shown refers to the aforementioned Figure 1C In this case, Figure 2 In the embodiment shown, the first unit obtains the first request information by receiving the first request information sent by other devices (such as the third unit), and Figure 2 After obtaining a plurality of response information, the first unit in the illustrated embodiment sends a plurality of response information or a first response information to other devices (eg, the third unit).

[0215] See also Figure 6 , another flow chart of the data processing method provided by the embodiment of the present application, the method may include steps 601 to 609. Steps 601 to 609 are described in detail below. Figure 6 The method shown can be applied to Figure 1B The system architecture shown.

[0216] Step 601: The first unit receives first request information input by a user.

[0217] Step 601 in this embodiment may refer to the aforementioned Figure 2 The description of step 201 in the illustrated embodiment will not be repeated here.

[0218] Step 602: The first unit sends first request information to the third unit.

[0219] After obtaining the first request information, the first unit sends the first request information to the third unit. Correspondingly, the third unit receives the first request information sent by the first unit.

[0220] Step 603: The third unit determines the first keyword in the first request information.

[0221] Step 604: The third unit obtains a plurality of second request information based on the first keyword and the first request information.

[0222] Step 603 and step 604 in this embodiment may refer to the aforementioned Figure 2 The description of step 205 in the illustrated embodiment will not be repeated here.

[0223] Step 605: The third unit sends a plurality of second request information to the first unit.

[0224] After acquiring the multiple second request information, the third unit may send the multiple second request information to the first unit. Correspondingly, the first unit receives the multiple second request information sent by the third unit.

[0225] It is understandable that the third unit may also send the first request information and multiple second request information to the first unit. Correspondingly, the first unit receives the first request information and multiple second request information sent by the third unit.

[0226] Step 606: The first unit sends a first request message and a plurality of second request messages to the second unit.

[0227] Step 607: The second unit obtains multiple response information through the neural network.

[0228] Step 608: The second unit sends a plurality of response messages to the first unit.

[0229] Step 609: The first unit presents response information.

[0230] Steps 606 to 609 in this embodiment may refer to the aforementioned Figure 2 The description of step 202, step 204, step 206, step 208 and step 209 in the illustrated embodiment will not be repeated here.

[0231] In an embodiment of the present application, multiple false request information corresponding to the real request information is obtained through a third-party device (i.e., the third unit) of the trusted unit. The first unit can use multiple false request information to confuse the real request information, so that the untrusted unit cannot clearly identify the real request information, thereby improving the user's privacy protection.

[0232] See also Figure 7 , another flow chart of the data processing method provided by the embodiment of the present application, the method may include steps 701 to 709. Steps 701 to 709 are described in detail below. Figure 7 The method shown can be applied to Figure 1C The system architecture shown.

[0233] Step 701: The first unit receives first request information input by a user.

[0234] Step 701 in this embodiment may refer to the aforementioned Figure 2 The description of step 201 in the illustrated embodiment will not be repeated here.

[0235] Step 702: The first unit sends first request information to the third unit.

[0236] After obtaining the first request information, the first unit sends the first request information to the third unit. Correspondingly, the third unit receives the first request information sent by the first unit.

[0237] Step 703: The third unit determines the first keyword in the first request information.

[0238] Step 704: The third unit obtains a plurality of second request information based on the first keyword and the first request information.

[0239] Step 703 and step 704 in this embodiment may refer to the aforementioned Figure 2 The description of step 205 in the illustrated embodiment will not be repeated here.

[0240] Step 705: The third unit sends a first request message and a plurality of second request messages to the second unit.

[0241] After the third unit obtains the multiple second request information, it can send the first request information and the multiple second request information to the second unit. Correspondingly, the first unit receives the multiple second request information sent by the third unit.

[0242] The third unit in this embodiment can communicate directly with the second unit.

[0243] Step 706: The second unit obtains multiple response information through the neural network.

[0244] Step 706 in this embodiment may refer to the aforementioned Figure 2 The description of step 203 and step 207 in the illustrated embodiment will not be repeated here.

[0245] Step 707: The second unit sends multiple response messages to the third unit.

[0246] After the second unit obtains the multiple response information, it can send the multiple response information to the third unit. Correspondingly, the third unit receives the multiple response information sent by the second unit.

[0247] Step 708: The third unit sends a plurality of response messages or a first response message to the first unit.

[0248] After receiving the multiple response messages sent by the second unit, the third unit may send multiple response messages or the first response message to the first unit. Correspondingly, the first unit receives the multiple response messages or the first response message sent by the third unit.

[0249] In a possible implementation manner, after obtaining the multiple response information, the third unit directly forwards the multiple response information to the first unit.

[0250] In another possible implementation, after obtaining the multiple response information, the third unit performs certain processing (such as screening, rendering, format conversion, etc.) on the multiple response information, and then sends the processed response information to the first unit.

[0251] Step 709: The first unit presents response information.

[0252] Step 709 in this embodiment may refer to the aforementioned Figure 2 The description of step 209 in the illustrated embodiment will not be repeated here.

[0253] In an embodiment of the present application, multiple false request information corresponding to real request information is obtained through a trusted unit or device (i.e., the third unit). The first unit can use multiple false request information to confuse the real request information, so that the untrusted unit cannot clearly identify the real request information, thereby improving the user's privacy protection.

[0254] Above Figure 6 and Figure 7 The embodiment shown describes a method for directly replacing keywords in real request information to obtain multiple fake request information. The following describes a method for replacing keywords in real response information to obtain multiple fake answers, and then reversely obtaining multiple fake request information based on the multiple fake answers.

[0255] See also Figure 8 , another flow chart of the data processing method provided by the embodiment of the present application, the method may include steps 801 to 811. Steps 801 to 811 are described in detail below. Figure 8 The method shown can be applied to Figures 1A to 1C In any of the system architectures shown.

[0256] Step 801: The first unit receives first request information input by a user.

[0257] Step 801 in this embodiment may refer to the aforementioned Figure 2 The description of step 201 in the illustrated embodiment will not be repeated here.

[0258] For example, the first request information is: How should I lose weight if I weigh 200 kilograms?

[0259] Step 802: The first unit sends first request information to the second unit.

[0260] After obtaining the first request information, the first unit sends the first request information to the second unit. Correspondingly, the second unit receives the first request information sent by the first unit.

[0261] Step 803: The second unit obtains first response information through a neural network.

[0262] After receiving the first request information, the second unit obtains the first response information through the neural network.

[0263] Optionally, the second unit inputs the first request information into the neural network to obtain first response information.

[0264] The neural network, the first request information, and the first response information in this embodiment can refer to the aforementioned Figure 2 The description in the illustrated embodiment will not be repeated here.

[0265] Exemplarily, continuing the example of the first request information above, the first response information is: eat less and exercise more.

[0266] Step 804: The second unit sends a first response message to the first unit.

[0267] After obtaining the first response information, the second unit sends the first response information to the first unit. Correspondingly, the first unit receives the first response information sent by the second unit.

[0268] Optionally, after obtaining the first response information, the first unit may present the first response information to the user.

[0269] Step 805: The first unit determines the second keyword in the first response information.

[0270] After obtaining the first response information, the first unit determines a second keyword in the first response information.

[0271] In this embodiment, there are multiple ways for the first unit to determine the second keyword in the first response information. It can be determined by the user's selection operation, by a third preset rule, or by sending the first response information to other devices (such as a third unit, etc.), and the other devices determine the second keyword and return it to the first unit, etc. The specifics are not limited here.

[0272] Among them, the third preset rule mentioned above is consistent with the above Figure 2The first preset rule used to determine the first keyword in the first request information in the illustrated embodiment is similar and is not limited here.

[0273] Exemplarily, continuing the example of the first request information and the first response information, the first unit can determine that the second keyword in "eat less and exercise more" includes "drink more" and "rest more" by means of homonyms and antonyms.

[0274] Step 806: The first unit obtains a plurality of second response information based on the second keyword and the first response information.

[0275] After the first unit obtains the keyword, it can obtain a plurality of second response information based on the keyword and the first response information.

[0276] In the embodiment of the present application, there are many ways for the first unit to obtain multiple second response information, which can be obtained through user operations, or through a fourth preset rule, or by sending the first response information and the second keyword to other devices (such as the third unit, etc.), and the other devices determine multiple second response information based on the first response information and the second keyword and return them to the first unit, etc., and the specifics are not limited here.

[0277] Among them, the fourth preset rule mentioned above is consistent with the above Figure 2 The second preset rule used to obtain multiple second request information in the illustrated embodiment is similar and is not limited here.

[0278] In a possible implementation, the first unit first determines a plurality of replacement characters based on the fourth preset rule and the second keyword; and then replaces the second keyword in the first response information with the plurality of replacement characters respectively to obtain a plurality of second response information.

[0279] Exemplarily, continuing the example of the first request information, the first response information, and the second keyword, the first unit determines that the replacement word for "eat less" includes "drink more". And determines that the replacement word for "exercise more" includes "rest more", so that the replacement words are brought into the first response information to replace the corresponding second keywords to obtain multiple second response information including: "drink more and exercise more", "drink more and rest more", and "drink less and rest more", etc.

[0280] In another possible implementation, the first unit sends the first response information and the second keyword to the third unit, and the third unit generates multiple second response information based on the first response information and the second keyword; the first unit receives the multiple second response information sent by the third unit. Of course, the first unit can also directly send the first response information to the third unit, and the third unit determines the second keyword of the first response information, obtains multiple second response information based on the second keyword, and returns the multiple second response information to the first unit.

[0281] Step 807: The first unit obtains multiple second request information based on the multiple second response information.

[0282] After the first unit obtains the multiple second response information, it can perform reverse deduction based on the multiple second response information to obtain the multiple second request information.

[0283] The above-mentioned reverse deduction process can be carried out by users, text generation networks, preset rules, etc., which are not limited here. For example, take the example that the first unit obtains the second request information corresponding to the second response information through the text generation network. The text generation network can be trained with the training data as input and the loss function value less than the threshold as the goal. The loss function is used to represent the difference between the output of the text generation network and the label of the training data.

[0284] In addition, in an embodiment of the present application, there are multiple ways for the first unit to obtain multiple second request information, which can be obtained through user operations, through the fifth preset rule, or by sending multiple second response information to other devices (such as the third unit, etc.), and the other devices determine multiple second request information based on the multiple second response information and return it to the first unit, etc. The specific methods are not limited here.

[0285] For example, continuing the example of the above-mentioned multiple second response information, the first unit reversely infers the second response information "drink more and exercise more" to obtain the second request information "what should I pay attention to for kidney stones?" The first unit reversely infers the second response information "drink more and rest more" to obtain the second request information "how should I adjust after surgery?" The first unit reversely infers the second response information "drink less and rest more" to obtain the second request information "what should I do if I often drink coffee and can't sleep at night?".

[0286] Step 808: The first unit sends a plurality of second request information to the second unit.

[0287] After acquiring the multiple second request information, the first unit sends the multiple second request information to the second unit. Correspondingly, the second unit receives the multiple second request information sent by the first unit.

[0288] Step 809: The second unit obtains a plurality of second response information through the neural network.

[0289] After the second unit obtains the plurality of second request information, it obtains the plurality of second response information through the neural network.

[0290] For example, continuing the example of the above multiple second request information, the second request information "What should I pay attention to for kidney stones?" corresponds to the second response information "Drink more water and exercise more." The second request information "How should I adjust after surgery?" corresponds to the second response information "Take medicine according to the doctor's instructions and have a good rest." The second request information "What should I do if I often drink coffee and can't sleep at night?" corresponds to the second response information "Drink less coffee and rest more."

[0291] Step 810: The second unit sends a plurality of second response messages to the first unit.

[0292] After the second unit obtains the multiple second response information, it can send the multiple second response information to the first unit. Correspondingly, the first unit receives the multiple second response information sent by the second unit.

[0293] Step 811: The first unit presents response information.

[0294] Step 811 in this embodiment may refer to the aforementioned Figure 2 The description of step 209 in the illustrated embodiment will not be repeated here.

[0295] In this embodiment, the real answer is first obtained through the real request information, and then the real answer is obfuscated to obtain multiple false answers, so that the multiple false answers can be reversed to obtain multiple false request information. In general, the second unit retrieves the log and other methods within a certain time period, as long as the time interval between the real request information and the false request information inferred by the second unit is less than the above time period. In the embodiment of the log, the first unit can confuse the real request information with multiple false request information, thereby improving the user's privacy protection.

[0296] Next, take the case where the request information is a question and the processing module implements the data processing function through LLM as an example (i.e., the request information is a question and the response information is the answer or reply corresponding to the question).

[0297] See also Fig. 9 , another flow chart of the data processing method provided in an embodiment of the present application.

[0298] Step 1: First, the user inputs a question, and then the question is decomposed into a series of sub-questions. The large model processes the data of these sub-questions separately to obtain corresponding multiple sub-answers.

[0299] Step 2: For the sub-answers we have obtained, we can look for words of the same category (such as synonyms, near-synonyms, and numbers and other information can be replaced or randomly shuffled, etc.) for each sub-answer, and then randomly combine these sub-answers to obtain multiple false answers, and reversely infer multiple false answers to obtain multiple false questions.

[0300] Step 3, confuse and verify: confuse the LLM with multiple false questions, initiate information retrieval to the LLM, and verify the degree of confusion.

[0301] The method provided in the embodiment of the present application is described.

[0302] See also Fig.10 , another flow chart of the data processing method provided by the embodiment of the present application. The method is applied to the coding processing module, which can be a program for automatically processing natural language, or other types, etc., which are not limited here.

[0303] Get the question, input it into the LLM, and give an answer. Input the question into the encoding processing module, which is located on the user side. It will first identify the words in the question that may be related to user privacy (this can be automatically identified by the encoder or manually selected by the user), modify these words (also called keywords) (including but not limited to replacing them with synonyms / similar words, randomly shuffling / changing information such as numbers, etc.), and generate a series of confusing questions. For each generated question, the LLM is required to generate an answer, and the false answer is mixed with the previous correct answer to confuse the model.

[0304] For example, the question is "What are the symptoms of stomach disease?", the keywords include "stomach disease" and "symptoms", and the generated confusing questions include: "What are the symptoms of stomach disease?", "What are the symptoms of kidney disease?", "What are the symptoms of liver disease?", "What are the treatments for liver disease?", "What are the treatments for stomach disease?". The final question set includes: "What are the treatments for liver disease?" gives three wrong answers, "What are the symptoms of stomach disease?" gives one correct answer, and "What are the hospitals for kidney disease" gives two random answers.

[0305] See also Fig.11 , another flow chart of the data processing method provided by the embodiment of the present application. The method is applied to the encoding processing module and the decoding processing module; both can be programs for automatically processing natural language, or other types, etc., which are not limited here.

[0306] The question is input into the encoding processing module, which is located on the user side. It will first identify the words in the question that may be related to user privacy (this can be automatic identification by the encoder or manual selection by the user), modify these words (including but not limited to replacing them with synonyms / similar words, randomly shuffling / changing numbers and other information, etc.), and generate a series of obfuscated questions.

[0307] For each generated question, LLM is required to generate several answers and agree on a certain position in the prompt through commands, specifying the answer at that position as the correct answer and the rest of the answers as incorrect answers; LLM performs reasoning and generates a series of answers.

[0308] The corresponding answer generated by the large model is input into the decoding processing module (which is also located on the user side). According to the above-mentioned agreed position, the correct answer to the real question is extracted and presented to the user.

[0309] For example, the question is "What are the symptoms of stomach disease?", the keywords include "stomach disease" and "symptoms", and the generated confusing questions include: "What are the symptoms of stomach disease?", "What are the symptoms of kidney disease?", "What are the symptoms of liver disease?", "What are the treatments for liver disease?", "What are the treatments for stomach disease?". The final question set includes: "What are the treatments for liver disease?" gives three wrong answers, "What are the symptoms of stomach disease?" gives one correct answer, and "What are the hospitals for kidney disease" gives two random answers.

[0310] As you can see, Fig.10 Compared with the illustrated embodiment, this method adds a decoding processing module, but the correct question and answer will no longer appear first, which is conducive to further protection of privacy.

[0311] In the embodiments of the present application, on the one hand, privacy protection is achieved by performing multiple obfuscations on questions. For example, when a user asks a question in the interactive interface, multiple questions will be asked in the privacy mode. There is no specific limitation on the way to generate obfuscated questions for better protection: random, table lookup, model generation, manual input, analogical reasoning, mathematical solution, etc. can all generate obfuscated questions. On the other hand, an algorithm flow for generating obfuscated questions through prompts - when a user asks a question in the interactive interface, a set of prompts will be attached in the privacy mode. There is no limitation on the specific text format of the prompt, and any prompt that can generate a false question is acceptable.

[0312] The present invention hides the real questions and answers among a large number of related but different questions and answers (i.e., false questions and false answers) in a confusing manner.

[0313] For untrusted LLM providers whose main purpose is to access private data, this method can circumvent the automated crawling scripts from crawling and filtering real questions and real answers, thus protecting privacy;

[0314] For untrusted LLM providers who use user Q&A data as model training data, a large number of conflicting questions and answers will reduce the accuracy of model training with opposite gradients, thereby preventing users’ private data from appearing in other inferences after training and protecting user privacy.

[0315] The above describes the data processing method in the embodiment of the present application, and the following describes the device in the embodiment of the present application. Fig.12 An embodiment of a data processing device in an embodiment of the present application includes: a transceiver unit 1201 and a processing unit 1202.

[0316] In one possible implementation, Fig.12 The data processing device shown is the aforementioned Figures 1A to 11 In the first device of the embodiment shown, the functions of each unit are as follows:

[0317] The transceiver unit 1201 is used to send first request information input by a user to the second unit;

[0318] The transceiver unit 1201 is further configured to receive first response information corresponding to the first request information sent by the second unit;

[0319] The transceiver unit 1201 is further configured to send a second request message to the second unit, where the second request message is used to confuse the first request message;

[0320] The transceiver unit 1201 is further configured to receive second response information corresponding to the second request information sent by the second unit.

[0321] Optionally, the first device further includes: a processing unit 1202, configured to obtain second request information based on the first request information.

[0322] Optionally, the processing unit 1202 is specifically configured to determine a first keyword in the first request information; the processing unit 1202 is specifically configured to obtain the second request information based on the first keyword and the first request information.

[0323] Optionally, the processing unit 1202 is specifically configured to determine a first replacement word corresponding to the first keyword; the processing unit 1202 is specifically configured to replace the first keyword in the first request information with the first replacement word to obtain the second request information.

[0324] Optionally, the processing unit 1202 is specifically configured to send the first keyword and the first request information to the third unit; the processing unit 1202 is specifically configured to receive the second request information sent by the third unit.

[0325] Optionally, the processing unit 1202 is specifically configured to determine a first keyword based on a first operation of the user or a first preset rule.

[0326] Optionally, the processing unit 1202 is specifically configured to send first request information to the third unit; the processing unit 1202 is specifically configured to receive second request information sent by the third unit.

[0327] Optionally, the first device further includes: a processing unit 1202, configured to generate second request information based on the first response information.

[0328] Optionally, the processing unit 1202 is specifically used to obtain third response information based on the first response information; the processing unit 1202 is specifically used to obtain the second request information based on the third response information.

[0329] Optionally, processing unit 1202 is specifically used to determine the second keyword in the first response information; processing unit 1202 is specifically used to determine the second replacement word corresponding to the second keyword; processing unit 1202 is specifically used to replace the second keyword in the first response information with the second replacement word to obtain a third response information.

[0330] Optionally, the first device further includes: a presentation unit 1203, configured to present multiple response information or first response information to the user, where the multiple response information includes the first response information and the second response information.

[0331] Optionally, the transceiver unit 1201 is also used to receive a first request message sent by the third unit; the transceiver unit 1201 is also used to send multiple response messages or first response messages to the third unit, and the multiple response messages include first response messages and second response messages.

[0332] Optionally, the first device mentioned above is a question answering device.

[0333] Optionally, the first request information includes at least one of the following: a question, a text processing request, and a translation request.

[0334] In this embodiment, the operations performed by each unit in the data processing device are the same as those described above. Figures 1A to 11 The description of the first unit in the illustrated embodiment is similar and will not be repeated here.

[0335] In this embodiment, false request information is obtained through a trusted unit or device (first unit or third unit), and the processing unit 1202 uses the false request information to confuse the real request information, so that the untrusted unit cannot clearly identify the real request information, thereby improving user privacy protection.

[0336] In another possible implementation, Fig.12 The data processing device shown is the aforementioned Figures 1A to 11 The third device in the illustrated embodiment has the following functions for each unit:

[0337] The transceiver unit 1201 is configured to receive a first request message sent by the first unit or a first response message corresponding to the first request message;

[0338] The transceiver unit 1201 is further used to send second request information to the first unit based on the first request information or the first response information, where the second request information is used to confuse the first request information.

[0339] Optionally, the third device is deployed on a network server, and the first unit is deployed on a terminal device.

[0340] Optionally, the transceiver unit 1201 is specifically used to obtain the second request information based on the first request information or the first response information; the transceiver unit 1201 is specifically used to send the second request information to the first unit.

[0341] In this embodiment, the operations performed by each unit in the data processing device are the same as those described above. Figures 1A to 11 The description of the third unit in the illustrated embodiment is similar and will not be repeated here.

[0342] In this embodiment, the transceiver unit 1201 in the trusted unit or device (third device) sends the second request information to the first unit based on the first request information or the first response information, so that the first unit can use the false request information to confuse the real request information, thereby improving the user's privacy protection.

[0343] In another possible implementation, Fig.12 The data processing device shown is the aforementioned Figures 1A to 11 The third device in the illustrated embodiment has the following functions for each unit:

[0344] The transceiver unit 1201 is used to receive the second response information sent by the first unit;

[0345] The processing unit 1202 is configured to obtain second request information based on the second response information;

[0346] The transceiver unit 1201 is further used to send a second request message to the first unit, where the second request message is used by the second unit to obtain second response message corresponding to the second request message.

[0347] In this embodiment, the operations performed by each unit in the data processing device are the same as those described above. Figures 1A to 11 The description of the third unit in the illustrated embodiment is similar and will not be repeated here.

[0348] In this embodiment, the transceiver unit 1201 reversely infers the false request information through the false response information. In general, the second device retrieves the log and other methods within a certain time period, as long as the time interval between the second device inferring the real request information and the false request information is less than the above time period. In the embodiment of the log, the transceiver unit 1201 can confuse the neural network in the untrusted unit (ie, the second device) through the false request information, thereby improving the user's privacy protection.

[0349] See also Fig.13 , the present application embodiment provides another data processing device, which can be the aforementioned Figures 1A to 11 For the convenience of explanation, the first device in the illustrated embodiment only shows the part related to the embodiment of the present application. For the specific technical details not disclosed, please refer to the method part of the embodiment of the present application. Take the first device as a mobile phone as an example:

[0350] Fig.13 FIG. 1 is a block diagram showing a partial structure of a mobile phone related to the first device provided in an embodiment of the present application. Fig.13 The mobile phone includes: a radio frequency (RF) circuit 1310, a memory 1320, an input unit 1330, a display unit 1340, a sensor 1350, an audio circuit 1360, a wireless fidelity (WiFi) module 1370, a processor 1380, and a power supply 1390. Those skilled in the art will understand that Fig.13 The mobile phone structure shown in the figure does not constitute a limitation on the mobile phone, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0351] Combine the following Fig.13 A detailed introduction to the various components of the mobile phone:

[0352] The RF circuit 1310 can be used for receiving and sending signals during the process of sending and receiving information or making calls. In particular, after receiving the downlink information of the base station, it is sent to the processor 1380 for processing; in addition, the designed uplink data is sent to the base station. Generally, the RF circuit 1310 includes but is not limited to an antenna, at least one amplifier, a transceiver, a coupler, a low noise amplifier (LNA), a duplexer, etc. In addition, the RF circuit 1310 can also communicate with the network and other devices through wireless communication. The above wireless communication can use any communication standard or protocol, including but not limited to the global system of mobile communication (GSM), general packet radio service (GPRS), code division multiple access (CDMA), wideband code division multiple access (WCDMA), long term evolution (LTE), email, short messaging service (SMS), etc.

[0353] The memory 1320 can be used to store software programs and modules. The processor 1380 executes various functional applications and data processing of the mobile phone by running the software programs and modules stored in the memory 1320. The memory 1320 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory 1320 can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0354] The input unit 1330 can be used to receive input digital or character information, and to generate key signal input related to the user settings and function control of the mobile phone. Specifically, the input unit 1330 may include a touch panel 1331 and other input devices 1332. The touch panel 1331, also known as a touch screen, can collect the user's touch operation on or near it (such as the user's operation on the touch panel 1331 or near the touch panel 1331 using any suitable object or accessory such as a finger, stylus, etc.), and drive the corresponding connection device according to a pre-set program. Optionally, the touch panel 1331 may include two parts: a touch detection device and a touch controller. Among them, the touch detection device detects the user's touch orientation, detects the signal brought by the touch operation, and transmits the signal to the touch controller; the touch controller receives the touch information from the touch detection device, converts it into contact coordinates, and then sends it to the processor 1380, and can receive and execute commands sent by the processor 1380. In addition, the touch panel 1331 can be implemented in various types such as resistive, capacitive, infrared, and surface acoustic waves. In addition to the touch panel 1331, the input unit 1330 may further include other input devices 1332. Specifically, the other input devices 1332 may include but are not limited to one or more of a physical keyboard, function keys (such as volume control keys, switch keys, etc.), a mouse, a joystick, etc.

[0355] The display unit 1340 can be used to display information input by the user or information provided to the user and various menus of the mobile phone. The display unit 1340 may include a display panel 1341. Optionally, the display panel 1341 may be configured in the form of a liquid crystal display (LCD), an organic light-emitting diode (OLED), etc. Further, the touch panel 1331 may cover the display panel 1341. When the touch panel 1331 detects a touch operation on or near it, it is transmitted to the processor 1380 to determine the type of touch event. Subsequently, the processor 1380 provides a corresponding visual output on the display panel 1341 according to the type of touch event. Although in Fig.13 In the embodiment, the touch panel 1331 and the display panel 1341 are used as two independent components to realize the input and output functions of the mobile phone, but in some embodiments, the touch panel 1331 and the display panel 1341 can be integrated to realize the input and output functions of the mobile phone.

[0356] The mobile phone may also include at least one sensor 1350, such as a light sensor, a motion sensor, and other sensors. Specifically, the light sensor may include an ambient light sensor and a proximity sensor, wherein the ambient light sensor may adjust the brightness of the display panel 1341 according to the brightness of the ambient light, and the proximity sensor may turn off the display panel 1341 and / or the backlight when the mobile phone is moved to the ear. As a type of motion sensor, the accelerometer sensor can detect the magnitude of acceleration in all directions (generally three axes), and can detect the magnitude and direction of gravity when stationary. It can be used for applications that identify the posture of the mobile phone (such as horizontal and vertical screen switching, related games, magnetometer posture calibration), vibration recognition related functions (such as pedometer, tapping), etc.; as for other sensors that can be configured in the mobile phone, such as gyroscopes, barometers, hygrometers, thermometers, infrared sensors, etc., they will not be repeated here.

[0357] The audio circuit 1360, the speaker 1361, and the microphone 1362 can provide an audio interface between the user and the mobile phone. The audio circuit 1360 can transmit the electrical signal converted from the received audio data to the speaker 1361, which is converted into a sound signal for output; on the other hand, the microphone 1362 converts the collected sound signal into an electrical signal, which is received by the audio circuit 1360 and converted into audio data, and then the audio data is output to the processor 1380 for processing, and then sent to, for example, another mobile phone through the RF circuit 1310, or the audio data is output to the memory 1320 for further processing. In the embodiment of the present application, the audio circuit 1360, the speaker 1361, and the microphone 1362 can cooperate to present response information to the user, for example, by presenting the response information to the user through voice broadcast.

[0358] WiFi is a short-range wireless transmission technology. The mobile phone can help users send and receive emails, browse web pages and access streaming media through the WiFi module 1370. It provides users with wireless broadband Internet access. Fig.13 A WiFi module 1370 is shown, but it is understandable that it is not an essential component of the mobile phone.

[0359] The processor 1380 is the control center of the mobile phone. It uses various interfaces and lines to connect various parts of the entire mobile phone. By running or executing software programs and / or modules stored in the memory 1320 and calling data stored in the memory 1320, it executes various functions of the mobile phone and processes data, thereby monitoring the mobile phone as a whole. Optionally, the processor 1380 may include one or more processing units; preferably, the processor 1380 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface and application programs, etc., and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 1380.

[0360] The mobile phone also includes a power supply 1390 (such as a battery) for supplying power to various components. Preferably, the power supply can be logically connected to the processor 1380 through a power management system, so that the power management system can manage functions such as charging, discharging, and power consumption management.

[0361] Although not shown, the mobile phone may also include a camera, a Bluetooth module, etc., which will not be described in detail here.

[0362] In the embodiment of the present application, the processor 1380 may execute the aforementioned Figures 1A to 11 The function of the first unit in the illustrated embodiment will not be described in detail here.

[0363] See also Fig.14 , a schematic diagram of the structure of another data processing device provided by the present application. The data processing device may be the aforementioned Figures 1A to 11 The second device or the third device in the illustrated embodiment.

[0364] The data processing device may include a processor 1401, a memory 1402 and a communication port 1403. The processor 1401, the memory 1402 and the communication port 1403 are interconnected via a line. The memory 1402 stores program instructions and data.

[0365] The memory 1402 stores the aforementioned Figures 1A to 11 In the corresponding embodiment shown, the steps executed by the data processing device correspond to program instructions and data.

[0366] Processor 1401, used to execute the above Figures 1A to 11 The steps shown in any of the illustrated embodiments are performed by a data processing device.

[0367] Communication port 1403 can be used to receive and send data, and to perform the above Figures 1A to 11 The steps related to obtaining, sending and receiving in any of the embodiments shown.

[0368] In one implementation, the data processing device may include Fig.14 More or fewer components are merely illustrative in this application and are not limiting.

[0369] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0370] A person skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0371] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, or it can be an electrical, mechanical or other form of connection.

[0372] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of the present application.

[0373] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0374] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or partly contributed to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory, a random access memory, a magnetic disk or an optical disk.

[0375] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A data processing method, the method being applied to a first unit, the method comprising: Sending first request information input by the user to the second unit; receiving first response information corresponding to the first request information sent by the second unit; Characterized in that the method further comprises: Sending second request information to the second unit, where the second request information is used to obfuscate the first request information; Receive second response information corresponding to the second request information sent by the second unit.

2. The method according to claim 1, characterized in that The method further comprises: The second request information is acquired based on the first request information.

3. The method according to claim 2, characterized in that The acquiring the second request information based on the first request information includes: Determining a first keyword in the first request information; The second request information is acquired based on the first keyword and the first request information.

4. The method according to claim 3, characterized in that The acquiring the second request information based on the first keyword and the first request information includes: Determine a first replacement word corresponding to the first keyword; The first keyword in the first request information is replaced by the first replacement word to obtain the second request information.

5. The method according to claim 3, characterized in that: The acquiring the second request information based on the first keyword and the first request information includes: sending the first keyword and the first request information to a third unit; The second request information sent by the third unit is received.

6. The method according to any one of claims 3 to 5, characterized in that The determining the first keyword in the first request information includes: The first keyword is determined based on a first operation of the user or a first preset rule.

7. The method according to claim 2, characterized in that The acquiring the second request information based on the first request information includes: Sending the first request information to the third unit; The second request information sent by the third unit is received.

8. The method according to claim 1, characterized in that The method further comprises: The second request information is generated based on the first response information.

9. The method according to claim 8, characterized in that The generating the second request information based on the first response information includes: Acquire third response information based on the first response information; The second request information is obtained based on the third response information.

10. The method according to claim 9, characterized in that The acquiring the third response information based on the first response information includes: determining a second keyword in the first response information; Determine a second replacement word corresponding to the second keyword; The second keyword in the first response information is replaced by the second replacement word to obtain the third response information.

11. The method according to any one of claims 1 to 10, characterized in that The method further comprises: A plurality of response messages or the first response message is presented to the user, wherein the plurality of response messages include the first response message and the second response message.

12. The method according to any one of claims 1-4, 8-10, characterized in that: The method further comprises: receiving the first request information sent by the third unit; Sending a plurality of response messages or the first response message to the third unit, where the plurality of response messages include the first response message and the second response message.

13. The method according to any one of claims 1 to 12, characterized in that The first unit is a question answering device.

14. The method according to any one of claims 1 to 13, characterized in that The first request information includes at least one of the following: a question, a text processing request, and a translation request.

15. A data processing method, characterized in that: The method is applied to the third unit, and the method comprises: Receiving a first request message sent by a first unit or a first response message corresponding to the first request message; Sending second request information to the first unit based on the first request information or the first response information, where the second request information is used to obfuscate the first request information.

16. The method according to claim 15, characterized in that The third unit is deployed on a network server, and the first unit is deployed on a terminal device.

17. A data processing method, the method comprising: The first unit sends first request information input by the user to the second unit; The second unit sends first response information corresponding to the first request information to the first unit; The second unit receives second request information sent by the first unit, where the second request information is used to obfuscate the first request information; The first unit receives second response information corresponding to the second request information sent by the second unit.

18. The method according to claim 17, characterized in that The method further comprises: The first unit obtains the second request information based on the first request information or the first response information.

19. The method according to claim 18, characterized in that The method further comprises: The first unit obtains the second request information based on the first response information.

20. A first device, characterized in that: The first device comprises: The transceiver unit is used to send the first request information input by the user to the second unit; The transceiver unit is further configured to receive first response information corresponding to the first request information sent by the second unit; The transceiver unit is further used to send second request information to the second unit, where the second request information is used to obfuscate the first request information; The transceiver unit is further configured to receive second response information corresponding to the second request information sent by the second unit.

21. A third device, characterized in that: The third device comprises: A transceiver unit, configured to receive a first request message sent by the first unit or a first response message corresponding to the first request message; The transceiver unit is further used to send second request information to the first unit based on the first request information or the first response information, where the second request information is used to confuse the first request information.

22. A first device, characterized in that: include: A processor, the processor is coupled to a memory, the memory is used to store programs or instructions, when the program or instructions are executed by the processor, the first device executes the method according to any one of claims 1 to 14.

23. A third device, characterized in that: include: A processor, wherein the processor is coupled to a memory, wherein the memory is used to store a program or an instruction, and when the program or the instruction is executed by the processor, the third device executes the method according to claim 15 or 16.

24. A communication system, characterized in that: The communication system includes: a first device and / or a third device; The first device is used to execute the method according to any one of claims 1 to 14, and the third device is used to execute the method according to claim 15 or 16.

25. A communication system, characterized in that: The communication system comprises: a first unit and a second unit; The first unit is used to send first request information input by a user to the second unit; The second unit is used to send first response information corresponding to the first request information to the first unit; The second unit is further used to receive second request information sent by the first unit, where the second request information is used to obfuscate the first request information; The first unit is further configured to receive second response information corresponding to the second request information sent by the second unit.

26. A computer-readable storage medium, characterized in that: The medium stores instructions, and when the instructions are executed by a computer, the method according to any one of claims 1 to 19 is implemented.

27. A computer program product, characterized in that The method comprises instructions which, when executed on a computer, cause the computer to execute the method according to any one of claims 1 to 19.

Images