Information processing method and device

Abstract

The invention discloses an information processing method and device, and belongs to the technical field of information processing. The method comprises the following steps: receiving a first input; obtaining first request information in response to the first input; extracting a first hidden state vector of the first request information; identifying privacy information in the first request information according to the first hidden state vector; processing privacy information in the first request information according to the virtual information to obtain second request information; the virtual information is generated according to the privacy information; and sending the second request information to the cloud server to obtain first response information.

Description

Technical Field

[0001] This application belongs to the technical field of information processing, and particularly relates to an information processing method and apparatus. Background Art

[0002] With the continuous development of the functions of electronic devices, users can input request information through a unified entrance to request services from a cloud server. Currently, the security review of the request information input by users is usually completed on the cloud server side.

[0003] However, uploading users' privacy information to the cloud server side poses certain security risks. On the one hand, during the transmission and storage of users' privacy information, it may be stolen or leaked due to security vulnerabilities of the cloud server; on the other hand, the cloud service provider may retain the data, and it is difficult for users to know information such as the scope of use and storage time of the privacy information.

[0004] Therefore, there is currently a lack of protection for the privacy information input by users. Summary of the Invention

[0005] The purpose of the embodiments of this application is to provide an information processing method and apparatus, which can solve the problem of the lack of protection for the privacy information input by users.

[0006] In a first aspect, the embodiments of this application provide an information processing method, which includes:

[0007] Receiving a first input;

[0008] In response to the first input, obtaining a first request information;

[0009] Extracting a first hidden state vector of the first request information;

[0010] Identifying the privacy information in the first request information according to the first hidden state vector;

[0011] Processing the privacy information in the first request information according to virtual information to obtain a second request information; the virtual information is generated according to the privacy information;

[0012] Sending the second request information to the cloud server to obtain a first response information.

[0013] In a second aspect, the embodiments of this application provide an information processing apparatus, which includes:

[0014] A first receiving module, configured to receive a first input;

[0015] An obtaining module, configured to obtain a first request information in response to the first input;

[0016] An extraction module for extracting a first hidden state vector of the first request information;

[0017] An identification module for identifying privacy information in the first request information according to the first hidden state vector;

[0018] A processing module for processing the privacy information in the first request information according to virtual information to obtain a second request information; the virtual information is generated according to the privacy information;

[0019] A sending module for sending the second request information to a cloud server to obtain a first response information.

[0020] In a third aspect, an embodiment of the present application provides an electronic device, which includes a processor and a memory. The memory stores a program or instruction that can run on the processor. When the program or instruction is executed by the processor, the steps of the method described in the first aspect are implemented.

[0021] In a fourth aspect, an embodiment of the present application provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the method described in the first aspect are implemented.

[0022] In a fifth aspect, an embodiment of the present application provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor, and the processor is used to run a program or instruction to implement the method described in the first aspect.

[0023] In a sixth aspect, an embodiment of the present application provides a computer program product, which is stored in a storage medium and is executed by at least one processor to implement the method described in the first aspect.

[0024] In the embodiments of the present application, by obtaining the first request information input by the user and extracting the first hidden state vector of the first request information, since the first hidden state vector can capture the key hidden features of the request information, which is convenient for subsequent information identification and processing operations. According to the first hidden state vector, the privacy information in the first request information is identified, and the privacy information in the first request information can be accurately found, providing a basis for subsequent privacy protection processing. The privacy information in the first request information is processed according to the virtual information to obtain the second request information. Since the virtual information is generated according to the privacy information, the privacy information of the user can be protected without affecting the business processing. The second request information is sent to the cloud server to obtain the first response information, which can avoid the leakage of privacy information during the transmission and processing process, and use the powerful computing power of the cloud server to provide the services required by the user while protecting the privacy of the user. Description of the Drawings

[0025] Figure 1 is a flowchart of an information processing method provided by an embodiment of the present application;

[0026] Figure 2 is a schematic diagram of a processing procedure of request intention information associated with an electronic device provided by an embodiment of the present application;

[0027] Figure 3 is a schematic diagram of a processing procedure of request intention information associated with a cloud server provided by an embodiment of the present application;

[0028] Figure 4 is a structural diagram of an information processing apparatus provided by an embodiment of the present application;

[0029] Figure 5 is one of the schematic diagrams of the hardware structure of an electronic device provided by an embodiment of the present application;

[0030] Figure 6 is the second schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present application. Detailed implementation manners

[0031] Next, the technical solutions of the embodiments of the present application will be clearly described in conjunction with the accompanying drawings of the embodiments of the present application. Obviously, the described embodiments are some, rather than all, of the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art belong to the scope of protection of the present application.

[0032] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, rather than to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are usually of the same category, and the number of objects is not limited. For example, the first object can be one or multiple. In addition, "and / or" in the specification and claims means at least one of the connected objects, and the character " / " generally indicates an "or" relationship between the associated objects before and after.

[0033] The information processing method provided by the embodiments of the present application can be applied to at least the following application scenarios, which will be described below.

[0034] Currently, taking the taxi-hailing scenario as an example, through the mobile phone interaction interface, the user inputs "Help me hail a taxi to Room 404, No. 66, Tanglong Road" at the unified entrance to initiate a taxi-hailing request. The text request input by the user is immediately uploaded to the intent control platform on the cloud server side. The intent control is a module that plays an overall coordinating and dispatching role in the information processing flow. The intent control on the cloud server side analyzes the text input by the user, identifies that the user's intent is "taxi-hailing", and extracts the corresponding intent slot as "Room 404, No. 66, Tanglong Road". The intent control sends the identified "taxi-hailing" intent back to the terminal side. After receiving the instruction, the terminal side calls the interface provided by the taxi-hailing software to perform the taxi-hailing operation and passes "Room 404, No. 66, Tanglong Road" as the destination parameter.

[0035] In this process, the intent control only needs to determine that "Room 404, No. 66, Tanglong Road" belongs to the address entity class, and without knowing the specific geographical location of this address, it can accurately identify the user's taxi-hailing intent. However, from the perspective of personal privacy protection, this information is extremely important.

[0036] Taking the scenario of copywriting generation as an example, through the mobile phone interaction interface, the user inputs "I want to post on a social media app to share my pet dog. Its name is Duoduo. We live in Jingyuan Community. Every day I take it for a walk in the community garden" at the unified entrance to initiate a copywriting generation request to the system. The text request input by the user is uploaded to the intent control platform on the cloud server side and waits for further processing. The intent control on the cloud server side analyzes the text input by the user and identifies that the user's intent is "copywriting generation". The intent control distributes the identified "copywriting generation" intent on the cloud server side and calls the large language model (LLM) in the server. The LLM generates the corresponding sharing copy on the social media app according to the text content input by the user.

[0037] During the copywriting generation process, information such as the name of the pet dog "Duoduo" and the user's address "Jingyuan Community" has no substantial impact on the quality of the generated copy. However, from the perspective of personal privacy protection, this information is extremely important.

[0038] The direct upload of the privacy information in the user's text to the cloud server side will cause the following significant problems:

[0039] When the user's privacy information is directly uploaded to the cloud server side, during the entire process of data transmission and storage, it faces the potential threat of being stolen or leaked. The servers of cloud service providers may be maliciously attacked due to security vulnerabilities, resulting in the illegal acquisition of user data. Such data leakage may cause serious damage to the user's personal privacy and rights.

[0040] After the user data is uploaded to the cloud server side, the cloud service provider often retains the data. During this process, users lack effective control means over the usage scope and storage duration of their own data. Even if the user actively deletes the relevant data, the cloud service provider may not be able to completely clear the data due to technical limitations or management oversights, etc. This poses a potential risk of data abuse, which may lead to the infringement of users' privacy, such as the data being used for unauthorized marketing activities or other improper purposes.

[0041] Taking the above-mentioned taxi-hailing scenario as an example, as an agent for intent distribution, the LLM only needs to identify the types of these privacy entities, without having to know the specific content of these privacy entities. Its core operation is just to transfer these privacy entities to the corresponding service providers on the terminal side. For example, in the taxi-hailing scenario, the LLM only needs to recognize that "Room 404, No. 66, Tanglong Road" is an address entity and then transfer it to the taxi service interface, without having to delve into the detailed information of this address.

[0042] Referring to the above-mentioned copywriting generation scenario, as an intelligent agent for text understanding, the LLM focuses on grasping the semantic structure of the context, rather than understanding the specific meaning of the privacy entities. When generating the copywriting, the LLM only needs to embed these privacy entities into the corresponding positions in the generated content. For example, when generating a sharing copywriting on a social media application, for the pet dog's name "Duoduo" and the address "Jingyuan Community", the LLM mainly places them in the appropriate expression positions, rather than deeply analyzing their specific references.

[0043] Therefore, from the perspective of actual business needs, uploading personal privacy information to the cloud server side is not a necessary condition for ensuring the normal operation of the business. On the premise of meeting the business functions, it is entirely possible to optimize the business process to avoid unnecessary cloud operations of privacy information, thereby better protecting the personal privacy of users.

[0044] In view of the problems that occur in related technologies, the embodiments of this application provide an information processing method and device, which can solve the problem of the lack of protection for the privacy information input by users in related technologies.

[0045] The following will combine the accompanying drawings to explain in detail the information processing method provided by the embodiments of this application through specific embodiments and their application scenarios.

[0046] Figure 1 It is a flowchart of an information processing method provided by the embodiments of this application.

[0047] As Figure 1 shown, this information processing method may include step 110 - step 160. This method is applied to an information processing device, and is specifically as follows:

[0048] Step 110, receive a first input;

[0049] Among them, the first input is an input operation performed by the user on the electronic device. The first input can be voice input, text input, or shortcut function input. For example, when the user says to the mobile phone, "I want to query the weather in Haidian District, Beijing tomorrow", the behavior of voice input belongs to the first input.

[0050] Step 120: In response to the first input, obtain first request information;

[0051] First request information: The text content input through a unified entry on the user's electronic device, which expresses the user's demand for obtaining a certain service.

[0052] By receiving the text content input by the user, it serves as the basis for subsequent processing. It can collect the user's needs and provide raw data for subsequent intent recognition and service processing.

[0053] Step 130: Extract the first hidden state vector of the first request information;

[0054] First hidden state vector: In natural language processing, usually with the help of deep learning models, such as recurrent neural network RNN, long short-term memory network LSTM, or Transformer, etc., the input text information is converted into a vector representation. This vector contains various features such as semantics and grammar of the input text, which is convenient for the computer to perform subsequent processing.

[0055] Through a pre-trained natural language processing model, the first request information is encoded to convert the text information into a numerical vector representation. The model learns the semantic and syntactic structures of the language during the training process, so the generated hidden state vector can capture the key features of the request information. It can convert the text information into a vector form that is easy for the computer to process, facilitating subsequent information recognition and processing operations.

[0056] Step 140: According to the first hidden state vector, identify the privacy information in the first request information;

[0057] Privacy information: Refers to information related to the user's personal identity, sensitive data, etc., such as name, phone number, ID number, home address, etc. Once these information are leaked, they may bring security risks to the user.

[0058] The first hidden state vector can be analyzed based on a trained privacy information recognition model. This model learns the characteristic patterns of privacy information through a large amount of labeled data, so as to be able to identify the corresponding privacy information from the vector. Accurately find out the privacy information in the first request information, providing a basis for subsequent privacy protection processing.

[0059] Step 150: Process the privacy information in the first request information according to the virtual information to obtain second request information; the virtual information is generated according to the privacy information.

[0060] Virtual information: Substitute information generated according to real privacy information, which may be somewhat related to the privacy information in form but does not contain real sensitive content, and is used to replace the privacy information in subsequent processing to play a role in protecting privacy.

[0061] Second request information: The request information obtained by replacing the privacy information in the first request information with virtual information, which can avoid the leakage of privacy information while ensuring that the semantics of the request are generally unchanged.

[0062] According to the identified privacy information, generate corresponding virtual information according to certain rules, and then replace the privacy information in the first request information with the virtual information to obtain the processed second request information. Without affecting the semantics of the request information and business processing, the privacy information of the user is protected, and it is avoided that it is leaked during transmission and processing.

[0063] Step 160: Send the second request information to the cloud server to obtain first response information.

[0064] First response information: The result returned by the cloud server after processing the second request information, which is used to respond to the user's initial request.

[0065] Transmit the processed second request information to the cloud server through the network. The cloud server processes the request according to the preset business logic and algorithm, and finally generates the corresponding response information. While protecting the user's privacy, the powerful computing power of the cloud server is utilized to complete the processing of the user's request and provide the required services for the user.

[0066] In a possible embodiment, step 130 may specifically include the following steps:

[0067] Perform regular matching processing on the first request information to obtain the annotation information of the first request information.

[0068] Perform representation learning on the annotation information of the first request information to obtain the first hidden state vector of the first request information.

[0069] Regular matching processing: A regular expression is a tool for describing string patterns. Regular matching processing is to use regular expressions to find strings that conform to specific patterns in the first request information, so as to add specific annotation information to the text.

[0070] Annotation information: Information obtained after analyzing and marking the first request information, which can reflect the characteristics, categories, and other attributes of different parts of the text, and is helpful for subsequent processing and understanding.

[0071] Representation learning: A machine learning technique that aims to convert annotation information into a more meaningful and computer - processable representation form, that is, a hidden state vector.

[0072] Regular expressions define a series of rules and patterns to match specific strings in the text. Using pre - defined business - related regular expressions, scan the first request information. When part of the text matches the pattern of the regular expression, add corresponding annotations to this part of the content.

[0073] Exemplarily, assume the first request information is "I want to book a flight to Beijing tomorrow". Define regular expressions to match the date and location. For example, the regular expression for the date may be \d{4}-\d{2}-\d{2}, and the regular expression for the location may be a list of common city names. After matching, label "tomorrow" as "date" and "Beijing" as "location".

[0074] Among them, the regular matching rules for digital - encoded sensitive personal information are shown in Table 1 specifically:

[0075] Table 1

[0076]

[0077]

[0078] Perform structured processing on the first request information, identify and mark the key information in the text, provide more targeted input for subsequent representation learning, and improve the efficiency and accuracy of information processing.

[0079] Representation learning uses models such as neural networks to learn and transform the annotation information. The model will automatically extract the features in the annotation information and map them into a low - dimensional vector space. The obtained vector is the first hidden state vector, which can comprehensively reflect the semantics and features of the first request information.

[0080] Exemplarily, use a pre - trained language model, such as BERT, and input the annotation information into the model. The model will encode the annotation information and, through multi - layer neural network calculations, finally output a fixed - length vector as the first hidden state vector.

[0081] Convert the annotation information into a numerical vector that is easy for a computer to process, facilitating further analysis and processing of the first request information, such as privacy information recognition, etc. At the same time, it can also eliminate the influence brought by the diversity of text expressions to a certain extent, improving the stability and reliability of information processing.

[0082] In a possible embodiment, step 140 may specifically include the following steps:

[0083] Identify the first hidden state vector to obtain a representation vector of at least one privacy entity;

[0084] Perform fusion processing on the first hidden state vector and the representation vector of the at least one privacy entity to obtain a fusion vector;

[0085] Identify the privacy information in the first request information according to the fusion vector.

[0086] Representation vector of a privacy entity: A privacy entity refers to a specific object containing privacy information, such as a person's name, phone number, etc. The representation vector is to convert these privacy entities into a numerical vector form for convenient computer processing and analysis, and this vector contains the feature information of the privacy entity.

[0087] Fusion processing: The process of merging vector information from different sources or different types into one vector, aiming to integrate information from multiple aspects for more comprehensive subsequent analysis.

[0088] Fusion vector: The new vector obtained after fusion processing, which synthesizes the information of the first hidden state vector and the privacy entity representation vector.

[0089] Use a pre-trained model or a specific algorithm to analyze the first hidden state vector. The model can identify the features related to privacy entities in the vector and convert them into corresponding representation vectors by learning a large amount of privacy entity data. This is based on the model's learning and capturing ability of the feature patterns of privacy entities.

[0090] Specifically, the text input by the user can be processed through Named Entity Recognition (NER) technology. The NER model will identify various named entities in the text according to its trained knowledge, such as person names, place names, organization names, etc. In this process, some potentially privacy-related entities may be identified. For example, a person's name may be part of the privacy information.

[0091] NER is a technology in natural language processing, aiming to identify named entities with specific meanings from text, such as person names, place names, organization names, dates, times, etc. It can help computers understand the entity information in text and provide a basis for subsequent analysis and processing.

[0092] Exemplarily, assume that the first hidden state vector is obtained after processing the user input "My name is Zhang San and my phone number is 13800138000". Using a trained named entity recognition model, this vector is recognized, and "Zhang San" and "13800138000" are recognized as privacy entities and converted into corresponding representation vectors respectively.

[0093] Extract the information related to privacy entities from the first hidden state vector, providing a key basis for subsequent privacy information recognition and helping to more accurately locate privacy information.

[0094] Integrate the overall features of the request information contained in the first hidden state vector with the privacy features contained in the privacy entity representation vector. Through fusion, the new vector can reflect both the overall semantics of the request information and the features of the privacy entities therein.

[0095] Specifically, based on a two-way interactive dual attention mechanism, the first hidden state vector and the representation vectors of the at least one privacy entity are fused to obtain a fused vector, so as to extract context semantic features and entity type features, thereby determining whether the text input by the user is privacy text, presenting the determination result to the user, and explicitly marking the privacy entities in the user input text in a text-highlighted manner on the interaction interface.

[0096] The dual attention mechanism will interact and allocate weights between the two information sources of the first hidden state vector and the representation vectors of the at least one privacy entity, comprehensively considering their features and correlations to determine which are the real privacy information. For example, it may further determine whether a phone number belongs to privacy information based on whether the phone number matched by regular expression is mentioned in the surrounding text in relation to relevant personal information. If certain information has a high correlation with privacy-related features, it will be determined as privacy information.

[0097] Dual Attention is a technology applied in fields such as natural language processing and computer vision for more refined feature extraction and analysis of data such as text or images. The dual attention mechanism is usually developed based on the attention mechanism. The core idea of the attention mechanism is to enable the model to automatically focus on important parts when processing data. Dual attention further considers the interaction and correlation between two different perspectives or modalities, and dynamically allocates attention weights by learning the correlation between different elements in the data, so as to better capture the semantic information and structural features in the data.

[0098] Thus, by integrating the overall features of the request information and the features of the privacy entity, the subsequent recognition process can take into account more information, improving the accuracy and comprehensiveness of privacy information recognition.

[0099] Use a classifier based on machine learning or deep learning to analyze the fused vector. The classifier learns the mapping relationship between different fused vector features and privacy information through training, so as to determine which parts of the first request information belong to privacy information.

[0100] Exemplarily, use a trained binary classifier and input the fused vector into the classifier. The classifier makes a judgment based on the features of the vector, and the output result shows that "Zhang San" and "13800138000" are privacy information.

[0101] Accurately identifying the privacy information in the first request information based on the fused information provides a basis for subsequent privacy protection processing and effectively protects the privacy security of users.

[0102] In a possible embodiment, before step 150, the following steps may further be included:

[0103] Identify the emotional feature information in the privacy information;

[0104] Generate the virtual information according to the privacy information and the emotional feature information;

[0105] Wherein, perform representation learning on the virtual information to obtain a second hidden state vector, and the similarity between the second hidden state vector and the first hidden state vector is less than a preset similarity threshold.

[0106] Emotional feature information: Refers to the features such as emotional tendency and tone contained in the privacy information. In this embodiment, the emotional feature may be reflected in the form of a nickname with a certain emotional color.

[0107] Virtual information: Information used to replace real privacy information. While ensuring that real privacy is not disclosed, it maintains the semantic coherence of the requested information as much as possible so that the cloud server can process the request normally.

[0108] Second hidden state vector: A vector obtained after performing representation learning on the generated virtual information. Similar to the first hidden state vector, it is also a numerical representation used to reflect the characteristics of the virtual information.

[0109] Preset similarity threshold: A numerically preset value used to measure the similarity between two vectors. When the similarity of two vectors is less than this threshold, it indicates that the difference between them is relatively large.

[0110] With the help of sentiment analysis technology in natural language processing, analyze the privacy information. This technology, through a trained model, identifies features such as vocabulary and grammatical structures in the text, thereby determining the implied sentiment tendency and sentiment features. For example, certain words may have obvious emotional colors such as intimacy, formality, and solemnity, and the model can identify them based on these features.

[0111] Exemplarily, assume the privacy information is "Dear Mom". Through the sentiment analysis model, it can be identified that "Dear" in it reflects the intimate sentiment feature, and "Mom" here may be a nickname with emotional color.

[0112] Accurately capture the emotional features in the privacy information, providing a basis for generating virtual information that better conforms to the user's emotional expression later. After the virtual information replaces the privacy information, it can better retain the emotional tone of the original requested information.

[0113] Combining the content of the privacy information and the identified emotional feature information, use preset rules or generation models to create virtual information. The rules can be based on methods such as semantic replacement and synonym replacement, while considering the emotional features, so that the generated virtual information not only does not disclose privacy but also maintains the original emotional tendency.

[0114] Exemplarily, for the privacy information "Dear Mom", according to the intimate emotional feature, the virtual information "Lovely elder" can be generated. Here, both the intimate emotion is retained, and the specific kinship title is replaced by "elder", protecting the privacy.

[0115] Generate virtual information that can effectively replace the privacy information. While protecting privacy, maintain the semantic and emotional coherence of the requested information as much as possible to ensure that the cloud server can understand and process the request normally.

[0116] The virtual information is converted into a second hidden state vector by performing the same representation learning process on the virtual information as that of the first request information. Then, the similarity between these two vectors is calculated and compared with a preset similarity threshold. If the similarity is less than the threshold, it indicates that the virtual information has sufficient differences from the original private information at the feature level, which can effectively protect privacy.

[0117] Exemplarily, assume that the cosine similarity calculation result between the first hidden state vector and the second hidden state vector is 0.2, while the preset similarity threshold is 0.3. Since 0.2 is less than 0.3, it indicates that the generated virtual information has significant differences from the original private information. Ensuring that the generated virtual information has sufficient differences from the original private information can prevent attackers from inferring the original private information by analyzing the second hidden state vector, further enhancing the security of privacy protection.

[0118] In a possible embodiment, after step 160, the following steps may further be included:

[0119] Receive the first response information;

[0120] Analyze the first request information to obtain request intent information and an application associated with the first request information;

[0121] When the processing of the request intent information is associated with the electronic device, process the first response information according to the privacy information to obtain a second response information, and send the second response information to the application;

[0122] When the processing of the request intent information is associated with the cloud server, send the first response information to the application.

[0123] First response information: The result returned by the cloud server after processing the second request information, which contains the response content to the user's request.

[0124] Request intent information: The actual purpose or requirement that the user wants to achieve obtained by analyzing the first request information, such as taking a taxi, booking a flight, obtaining a knowledge Q&A, etc.

[0125] Application associated with the first request information: The specific application program that can implement the user's request intent, such as a taxi application, a flight booking application, a knowledge Q&A application, etc.

[0126] Second response information: The result obtained by processing the first response information according to the privacy information when the processing of the request intent information is associated with the electronic device, which is used to respond to the user's request while protecting privacy.

[0127] The electronic device receives the first response message sent back by the cloud server through a network connection. The network communication protocol is responsible for data transmission and reception, ensuring that information can be accurately transmitted from the cloud server to the electronic device. Obtain the processing result of the cloud server for the user request, providing a data basis for subsequent further processing of the response message and responding to the user request.

[0128] Utilize natural language processing techniques, such as semantic analysis, named entity recognition, intent classification, etc., to analyze the first request message. Semantic analysis understands the meaning of the text, named entity recognition finds key entities, and intent classification determines the user's request intent based on text features. According to the request intent, match the corresponding application program. Exemplarily, when the user inputs "Help me book a flight from Shanghai to Beijing tomorrow", through the analysis of natural language processing techniques, it is recognized that the request intent is to book a flight, and the associated application is the flight booking application.

[0129] Thus, the user's request intent and the corresponding application can be clarified, providing a basis for subsequent processing of the response message according to different situations, ensuring that the response message can be accurately sent to the appropriate application for processing.

[0130] When the processing of the request intent is mainly carried out on the electronic device side, in order to protect privacy, the first response message is processed according to the previously identified privacy information. For example, the privacy content involved in the response message is replaced with the original privacy information or other appropriate content. After processing, the second response message is sent to the application associated with the request for subsequent display or processing by the application.

[0131] Thus, on the premise of protecting user privacy, the application on the electronic device side can accurately process and display the response message to meet the user's request needs.

[0132] When the processing of the request intent is mainly carried out on the cloud server side, it indicates that the first response message has been fully processed on the cloud server side and no additional privacy-related processing is required. The first response message is directly sent to the application associated with the request for subsequent operations by the application. Exemplarily, for the user request "Help me generate a travel promotion copywriting", the first response message returned after the cloud server processes it is the generated copywriting. Since the copywriting generation is completed on the cloud server side, this first response message is directly sent to the copywriting editing application for the user to view or further edit.

[0133] Thus, the processing flow can be simplified, the processing result of the cloud server can be quickly sent to the application, the processing efficiency of the system can be improved, and the user can obtain the required service result in a timely manner.

[0134] The following combines Figure 2 to illustrate the processing process of the request intent information in association with the electronic device:

[0135] The electronic device receives a ride-hailing request input by the user, such as "Help me hail a taxi to Building B, No. 15, Sunshine Garden" or "Help me hail a taxi to Room 404, No. 66, Tanglong Road". The intent control center is responsible for managing and scheduling the entire intent processing flow. The electronic device uses a built-in algorithm to analyze the input request information and identify the privacy entity information therein, such as the address "Room 404, No. 66, Tanglong Road". After identifying the privacy entity, the electronic device will blur or replace the sensitive part. For example, only send "No. 66, Tanglong Road" to the ride-hailing application and hide more precise privacy information such as "Room 404", which can not only meet the ride-hailing needs but also protect the user's privacy.

[0136] Thus, the processed second response information is sent to the ride-hailing application, and the electronic device communicates with the ride-hailing application. The electronic device completes the ride-hailing operation on the premise of protecting the user's privacy.

[0137] The following Figure 3 illustrates the situation where the processing process of the request intent information is associated with the cloud server:

[0138] The user issues an intent request such as "I want to post on a social media application and share my pet Xixi. We take a walk in Shawan Community". The intent control module in the cloud server is responsible for overall management, receiving and scheduling these request information. The cloud server analyzes information such as "my pet Duoduo, we take a walk in Jingyuan Community" input by the user through a built-in algorithm and model, and identifies the privacy entities therein, such as the pet name "Duoduo", the community name "Jingyuan Community", etc. After identifying the privacy entity, the cloud server performs privacy protection processing, blurs or replaces the sensitive information. For example, retains non-sensitive information such as the community type, and blurs the specific precise location such as "Jingyuan Community".

[0139] Thus, the first response information is sent to the social media application to complete operations such as user sharing, while protecting the user's privacy.

[0140] Among them, in the step of processing the first response information according to the privacy information to obtain the second response information involved above, it may specifically include the following steps:

[0141] Replace the virtual information in the first response information with the privacy information to obtain a third response information;

[0142] Blur the privacy information in the third response information to obtain the second response information.

[0143] Third response information: The response information obtained by replacing the virtual information in the first response information with the original privacy information. It is closer in content to the result that should be obtained based on the processing of the original privacy information, but the privacy information is in a completely exposed state.

[0144] Fuzzy processing: Modify the privacy information in a certain form to reduce the recognizability of the privacy information and the risk of privacy leakage without affecting the overall usability and semantic understanding of the information. For example, fuzzily processing the real name "Zhang San" of a person to "Mr. Zhang", or replacing some digits of the ID number with "*", etc.

[0145] Since in the second request information sent to the cloud server before, virtual information was used to replace the real privacy information to protect privacy. And the first response information obtained by the cloud server based on the virtual information also contains this virtual information. At this time, in order to make the response information conform to the actual situation, it is necessary to replace the virtual information back with the original privacy information. This is based on the requirements of information restoration and integrity to ensure that the response information can accurately correspond to the original request.

[0146] Exemplarily, the user inputs "My name is Zhang San, help me check Zhang San's express delivery information", and in the second request information sent to the cloud server, "Zhang San" is replaced with the virtual information "Mr. Somebody". The first response information returned by the cloud server is "Mr. Somebody's express delivery has arrived at the pick-up point". Replacing "Mr. Somebody" back with "Zhang San" gives the third response information "Zhang San's express delivery has arrived at the pick-up point".

[0147] Thus, the response information corresponds to the privacy information in the original request, ensuring the accuracy and coherence of the information, and making the response content for subsequent processing and display more in line with the actual situation.

[0148] To further protect user privacy, it is necessary to perform fuzzy processing on the privacy information in the third response information. There are various ways of fuzzy processing, such as partial hiding, desensitization, etc. Through these ways, the recognizability of the privacy information can be reduced to a certain extent without affecting the overall usability of the response information and the understanding of it.

[0149] Exemplarily, for the third response information "Zhang San's express delivery has arrived at the pick-up point", fuzzily processing "Zhang San" can display it as "Zhang *", obtaining the second response information "Zhang *'s express delivery has arrived at the pick-up point".

[0150] Because the data requirements in different business scenarios are different, it is necessary to communicate and discuss with the party responsible for the specific business. For example, social software and financial software have different requirements for user privacy information, so it is necessary to jointly discuss and determine a suitable solution.

[0151] Before the step of obfuscating the privacy information in the third response message, a privacy-intention filtering table can be defined. The privacy-intention filtering table is a specially defined table. "Privacy" refers to various types of user privacy information, such as name, age, address, contact information, etc.; "intention" can be understood here as the purpose of business function implementation. For example, a social software wants to implement a friend recommendation function, and a financial software wants to implement a transfer function, etc. The privacy-intention filtering table should clarify which types of privacy entities need to be collected at least and the quantity of each type of privacy entity in each specific business scenario in order to implement the corresponding business function.

[0152] The principle of minimum necessity is an important principle in privacy protection, which means that when collecting and using user privacy information, only the minimum amount of information necessary for implementing specific business functions should be collected and used, and over-collection is not allowed. By defining the privacy-intention filtering table, operations can be carried out according to the regulations of this table to ensure that the collected privacy information is the least and necessary, thus meeting the minimum necessity requirement and protecting the privacy rights and interests of users.

[0153] Thereby, on the basis of ensuring that the response message can meet the basic needs and understanding of users, the degree of privacy protection is further enhanced, and the risk of privacy information being leaked during subsequent display or use is reduced.

[0154] In a possible embodiment, after step 150, the following steps may further be included:

[0155] Record the mapping relationship between the privacy information and the virtual information;

[0156] In the step of processing the first response message according to the privacy information to obtain the second response message as mentioned above, the following steps may specifically be included:

[0157] Process the first response message according to the mapping relationship and the privacy information to obtain the second response message;

[0158] Delete the mapping relationship.

[0159] Mapping relationship: refers to the one-to-one corresponding association relationship between privacy information and virtual information. Through this relationship, the corresponding privacy information can be found according to the virtual information when needed, or vice versa.

[0160] When replacing the privacy information with virtual information and sending it to the cloud server for processing, in order to accurately restore the response message returned by the cloud server to the content related to the original privacy information subsequently, the mapping relationship between the two needs to be recorded. This is an information association and storage mechanism, which is convenient for subsequent information restoration operations.

[0161] Exemplarily, when the user inputs "My phone number is 13800138000", replace "13800138000" with the virtual information "1xxxxxxxxxx", and at this time, record the mapping relationship between "13800138000" and "1xxxxxxxxxx".

[0162] Thereby, it provides a basis for subsequent processing of the response information returned by the cloud server, ensuring that the virtual information can be accurately restored to the private information and guaranteeing the accuracy and coherence of the information.

[0163] After receiving the first response information returned by the cloud server, use the previously recorded mapping relationship to replace the virtual information in the first response information with the corresponding private information. Then, according to actual needs, the private information may be further processed to finally obtain the second response information. This is done to enable the response information to accurately reflect the user's original request and related information while protecting privacy.

[0164] Exemplarily, the first response information returned by the cloud server is "There is a new message for your number 1xxxxxxxxxx". According to the mapping relationship, replace "1xxxxxxxxxx" with "13800138000" to get "There is a new message for your number 13800138000". If further obfuscation processing of the private information is required, it can be processed as "There is a new message for your number 138****8000", which is the second response information.

[0165] Thereby, while protecting privacy, the response information can accurately correspond to the original request, meet the user's needs, and further enhance the effect of privacy protection.

[0166] After completing the processing of the response information, successfully restoring the virtual information to the private information and obtaining the final second response information, the mapping relationship is no longer needed. To further protect the user's privacy and avoid the leakage or abuse of the mapping relationship, it needs to be deleted. This is a security measure for privacy protection.

[0167] For example, after processing the first response information into the second response information, delete the previously recorded mapping relationship between "13800138000" and "1xxxxxxxxxx" from the database or cache storing the mapping relationship.

[0168] Thereby, the risk of private information leakage can be reduced, the potential security hazards caused by the storage of the mapping relationship can be minimized, and the privacy security of users can be further guaranteed.

[0169] In an embodiment of the present application, by obtaining the first request information input by the user and extracting the first hidden state vector of the first request information, since the first hidden state vector can capture the key hidden features of the request information, facilitating subsequent information recognition and processing operations, based on the first hidden state vector, the privacy information in the first request information is recognized, and the privacy information in the first request information can be accurately found, providing a basis for subsequent privacy protection processing. The privacy information in the first request information is processed according to the virtual information to obtain the second request information. Since the virtual information is generated based on the privacy information, it can protect the user's privacy information without affecting business processing. The second request information is sent to the cloud server for the cloud server to process the second request information to obtain the first response information, which can avoid the leakage of privacy information during transmission and processing and provide the required services for the user using the powerful computing power of the cloud server while protecting the user's privacy.

[0170] For the information processing method provided by the embodiment of the present application, the execution subject may be an information processing device. In the embodiment of the present application, taking the information processing device executing the information processing method as an example, the information processing device provided by the embodiment of the present application is described.

[0171] Figure 4 It is a block diagram of an information processing device provided by an embodiment of the present application. The device 400 includes:

[0172] The first receiving module 410 is configured to receive the first input;

[0173] The obtaining module 420 is configured to obtain the first request information in response to the first input;

[0174] The extracting module 430 is configured to extract the first hidden state vector of the first request information;

[0175] The identifying module 440 is configured to identify the privacy information in the first request information according to the first hidden state vector;

[0176] The processing module 450 is configured to process the privacy information in the first request information according to the virtual information to obtain the second request information; the virtual information is generated according to the privacy information;

[0177] The sending module 460 is configured to send the second request information to the cloud server to obtain the first response information.

[0178] In a possible embodiment, the extracting module 430 is specifically configured to:

[0179] Perform regular matching processing on the first request information to obtain the annotation information of the first request information;

[0180] Perform representation learning on the annotation information of the first request information to obtain the first hidden state vector of the first request information.

[0181] In a possible embodiment, the recognition module 440 is specifically configured to:

[0182] Recognize the first hidden state vector to obtain the representation vectors of at least one privacy entity;

[0183] Perform fusion processing on the first hidden state vector and the representation vectors of the at least one privacy entity to obtain a fusion vector;

[0184] Identify the privacy information in the first request information according to the fusion vector.

[0185] In a possible embodiment, the recognition module 440 is further configured to recognize the emotional feature information in the privacy information;

[0186] The device 400 may further include:

[0187] A generation module, configured to generate the virtual information according to the privacy information and the emotional feature information;

[0188] Wherein, perform representation learning on the virtual information to obtain a second hidden state vector, and the similarity between the second hidden state vector and the first hidden state vector is less than a preset similarity threshold.

[0189] In a possible embodiment, the device 400 may further include:

[0190] A second receiving module, configured to receive the first response information;

[0191] An analysis module, configured to analyze the first request information to obtain request intention information and an application associated with the first request information;

[0192] The processing module 450 is further configured to, when the processing process of the request intention information is associated with the electronic device, process the first response information according to the privacy information to obtain a second response information, and send the second response information to the application;

[0193] The sending module 460 is further configured to, when the processing process of the request intention information is associated with the cloud server, send the first response information to the application.

[0194] In an embodiment of the present application, by obtaining the first request information input by the user and extracting the first hidden state vector of the first request information, since the first hidden state vector can capture the key hidden features of the request information, facilitating subsequent information recognition and processing operations, based on the first hidden state vector, the private information in the first request information is recognized, and the private information in the first request information can be accurately found, providing a basis for subsequent privacy protection processing. The private information in the first request information is processed according to the virtual information to obtain the second request information. Since the virtual information is generated based on the private information, it can protect the user's private information without affecting business processing. The second request information is sent to the cloud server to obtain the first response information, which can avoid the leakage of private information during transmission and processing, and use the powerful computing power of the cloud server to provide the required services for the user while protecting the user's privacy.

[0195] The information processing device in the embodiment of the present application can be an electronic device or a component in the electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal or other devices other than the terminal. Exemplarily, the electronic device can be a mobile phone, a tablet computer, a laptop computer, a handheld computer, a vehicle-mounted electronic device, a Mobile Internet Device (MID), an augmented reality (AR) / virtual reality (VR) device, a robot, a wearable device, an ultra-mobile personal computer (UMPC), a netbook, or a personal digital assistant (PDA), etc., and can also be a server, a Network Attached Storage (NAS), a personal computer (PC), a television (TV), a teller machine, or a self-service machine, etc. The embodiment of the present application does not make specific limitations.

[0196] The information processing device in the embodiment of the present application can be a device with an operating system. The operating system can be an Android operating system, an iOS operating system, or other possible operating systems. The embodiment of the present application does not make specific limitations.

[0197] The information processing device provided in the embodiment of the present application can implement each process implemented in the above method embodiment. To avoid repetition, it will not be elaborated here.

[0198] Optionally, as Figure 5As shown in the figure, an embodiment of the present application further provides an electronic device 510, including a processor 511, a memory 512, a program or instruction stored on the memory 512 and executable on the processor 511. When the program or instruction is executed by the processor 511, it implements each step of any of the above information processing method embodiments and can achieve the same technical effects. To avoid repetition, it will not be elaborated here.

[0199] It should be noted that the electronic devices in the embodiments of the present application include the above-mentioned mobile electronic devices and non-mobile electronic devices.

[0200] Figure 6 FIG. [FIGURE NUMBER] is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present application.

[0201] The electronic device 600 includes, but is not limited to: a radio frequency unit 601, a network module 602, an audio output unit 603, an input unit 604, a sensor 605, a display unit 606, a user input unit 607, an interface unit 608, a memory 609, and a processor 610 and other components.

[0202] Those skilled in the art can understand that the electronic device 600 may further include a power source (such as a battery) for supplying power to each component. The power source can be logically connected to the processor 610 through a power management system, so as to implement functions such as management of charging, discharging, and power consumption management through the power management system. Figure 6 The structure of the electronic device shown in [FIGURE NUMBER] does not constitute a limitation on the electronic device. The electronic device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements, which will not be elaborated here.

[0203] Among them, the user input unit 607 is used to receive a first input;

[0204] The processor 610 is configured to obtain first request information in response to the first input;

[0205] The processor 610 is further configured to extract a first hidden state vector of the first request information;

[0206] The processor 610 is further configured to identify privacy information in the first request information according to the first hidden state vector;

[0207] The processor 610 is further configured to process the privacy information in the first request information according to virtual information to obtain second request information; the virtual information is generated according to the privacy information;

[0208] The network module 602 is configured to send the second request information to a cloud server to obtain first response information.

[0209] Please note that the "FIG. [FIGURE NUMBER]" in the translation of and should be replaced with the actual figure number. Also, the " ", "

[0199] ", etc. tags are preserved as they are in the original text.Optionally, the processor 610 is further configured to perform regular matching processing on the first request information to obtain annotation information of the first request information;

[0210] The processor 610 is further configured to perform representation learning on the annotation information of the first request information to obtain a first hidden state vector of the first request information.

[0211] Optionally, the processor 610 is further configured to identify the first hidden state vector to obtain representation vectors of at least one privacy entity;

[0212] The processor 610 is further configured to perform fusion processing on the first hidden state vector and the representation vectors of the at least one privacy entity to obtain a fusion vector;

[0213] The processor 610 is further configured to identify privacy information in the first request information according to the fusion vector.

[0214] Optionally, the processor 610 is further configured to identify emotional feature information in the privacy information;

[0215] The processor 610 is further configured to generate the virtual information according to the privacy information and the emotional feature information;

[0216] Wherein, a second hidden state vector is obtained by performing representation learning on the virtual information, and the similarity between the second hidden state vector and the first hidden state vector is less than a preset similarity threshold.

[0217] Optionally, the network module 602 is further configured to receive the first response information;

[0218] The processor 610 is further configured to analyze the first request information to obtain request intention information and an application associated with the first request information;

[0219] The processor 610 is further configured to, when the processing process of the request intention information is associated with the electronic device, process the first response information according to the privacy information to obtain a second response information, and send the second response information to the application;

[0220] The network module 602 is further configured to, when the processing process of the request intention information is associated with the cloud server, send the first response information to the application.

[0221] In an embodiment of the present application, by obtaining first request information input by a user and extracting a first hidden state vector of the first request information, since the first hidden state vector can capture key hidden features of the request information, facilitating subsequent information recognition and processing operations, based on the first hidden state vector, private information in the first request information is recognized, and the private information in the first request information can be accurately found, providing a basis for subsequent privacy protection processing. The private information in the first request information is processed according to virtual information to obtain second request information. Since the virtual information is generated based on the private information, it can protect the user's private information without affecting business processing. The second request information is sent to a cloud server to obtain a first response message, which can avoid the leakage of private information during transmission and processing, and use the powerful computing power of the cloud server to provide the required services for the user while protecting the user's privacy.

[0222] It should be understood that in an embodiment of the present application, the input unit 604 may include a Graphics Processing Unit (GPU) 6041 and a microphone 6042. The graphics processor 6041 processes image data of static pictures or video images obtained by an image capture device (such as a camera) in a video image capture mode or an image capture mode. The display unit 606 may include a display panel 6061, and the display panel 6061 may be configured in the form of a liquid crystal display, an organic light-emitting diode, etc. The user input unit 607 includes at least one of a touch panel 6071 and other input devices 6072. The touch panel 6071 is also called a touch screen. The touch panel 6071 may include two parts: a touch detection device and a touch controller. The other input devices 6072 may include, but are not limited to, a physical keyboard, function keys (such as volume control keys, switch keys, etc.), a trackball, a mouse, and an action lever, which will not be elaborated here. The memory 609 may be used to store software programs and various data, including but not limited to application programs and operating systems. The processor 610 may integrate an application processor and a modem processor. Among them, the application processor mainly processes the operating system, user interfaces, and application programs, and the modem processor mainly processes wireless communication. It can be understood that the above modem processor may not be integrated into the processor 610.

[0223] The memory 609 can be used to store software programs and various data. The memory 609 may mainly include a first storage area for storing programs or instructions and a second storage area for storing data. Among them, the first storage area can store an operating system, application programs or instructions required for at least one function (such as a sound playback function, an image playback function, etc.). In addition, the memory 609 can include volatile memory or non-volatile memory, or the memory 609 can include both volatile and non-volatile memory. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM), a static random access memory (SRAM), a dynamic random access memory (DRAM), a synchronous dynamic random access memory (SDRAM), a double data rate synchronous dynamic random access memory (DDR SDRAM), an enhanced synchronous dynamic random access memory (ESDRAM), a synch link dynamic random access memory (SLDRAM), and a direct rambus random access memory (DRRAM). The memory 609 in the embodiments of the present application includes but is not limited to these and any other suitable types of memory.

[0224] The processor 610 may include one or more processing units; optionally, the processor 610 integrates an application processor and a modem processor. Among them, the application processor mainly processes operations related to the operating system, user interface, and application programs, etc., and the modem processor mainly processes wireless communication signals, such as a baseband processor. It can be understood that the above-mentioned modem processor may not be integrated into the processor 610 either.

[0225] The embodiments of the present application also provide a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, it realizes each process of the above information processing method embodiment and can achieve the same technical effect. To avoid repetition, it will not be elaborated here.

[0226] Among them, the processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media such as computer read-only memory ROM, random access memory RAM, magnetic disks, or optical discs, etc.

[0227] Another embodiment of the present application provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor. The processor is used to run programs or instructions to implement each process of the above information processing method embodiment, and can achieve the same technical effects. To avoid repetition, it will not be elaborated here.

[0228] It should be understood that the chip mentioned in the embodiments of the present application may also be referred to as a system-on-chip, system chip, chip system, or system-on-chip, etc.

[0229] The embodiments of the present application provide a computer program product. The program product is stored in a storage medium. The program product is executed by at least one processor to implement each process of the above information processing method embodiment, and can achieve the same technical effects. To avoid repetition, it will not be elaborated here.

[0230] It should be noted that in this article, the term "including", "comprising", or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article, or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such a process, method, article, or device. Without further limitations, an element defined by the statement "including a..." does not exclude the existence of additional identical elements in the process, method, article, or device including that element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed. It may also include performing functions in a substantially simultaneous manner or in the reverse order according to the functions involved. For example, the described methods may be performed in an order different from that described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0231] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-described example methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. The computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions for causing a terminal (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in various embodiments of the present application.

[0232] The embodiments of the present application have been described above in conjunction with the accompanying drawings. However, the present application is not limited to the above specific implementation manners. The above specific implementation manners are merely illustrative and not restrictive. Under the inspiration of the present application, those of ordinary skill in the art can also make many forms without departing from the purpose of the present application and the scope protected by the claims, and all of them belong to the protection scope of the present application.

Claims

1. An information processing method, characterized in that, The method includes: Receiving a first input; In response to the first input, obtaining first request information; Extracting a first hidden state vector of the first request information; Identifying privacy information in the first request information according to the first hidden state vector; Processing the privacy information in the first request information according to virtual information to obtain second request information; the virtual information is generated according to the privacy information; Sending the second request information to a cloud server to obtain first response information.

2. The method according to claim 1, characterized in that, The extracting the first hidden state vector of the first request information includes: Performing regular matching processing on the first request information to obtain annotation information of the first request information; Performing representation learning on the annotation information of the first request information to obtain a first hidden state vector of the first request information.

3. The method according to claim 1, wherein The identifying the privacy information in the first request information according to the first hidden state vector includes: Identifying the first hidden state vector to obtain representation vectors of at least one privacy entity; Performing fusion processing on the first hidden state vector and the representation vectors of the at least one privacy entity to obtain a fusion vector; Identifying the privacy information in the first request information according to the fusion vector.

4. The method according to claim 1, wherein Before the processing the privacy information in the first request information according to virtual information to obtain second request information, the method further includes: Identifying emotional feature information in the privacy information; Generating the virtual information according to the privacy information and the emotional feature information; Wherein, performing representation learning on the virtual information to obtain a second hidden state vector, and the similarity between the second hidden state vector and the first hidden state vector is less than a preset similarity threshold.

5. The method according to claim 1, characterized in that After the sending the second request information to the cloud server, the method further includes: Receiving the first response information; Analyzing the first request information to obtain request intention information and an application associated with the first request information; When the processing process of the request intention information is associated with an electronic device, processing the first response information according to the privacy information to obtain a second response information, and sending the second response information to the application; When the processing process of the request intention information is associated with the cloud server, sending the first response information to the application.

6. An information processing apparatus, characterized in that, The device includes: A first receiving module, configured to receive a first input; An obtaining module, configured to obtain first request information in response to the first input; An extracting module, configured to extract a first hidden state vector of the first request information; An identifying module, configured to identify privacy information in the first request information according to the first hidden state vector; A processing module, configured to process the privacy information in the first request information according to virtual information to obtain second request information; the virtual information is generated according to the privacy information; A sending module, configured to send the second request information to a cloud server to obtain first response information.

7. The device according to claim 6, characterized in that, The extracting module is specifically configured to: Perform regular matching processing on the first request information to obtain annotation information of the first request information; Perform representation learning on the annotation information of the first request information to obtain a first hidden state vector of the first request information.

8. The device according to claim 6, characterized in that The recognition module is specifically configured to: Recognize the first hidden state vector to obtain representation vectors of at least one privacy entity; Perform a fusion process on the first hidden state vector and the representation vectors of the at least one privacy entity to obtain a fusion vector; Identify the privacy information in the first request information according to the fusion vector.

9. The device according to claim 6, wherein The recognition module is further configured to identify the emotional feature information in the privacy information; The apparatus further includes: A generation module, configured to generate the virtual information according to the privacy information and the emotional feature information; Wherein, representation learning is performed on the virtual information to obtain a second hidden state vector, and the similarity between the second hidden state vector and the first hidden state vector is less than a preset similarity threshold.

10. The device according to claim 6, characterized in that, The apparatus further includes: A second receiving module, configured to receive the first response information; An analysis module, configured to analyze the first request information to obtain request intention information and an application associated with the first request information; The processing module is further configured to, when the processing process of the request intention information is associated with the electronic device, process the first response information according to the privacy information to obtain a second response information, and send the second response information to the application; The sending module is further configured to, when the processing process of the request intention information is associated with the cloud server, send the first response information to the application.

Images