Non-privacy data acquisition and message push method and system

Abstract

The present invention discloses a method and system for obtaining non-private data and message pushing, belonging to the technical field of big data processing. The technical problem to be solved is how to push messages to users based on user behavior while protecting user privacy. The method includes the following steps: obtaining private data and non-private data; obtaining user behavior data and interface data in non-private data in real time, identifying the current usage status of the user, obtaining user operation habits and usage intentions, and predicting the user operation gap time; constructing a usage habit reference database of the user based on the user operation habits and usage intentions, and establishing a time prediction model; classifying users according to preset user tags to obtain a general database corresponding to the user tags, and training the time prediction model based on the general database; pushing messages when the user uses the smart device based on the promotion time and the message pushing position.

Description

Technical Field

[0001] The present invention relates to the technical field of big data processing, and more specifically, to a method and system for obtaining non-private data and pushing messages. Background Art

[0002] During the use of application programs on user terminals, a lot of application program behavior data will be generated, such as user operation behavior data, user browsing behavior data, user setting behavior data, etc. How to effectively mine these application program data to determine relevant user portraits for subsequent message pushing is a big data mining and analysis method that most existing social, shopping, reading, and short video software are currently building. However, the existing big data mining and processing methods do not fully consider user privacy, obtain user-related data without restriction, and when pushing relevant messages to users, they usually use rough methods such as advertising pop-ups or directly construct an information cocoon for users based on their browsing records, seriously damaging the user experience.

[0003] How to push messages to users based on user behavior while protecting user privacy is a technical problem that needs to be solved. Summary of the Invention

[0004] The technical task of the present invention is to address the above deficiencies and provide a method and system for obtaining non-private data and pushing messages to solve the technical problem of how to push messages to users based on user behavior while protecting user privacy.

[0005] In a first aspect, a method for obtaining non-private data and pushing messages according to the present invention includes the following steps:

[0006] Process the user data in the user file through a data segmentation algorithm to obtain private data and non-private data, and perform integrity detection on the private data, where the user data is the relevant data generated by the user using a smart device;

[0007] Store the non-private data in a data transfer station and use the data transfer station as the user's big data source. The cloud data center continuously obtains and processes the non-private data from the user's big data source;

[0008] Continuously obtain the user behavior data and interface data in the non-private data, identify the user's current usage status, obtain the user operation frequency parameter and the smart device display interface layout parameter, analyze the user's usage behavior based on the user operation frequency parameter and the smart device display interface layout parameter to obtain the user's operation habits and usage intentions, and predict the user's operation gap time;

[0009] Build a reference database of user usage habits based on user operation habits and usage intentions, and establish a time prediction model. The time prediction model is used to predict and output user operation gap data based on user operation habits and usage intentions, and the user operation gap time is used as the promotion time.

[0010] Classify users according to preset user tags, aggregate the reference databases of usage habits of at least one user under the same user tag to obtain a general database corresponding to the user tag, and train the time prediction model based on the general database to obtain a trained time prediction model.

[0011] Predict the user operation gap time through the trained time prediction model based on the obtained user operation habits and usage intentions to obtain the promotion time, and predict the message push position based on the intelligent device display interface layout parameters. Push messages when the user uses the intelligent device based on the promotion time and the message push position.

[0012] Preferably, the integrity detection of privacy data is performed through the following steps:

[0013] Obtain the location information of the privacy data in the user file.

[0014] Establish a mapping table between the privacy data and the original file according to the privacy data and the location information, establish an index table according to the original file and the mapping table, and store the mapping table and the index table in a relational database at the same time.

[0015] Detect whether the data content in the privacy data has been tampered with by detecting whether the digital digests before and after the data are consistent, so as to achieve the integrity detection of the privacy data content.

[0016] Preferably, the cloud data center obtains and processes non-privacy data from the user big data source in real time, and further includes the following steps:

[0017] L1. Obtain all network search data through the non-privacy data. The network search data is the search terms entered by the user in each search engine, and construct a search term group based on the obtained search terms.

[0018] L2. Obtain the search term group at the network data acquisition end

[0019] L3. Select any search term in the search term group.

[0020] L4. Obtain the key data in the search term group, including obtaining the name, usage keyword, and feature description word in the key data, and perform fitting analysis on the key data and the search terms.

[0021] L5. Select the next search term and repeat step L4.

[0022] L6. Repeat step L5 until all search terms are processed, obtaining a target word group composed of several target words;

[0023] L7. Retrieve relevant push content based on the target word group.

[0024] Preferably, when analyzing the user's usage behavior based on the user operation frequency parameter and the intelligent device display interface layout parameter, analyze the corresponding operation frequency, operation content, and stay time when the user uses the application based on the user operation frequency parameter, and analyze the user's demand for the application or the theme content of the interface and the user's usage intention. Among them, the user's usage intention includes the search frequency of the user for the specified keyword, its stay time, and the corresponding operation behavior data.

[0025] Preferably, classify users according to the preset user tags, aggregate the usage habit reference databases of at least one user under the same user tag to obtain a general database corresponding to the user tag, classify users according to the preset user tags to form a specified user tag group, for each user tag group, collect data from the usage habit reference databases corresponding to all users under the user tag group, and perform an intersection based on the user operation habits and usage intentions to establish a homogeneous general database for the user tag group.

[0026] In a second aspect, the present invention provides a non-private data acquisition and message push system for pushing messages through the non-private data acquisition and message push method described in any item of the first aspect. The system includes:

[0027] A data processing module, which is used to process the user data in the user file through a data segmentation algorithm to obtain private data and non-private data, and perform integrity detection on the private data, where the user data is the relevant data generated by the user using the intelligent device;

[0028] A data transfer module, which is used to store the non-private data in the data transfer station and use the data transfer station as the user big data source. The cloud data center retrieves and processes the non-private data from the user big data source in real time;

[0029] A data analysis module, which is used to obtain the user behavior data and interface data in the non-private data in real time, identify the user's current usage status, obtain the user operation frequency parameter and the intelligent device display interface layout parameter, analyze the user's usage behavior based on the user operation frequency parameter and the intelligent device display interface layout parameter to obtain the user operation habit and usage intention, and predict the user operation gap time;

[0030] A model construction module, which is used to construct a usage habit reference database of a user based on the user's operation habits and usage intentions, and establish a time prediction model. The time prediction model is used to predict and output user operation gap data based on the user's operation habits and usage intentions, and the user operation gap time is used as the promotion time;

[0031] A model training module, which is used to classify users according to preset user tags, aggregate the usage habit reference databases of at least one user under the same user tag to obtain a general database corresponding to the user tag, and perform model training on the time prediction model based on the general database to obtain a trained time prediction model;

[0032] A prediction module, which is used to predict the user operation gap time based on the obtained user operation habits and usage intentions through the trained time prediction model to obtain the promotion time, and predict the message push position based on the intelligent device display interface layout parameters, and perform message push when the user uses the intelligent device based on the promotion time and the message push position.

[0033] Preferably, the data processing module is used to perform integrity detection on privacy data through the following steps:

[0034] Obtain the location information of the privacy data in the user file;

[0035] Establish a mapping table between the privacy data and the original file according to the privacy data and the location information, establish an index table according to the original file and the mapping table, and store the mapping table and the index table in the relational database at the same time;

[0036] Detect whether the data content in the privacy data has been tampered with by detecting whether the digital digests before and after the data are consistent, so as to achieve the integrity detection of the privacy data content.

[0037] Preferably, the data transfer module is used to perform the following operations to enable the cloud data center to obtain and process non-privacy data from the user big data source in real time:

[0038] L1. Obtain all network search data through non-privacy data. The network search data is the search terms entered by the user in various search engines, and construct a search term group based on the obtained search terms;

[0039] L2. Obtain the search term group at the network data acquisition end

[0040] L3. Select any search term in the search term group;

[0041] L4. Obtain the key data in the search term group, including obtaining the name, usage keyword, and feature description word in the key data, and perform fitting analysis on the key data and the search terms;

[0042] L5. Optionally select the next search term and repeat step L4;

[0043] L6. Repeat step L5 until all search terms are processed, obtaining a target term group composed of several target terms;

[0044] L7. Retrieve relevant push content based on the target term group.

[0045] Preferably, when analyzing the user's usage behavior based on the user operation frequency parameter and the intelligent device display interface layout parameter, the data analysis module is used to analyze the corresponding operation frequency, operation content, and stay time when the user uses the application program based on the user operation frequency parameter, and analyze the user's demand degree for the application program or the theme content of the interface and the user's usage intention. Among them, the user's usage intention includes the search frequency of the user for the specified keyword, its stay time, and the corresponding operation behavior data.

[0046] Preferably, the model training module is used to classify users according to the preset user tags, aggregate the usage habit reference databases of at least one user under the same user tag to obtain a general database corresponding to the user tag, classify users according to the preset user tags to form a specified user tag group, for each user tag group, collect data from the usage habit reference databases corresponding to all users under the user tag group, and perform an intersection based on the user operation habits and usage intentions to establish a homogeneous general database for the user tag group.

[0047] The non-privacy data acquisition and message push method and system of the present invention have the following advantages: By organizing the relevant data generated by the user using the intelligent device into a user file, dividing the user-related data in the user file into non-privacy data and privacy data through a data segmentation algorithm, uploading the non-privacy data, establishing a time prediction model, continuously collecting and continuously learning and training the user's usage habits, screening the usage habits of the classified user groups through the usage habit reference databases of a large number of users to complete the intersection of habits, completing the establishment of a homogeneous general database, and then reversely learning and training the time prediction model, and finally pushing messages with the best time prediction to improve the user's usage experience. On the premise of not affecting the user's use, complete the message push with a high degree of user attention at the best position to improve the overall user experience. Description of the Drawings

[0048] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for use in the embodiments or the description of the prior art. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0049] The present invention will be further described below with reference to the accompanying drawings.

[0050] Figure 1 It is a flowchart of the method for obtaining non-private data and pushing messages in Embodiment 1. Specific embodiments

[0051] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand the present invention and be able to implement it. However, the specific embodiments cited do not limit the present invention. Without conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0052] The embodiments of the present invention provide a method and system for obtaining non-private data and pushing messages, which are used to solve the technical problem of how to push messages to users based on user behavior while protecting user privacy.

[0053] Embodiment 1:

[0054] A method for obtaining non-private data and pushing messages according to the present invention includes the following steps:

[0055] S100. Process the user data in the user file through a data segmentation algorithm to obtain private data and non-private data, and perform integrity detection on the private data, where the user data is the relevant data generated by the user using the smart device;

[0056] S200. Store the non-private data in the data transfer station, and use the data transfer station as the user's big data source. The cloud data center retrieves and processes the non-private data from the user's big data source in real time;

[0057] S300. Retrieve the user behavior data and interface data in the non-private data in real time, identify the current usage status of the user, obtain the user operation frequency parameter and the smart device display interface layout parameter, analyze the user's usage behavior based on the user operation frequency parameter and the smart device display interface layout parameter to obtain the user's operation habits and usage intentions, and predict the user's operation gap time;

[0058] S400. Construct a usage habit reference database for the user based on the user's operation habits and usage intentions, and establish a time prediction model. The time prediction model is used to predict and output user operation gap data based on the user's operation habits and usage intentions, and the user operation gap time is used as the promotion time.

[0059] S500. Classify users according to preset user tags, aggregate the usage habit reference databases of at least one user under the same user tag to obtain a general database corresponding to the user tag, and train the time prediction model based on the general database to obtain a trained time prediction model.

[0060] S600. Based on the obtained user operation habits and usage intentions, predict the user operation gap time through the trained time prediction model to obtain the promotion time, and predict the message push position based on the intelligent device display interface layout parameters. Push messages when the user uses the intelligent device based on the promotion time and the message push position.

[0061] In step S100 of this embodiment, the integrity of the privacy data is detected through the following steps:

[0062] (1) Obtain the location information of the privacy data in the user file.

[0063] (2) Establish a mapping table between the privacy data and the original file according to the privacy data and the location information, establish an index table according to the original file and the mapping table, and store the mapping table and the index table in the relational database at the same time.

[0064] (3) Detect whether the data content in the privacy data has been tampered with by detecting whether the digital digests before and after the data are the same, so as to achieve the integrity detection of the privacy data content.

[0065] In this step operation, the original file is the screenshot or video data captured from the interface. For example, for an APP, to record the position of the APP login interface relative to the entire APP interface, the data at that position needs to be removed during data processing to protect user privacy. The index table is the general table of information related to the time and storage location corresponding to the original file.

[0066] The user file refers to the captured screenshot or video, and the original file can be changed to the user file.

[0067] The purpose of establishing the table is to detect whether the digital digests before and after the data are the same to detect whether the data content has been tampered with, so as to achieve the integrity detection of the privacy data content.

[0068] Detect whether the data content has been tampered with by detecting whether the digital digests before and after the data are the same, so as to achieve the integrity detection of the privacy data content.

[0069] Through the integrity detection of the privacy data, it is determined that the user's privacy data will not be collected by the cloud data center, ensuring the user's privacy. At the same time, it prevents the data tampering by maliciously implanted programs after the data segmentation algorithm processes the large data in the user's file.

[0070] In step S200, the data transfer station can be understood as a transfer station for caching. Not all of these cached data need to be stored, but those that have been used are discarded.

[0071] In step S200, the cloud data center continuously obtains and processes non-privacy data from the user's large data source, and further includes the following steps:

[0072] L1. Obtain all network search data through non-privacy data. The network search data is the search terms entered by the user in various search engines, and construct a search term group based on the obtained search terms;

[0073] L2. Obtain the search term group at the network data acquisition end

[0074] L3. Select any search term within a search term group;

[0075] L4. Obtain the key data within the search term group, including obtaining the name, usage keyword, and feature description word within the key data, and perform fitting analysis on the key data and the search terms;

[0076] L5. Select the next search term and repeat step L4;

[0077] L6. Repeat step L5 until all search terms are processed, obtaining a target term group composed of several target terms;

[0078] L7. Retrieve relevant push content based on the target term group.

[0079] As a specific implementation of this step, the operations are as follows:

[0080] (1) Obtain all network search data through non-privacy data. The network search data is the search terms entered by the user in various search engines, obtaining a search term group composed of several search terms;

[0081] (2) The action analysis unit obtains the search term group at the network data acquisition end;

[0082] Among them, the action analysis unit is used to analyze the input instructions, input frequency, input time, and interval time of the user's keyboard and mouse. The network data acquisition end corresponds to the network search data and is the execution end;

[0083] (3) Select any search term within a search term group;

[0084] (4) Obtain the key data in the action analysis unit (key data includes name, usage keywords and feature descriptors. Set the content as needed. After completing data acquisition, proceed directly to the subsequent push steps. Please read the briefing content carefully); obtain the name, usage keywords and feature descriptors in the key data (based on AI language recognition, perform keyword extraction, and existing conventional methods can be used); perform fitting analysis on the key data and search terms to form a target word group;

[0085] (5) Select the next search term and repeat step (4);

[0086] (6) Repeat step (5) until all search terms are processed, and a target term group consisting of several target terms is obtained. The target term group is the semantics finally recognized. The content of the key data is identified through AI, and the analysis result of the corresponding user's behavior preference is output. This accurate result is directly used to retrieve the push content;

[0087] (7) Retrieve relevant push content based on the target word group, retrieve relevant content, and push it according to the predicted time and the current user behavior and interface.

[0088] The cloud data center determines user preferences through the non-privacy data it obtains. It determines whether the search term is the user's preference by the search frequency of the search term, the dwell time on the search results page, and the number of clicks on the search results page. This avoids the problem of pushing related messages to all the user's search terms, causing the user to receive a lot of useless information and a poor user experience.

[0089] Step S300 analyzes the operation frequency, operation content and dwell time corresponding to the user's use of a certain APP through user operation frequency parameters, analyzes the user's demand for the APP or the interface theme content and the user's usage intention, wherein the user's usage intention includes the user's search frequency for specified keywords, its dwell time and the corresponding operation behavior data.

[0090] By analyzing the user's usage behavior, we can determine the user's intention towards the APP or the interface theme content. At the same time, we can identify the user's operation frequency. The operation frequency is closely related to the user's page reading speed. Therefore, by analyzing the data parameters related to the operation frequency, operation content and dwell time, we can obtain the user's attention to certain content, as well as the importance of this certain content to the user. At the same time, we can predict the interval time between operations based on the user, laying the foundation for the preset time of subsequent push messages.

[0091] In step S4, a time prediction model for the user is established to analyze the user's operation habits and preference tendencies, and a reference database of the user's usage habits exclusive to the user is established.

[0092] The time prediction model continuously learns and trains through the reference database of usage habits. Moreover, the longer the user uses the intelligent device, the more data the reference database of usage habits obtains, and the closer the learning and training results of the time prediction model will be to the user's real usage habits and preference tendencies.

[0093] In step S500, the users are classified according to the preset user tags. The reference databases of the usage habits of at least one user under the same user tag are aggregated to obtain a general database corresponding to the user tag. The users are classified according to the preset user tags to form a specified user tag group. For each user tag group, data collection is performed on the reference databases of the usage habits corresponding to all users under the user tag group, and the intersection is made based on the user's operation habits and usage intentions to establish a homogeneous general database for the user tag group.

[0094] In this specific implementation, according to the preset tags, the users are classified to form a specified user tag group. Data collection is performed on the reference databases of the usage habits corresponding to all users under the user tag group, and the intersection is made based on the habitual feature data and preference tendency data to establish a homogeneous general database for the user tag group. The intelligent promotion prediction time model under the user tag group is continuously learned and trained based on the homogeneous general database through retraining the model.

[0095] By setting tags for the users, the setting basis depends on specific needs. For example: preference tendencies, social identities, genders, ages, etc. Data collection is performed on all the reference databases of the usage habits of the user groups under different tags to obtain the same habitual feature data and preference tendencies that most users under the tag will have, and the corresponding data is stored in the homogeneous general database. After a new user first uses the intelligent device, the intelligent promotion prediction time model of the new user uses the shortest time to determine the tag of the new user, and learns and trains from the corresponding homogeneous general database according to the tag to adapt to the usage habits of the new user in the shortest time. At the same time, non-private data of the new user is continuously obtained, and the intelligent promotion prediction time model of the new user is continuously improved to enhance the usage experience of the new user.

[0096] The retrained model performs reverse learning on the user habits under the corresponding tags based on a general database of the same type, anticipates the usage habits of the user habit reference database, and presents the habit characteristics in advance in the case where the user has not shown or has rarely shown this habit. After uploading non-private data, it determines whether the user's anticipatory behavior for this habit characteristic is adaptable or conforms to the user's usage habits, and further learns and trains the habit reference database based on the judgment result.

[0097] Step S600 identifies the layout of the display interface of the smart device and selects the best position as the display position for the push message, avoiding the simple and violent display of traditional push messages in the center of the interface during pushing, which may cause users' disgust with the push information.

[0098] Embodiment 2:

[0099] A non-private data acquisition and message push system of the present invention includes a data processing module, a data transfer module, a data analysis module, a model construction module, a model training module, and a prediction module. This system can execute the method disclosed in Embodiment 1.

[0100] The data processing module is used to process the user data in the user file through a data segmentation algorithm to obtain private data and non-private data, and perform integrity detection on the private data, where the user data is the relevant data generated by the user using the smart device.

[0101] In this embodiment, the data processing module is used to perform integrity detection on the private data through the following steps:

[0102] (1) Obtain the location information of the private data in the user file;

[0103] (2) Establish a mapping table between the private data and the original file according to the private data and the location information, establish an index table according to the original file and the mapping table, and store the mapping table and the index table in a relational database;

[0104] (3) Detect whether the data content in the private data has been tampered with by detecting whether the digital digests before and after the data are consistent, so as to achieve the integrity detection of the private data content.

[0105] In this module, the original file is the screenshot or video data captured from the interface. For example, for an APP, to record the position of the APP login interface relative to the entire APP interface, the data at that position needs to be removed during data processing to protect user privacy. The index table is the general table of information related to the time and storage location corresponding to the original file.

[0106] The user file refers to the captured screenshot or video, and the original file can be changed to the user file.

[0107] The purpose of establishing the table is to detect whether the digital digest before and after the data is consistent to detect whether the data content has been tampered with, so as to achieve the integrity detection of the private data content.

[0108] By detecting whether the digital digest before and after the data is consistent to detect whether the data content has been tampered with, so as to achieve the integrity detection of the private data content.

[0109] Through the integrity detection of private data, it is determined that the user's private data will not be collected by the cloud data center, ensuring the privacy of the user. At the same time, it prevents the data tampering by maliciously implanted programs after the data splitting algorithm processes the large data in the user's file.

[0110] The data transfer module is used to store non-private data in the data transfer station and use the data transfer station as the user's large data source. The cloud data center retrieves and processes non-private data from the user's large data source in real time.

[0111] The data transfer station can be understood as a transfer station for caching. Not all of these cached data need to be stored, but those that have been used are discarded.

[0112] In this embodiment, the data transfer module is used to execute the following to enable the cloud data center to retrieve and process non-private data from the user's large data source in real time:

[0113] L1. Obtain all network search data through non-private data. The network search data is the search terms entered by the user in various search engines, and construct a search term group based on the obtained search terms;

[0114] L2. Obtain the search term group at the network data acquisition end

[0115] L3. Select any search term within a search term group;

[0116] L4. Obtain the key data within the search term group, including obtaining the name, usage keyword, and feature description word within the key data, and perform fitting analysis on the key data and the search terms;

[0117] L5. Select the next search term and repeat step L4;

[0118] L6. Repeat step L5 until all search terms are processed to obtain a target term group composed of several target terms;

[0119] L7. Retrieve relevant push content based on the target term group.

[0120] As the specific implementation of this step, the operation is as follows:

[0121] (1) Obtain all network search data through non-privacy data, where the network search data is the search terms entered by users in various search engines, and obtain a search term group consisting of several search terms;

[0122] (2) the action parsing unit obtains the search term group from the network data acquisition terminal;

[0123] Among them, the action analysis unit is used to analyze the user's keyboard and mouse input instructions, input frequency, input time and interval time. The network data acquisition end corresponds to the network search data and is the execution end;

[0124] (3) Select any search term in the search term group;

[0125] (4) Obtain the key data in the action analysis unit (key data includes name, usage keywords and feature descriptors. Set the content as needed. After completing data acquisition, proceed directly to the subsequent push steps. Please read the briefing content carefully); obtain the name, usage keywords and feature descriptors in the key data (based on AI language recognition, perform keyword extraction, and existing conventional methods can be used); perform fitting analysis on the key data and search terms to form a target word group;

[0126] (5) Select the next search term and repeat step (4);

[0127] (6) Repeat step (5) until all search terms are processed, and a target term group consisting of several target terms is obtained. The target term group is the semantics finally recognized. The content of the key data is identified through AI, and the analysis result of the corresponding user's behavior preference is output. This accurate result is directly used to retrieve the push content;

[0128] (7) Retrieve relevant push content based on the target word group, retrieve relevant content, and push it according to the predicted time and the current user behavior and interface.

[0129] The cloud data center uses the non-privacy data it obtains to determine the user's preference. It determines whether the search term is the user's preference by the search frequency, the time spent on the search results page, and the number of clicks on the search results page. This avoids the problem of pushing relevant messages to all the user's search terms, causing the user to receive a lot of useless information and a poor user experience.

[0130] The data analysis module is used to obtain user behavior data and interface data in non-privacy data in real time, identify the user's current usage status, obtain user operation frequency parameters and smart device display interface layout parameters, analyze user usage behavior based on user operation frequency parameters and smart device display interface layout parameters, obtain user operation habits and usage intentions, and predict user operation interval time.

[0131] When analyzing the user's usage behavior based on the user operation frequency parameter and the intelligent device display interface layout parameter, the data analysis module is used to analyze the operation frequency, operation content, and stay time corresponding to the user's use of the application based on the user operation frequency parameter, and analyze the user's demand degree for the application or the theme content of the interface and the user's usage intention. Among them, the user's usage intention includes the search frequency of the user for the specified keyword, its stay time, and the corresponding operation behavior data.

[0132] In this embodiment, the data analysis module is used to analyze the operation frequency, operation content, and stay time corresponding to the user's use of a certain APP through the user operation frequency parameter, and analyze the user's demand degree for the APP or the theme content of the interface and the user's usage intention. Among them, the user's usage intention includes the search frequency of the user for the specified keyword, its stay time, and the corresponding operation behavior data.

[0133] The data analysis module is used to judge the user's intention for the APP or the theme content of the interface by analyzing the user's usage behavior. At the same time, it identifies the user's operation frequency, which has a great relationship with the user's page reading speed. Therefore, by analyzing the data parameters related to the operation frequency, operation content, and stay time, the attention degree of the user to certain content can be obtained, which also means the importance of certain content to the user. At the same time, the operation interval time of the user can be predicted, laying a foundation for the preset time of the subsequent push message.

[0134] The model construction module is used to construct a user usage habit reference database based on the user operation habit and usage intention, and establish a time prediction model. The time prediction model is used to predict and output the user operation gap data based on the user operation habit and usage intention, and the user operation gap time is used as the promotion time.

[0135] The model construction module is used to establish a time prediction model for the user, analyze the user's operation habit and the user's preference tendency, and establish a user-specific user usage habit reference database.

[0136] The time prediction model is continuously learned and trained through the usage habit reference database. The longer the user uses the intelligent device, the more data the usage habit reference database obtains, and the closer the learning and training result of the time prediction model will be to the user's real usage habit and the user's preference tendency.

[0137] The model training module is used to classify users according to preset user tags, collect the usage habit reference database of at least one user under the same user tag, obtain a general database corresponding to the user tag, and train the time prediction model based on the general database to obtain a trained time prediction model.

[0138] The model training module is used to classify users according to preset user tags, collect the usage habit reference database of at least one user under the same type of user tag, obtain a general database corresponding to the user tag, classify users according to the preset user tags, and form designated user tag groups. For each user tag group, data is collected from the usage habit reference database corresponding to all users under the user tag group, and the intersection is performed based on user operation habits and usage intentions to establish a general database of the same type for the user tag group.

[0139] In this specific implementation, users are classified according to preset labels to form designated user label groups, data is collected from the usage habit reference database corresponding to all users under the user label group, and based on the intersection of habit feature data and preference tendency data, a general database of the same type for the user label group is established, and the intelligent promotion prediction time model for the user label group is continuously learned and trained based on the general database of the same type through the retraining model.

[0140] By setting labels for users, the basis for setting labels is determined according to specific needs, such as preference tendencies, social identity, gender, age, etc., data is collected from all habit reference databases of user groups under different labels, and the same habit feature data and preference tendencies that most users will have under the label are obtained, and the corresponding data is stored in a general database of the same type. After a new user uses the smart device for the first time, the new user's intelligent promotion prediction time model uses the shortest time to determine the label of the new user, and learns and trains the corresponding general database of the same type based on the label to adapt to the new user's usage habits in the shortest time. At the same time, the new user's non-privacy data is continuously obtained, and the new user's intelligent promotion prediction time model is continuously improved to improve the new user's usage experience.

[0141] The retraining model performs reverse learning on the user habits under the corresponding labels based on the same type of general database, predicts the usage habits of the user habit reference database, and performs previous performance of the habit feature when the user has not yet shown the habit or has rarely shown the habit. After uploading non-privacy data, it determines whether the user's predicted behavior for the habit feature is adapted to or consistent with the user's usage habits, and further studies and trains the habit reference database based on the judgment results.

[0142] The prediction module is used to predict the user operation interval time based on the obtained user operation habits and usage intentions through a trained time prediction model to obtain the promotion time, and predict the message push position based on the intelligent device display interface layout parameters, and push messages when the user uses the intelligent device based on the promotion time and the message push position.

[0143] In this embodiment, the prediction module is used to identify the layout of the intelligent device display interface and select the best position as the display position of the pushed message, avoiding the simple and violent display of traditional pushed messages in the center of the interface, which causes the user's aversion to the pushed information.

[0144] The above has detailedly demonstrated and described the present invention through the drawings and preferred embodiments. However, the present invention is not limited to these disclosed embodiments. Based on the above-mentioned multiple embodiments, those skilled in the art can know that more embodiments of the present invention can be obtained by combining the code review means in the above different embodiments, and these embodiments are also within the protection scope of the present invention.

Claims

1. A method for obtaining non-private data and pushing messages, characterized in that, It includes the following steps: Process the user data in the user file through a data segmentation algorithm to obtain private data and non-private data, and perform integrity detection on the private data, where the user data is the relevant data generated by the user using the smart device; Store the non-private data in the data transfer station, and use the data transfer station as the user's big data source. The cloud data center retrieves and processes the non-private data from the user's big data source in real time; Retrieve the user behavior data and interface data in the non-private data in real time, identify the user's current usage status, obtain the user operation frequency parameter and the smart device display interface layout parameter, analyze the user's usage behavior based on the user operation frequency parameter and the smart device display interface layout parameter, obtain the user operation habit and usage intention, and predict the user operation gap time; Construct a user usage habit reference database based on the user operation habit and usage intention, and establish a time prediction model. The time prediction model is used to predict and output the user operation gap data based on the user operation habit and usage intention. The user operation gap time is used as the promotion time; Classify the users according to the preset user tags, aggregate the usage habit reference databases of at least one user under the same user tag to obtain the general database corresponding to the user tag, and train the time prediction model based on the general database to obtain the trained time prediction model; Predict the user operation gap time based on the obtained user operation habit and usage intention through the trained time prediction model to obtain the promotion time, predict the message push position based on the smart device display interface layout parameter, and push messages when the user uses the smart device based on the promotion time and the message push position.

2. The method for obtaining non-private data and pushing messages according to claim 1, characterized in that, Perform integrity detection on the private data through the following steps: Obtain the location information of the private data in the user file; Establish a mapping table between the private data and the original file according to the private data and the location information, establish an index table according to the original file and the mapping table, and store the mapping table and the index table in the relational database at the same time; Detect whether the data content in the private data has been tampered with by detecting whether the digital digests before and after the data are consistent, so as to achieve the integrity detection of the private data content.

3. The method for obtaining non-private data and pushing messages according to claim 1, characterized in that, The cloud data center retrieves and processes the non-private data from the user's big data source in real time, and further includes the following steps: L1. Obtain all network search data through the non-private data. The network search data is the search terms entered by the user in each search engine, and construct a search term group based on the obtained search terms; L2. Obtain the search term group at the network data acquisition end L3. Select any search term in the search term group; L4. Obtain the key data in the search term group, including obtaining the name, usage keyword, and feature description word in the key data, and perform fitting analysis on the key data and the search term; L5. Select the next search term and repeat step L4; L6. Repeat step L5 until all search terms are processed to obtain a target term group composed of several target terms; L7. Retrieve relevant push content based on the target term group.

4. The method for obtaining non-private data and pushing messages according to claim 1, characterized in that, When analyzing the user's usage behavior based on the user operation frequency parameter and the intelligent device display interface layout parameter, analyze the operation frequency, operation content, and stay time corresponding to the user's use of the application based on the user operation frequency parameter, and analyze the user's demand degree for the application or the theme content of the interface and the user's usage intention. Among them, the user's usage intention includes the search frequency of the user for the specified keyword, its stay time, and the corresponding operation behavior data.

5. The method for obtaining non-private data and pushing messages according to claim 1, characterized in that, Classify users according to the preset user tags, aggregate the usage habit reference databases of at least one user under the same user tag to obtain the general database corresponding to the user tag, classify users according to the preset user tags to form a specified user tag group, and for each user tag group, collect data from the usage habit reference databases corresponding to all users under the user tag group, and perform an intersection based on the user operation habit and usage intention to establish a homogeneous general database for the user tag group.

6. A system for obtaining non-private data and pushing messages, characterized in that, For message pushing by using the non-privacy data acquisition and message pushing method according to any one of claims 1-5, the system includes: A data processing module, which is used to process the user data in the user file through a data segmentation algorithm to obtain privacy data and non-privacy data, and perform integrity detection on the privacy data, where the user data is the relevant data generated by the user using the intelligent device; A data transfer module, which is used to store the non-privacy data in the data transfer station and use the data transfer station as the user big data source, and the cloud data center retrieves and processes the non-privacy data from the user big data source in real time; A data analysis module, which is used to obtain the user behavior data and interface data in the non-privacy data in real time, identify the user's current usage status, obtain the user operation frequency parameter and the intelligent device display interface layout parameter, analyze the user's usage behavior based on the user operation frequency parameter and the intelligent device display interface layout parameter, obtain the user operation habit and usage intention, and predict the user operation gap time; A model construction module, which is used to construct a user usage habit reference database based on the user operation habit and usage intention, and establish a time prediction model, where the time prediction model is used to predict and output user operation gap data based on the user operation habit and usage intention, and the user operation gap time is used as the promotion time; A model training module, which is used to classify users according to the preset user tags, aggregate the usage habit reference databases of at least one user under the same user tag to obtain the general database corresponding to the user tag, and perform model training on the time prediction model based on the general database to obtain the trained time prediction model; A prediction module, which is used to predict the user operation interval time based on the obtained user operation habits and usage intentions through a trained time prediction model to obtain the promotion time, and predict the message push position based on the intelligent device display interface layout parameters, and push messages when the user uses the intelligent device based on the promotion time and the message push position.

7. The system for obtaining non-private data and pushing messages according to claim 6, characterized in that, The data processing module is used to perform integrity detection on the privacy data through the following steps: Obtain the location information of the privacy data in the user file; Establish a mapping table of the privacy data and the original file according to the privacy data and the location information, and establish an index table according to the original file and the mapping table, and at the same time store the mapping table and the index table in the relational database; Detect whether the data content in the privacy data has been tampered with by detecting whether the digital digests before and after the data are consistent, so as to achieve the integrity detection of the privacy data content.

8. The non-privacy data acquisition and message push system according to claim 6, wherein, The data transfer module is used to perform the following operations to enable the cloud data center to obtain and process non-privacy data from the user big data source in real time: L1. Obtain all network search data through the non-privacy data. The network search data is the search terms entered by the user in each search engine, and construct a search term group based on the obtained search terms; L2. Obtain the search term group at the network data acquisition end L3. Select an arbitrary search term within a search term group; L4. Obtain the key data within the search term group, including obtaining the name, usage keywords, and feature description words within the key data, and perform fitting analysis on the key data and the search terms; L5. Select the next search term and repeat step L4; L6. Repeat step L5 until all search terms are processed to obtain a target term group composed of several target terms; L7. Retrieve relevant push content based on the target term group.

9. The non-privacy data acquisition and message push system according to claim 6, wherein, When analyzing the user usage behavior based on the user operation frequency parameter and the intelligent device display interface layout parameter, the data analysis module is used to analyze the corresponding operation frequency, operation content, and stay time when the user uses the application program based on the user operation frequency parameter, and analyze the user's demand degree for the application program or the theme content of the interface and the user's usage intention. Among them, the user's usage intention includes the user's search frequency for the specified keyword, its stay time, and the corresponding operation behavior data.

10. The non-privacy data acquisition and message push system according to claim 6, wherein, The model training module is used to classify users according to the preset user tags, aggregate the usage habit reference databases of at least one user under the same user tag to obtain the general database corresponding to the user tag, classify users according to the preset user tags to form a specified user tag group, for each user tag group, collect the usage habit reference databases corresponding to all users under the user tag group, and perform an intersection based on the user operation habits and usage intentions to establish a homogeneous general database for the user tag group.

Images