Social network user alignment method, apparatus, and medium based on relevance information
By extracting relevant information from public information sources and using neural network models for automatic filtering and verification, the problem of low matching accuracy in social network user alignment is solved, achieving efficient and accurate user alignment and enhancing the practicality and scalability of the method.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NAT COMP NETWORK & INFORMATION SECURITY MANAGEMENT CENT
- Filing Date
- 2022-07-07
- Publication Date
- 2026-05-15
AI Technical Summary
Existing social network user alignment methods suffer from low matching accuracy, difficulty in verifying results, and lack of scalability and portability, especially due to the fragmentation of user data caused by the similarity, sparsity, and falsity of virtual information.
By extracting relevant information from public information sources and using a webpage parsing neural network model to automatically filter websites containing social network application account details, automated collection and verification are performed to align accounts of the same user.
It improves the accuracy and efficiency of social network user alignment, enhances the practicality and scalability of the method, and eliminates the need to pre-specify the social network applications to be matched.
Smart Images
Figure CN115186240B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of social media data mining, and particularly to a method, apparatus, and medium for aligning social network users based on correlation information. Background Technology
[0002] With the booming development of mobile internet and the increasingly diverse needs of people using mobile internet, various internet applications have emerged, forming a digital lifestyle centered on social networking, with different types of applications working together to meet the needs of different users. The same user may register accounts on multiple different social networking applications due to different usage needs. However, these different social networking applications are often independent of each other and lack sharing mechanisms, resulting in fragmented user data. This characteristic hinders social networking applications from providing effective personalized recommendations to newly joined users, leading to a poor user experience. Therefore, accurately identifying the virtual accounts of the same user across multiple social networking applications and efficiently integrating multi-source user information will provide strong support for commercial applications and recommendation system optimization.
[0003] Currently, the core strategy for aligning users on mainstream social networks is to leverage the similarity between virtual information (including static attribute information and dynamic social information) of different application accounts to achieve the goal of matching social network application accounts. However, the aforementioned virtual information of application accounts often has the following characteristics:
[0004] Similarity. This manifests both in the existence of a large number of accounts on social networks that have similar or identical attributes (such as similar or identical nicknames or profile pictures) but are not associated with the same user, and in the existence of a large number of accounts on social networks that have similar or identical interest circles and thus similar follower / fan relationship structures but are not associated with the same user.
[0005] Sparsity. From a user-friendly perspective, most social networking applications allow users to fill in other attribute information, except for a few necessary attributes. This results in sparse virtual information for application accounts, and very few attributes shared across applications.
[0006] Falsehood. Due to privacy concerns or malicious intent such as impersonation, social network users often fill in false and arbitrary attribute values when entering their information, resulting in false and inconsistent virtual information about their application accounts and low data usability.
[0007] The virtual information characteristics of the application accounts mentioned above result in low matching accuracy and difficulty in verifying the matching results of existing methods. In addition, existing technologies mostly require pre-specifying the application source of the virtual accounts to be matched, that is, they are only applicable to pre-specified social network applications to achieve matching and alignment of virtual accounts between them, which does not have good scalability and portability in practical applications. Summary of the Invention
[0008] To address the aforementioned technical problems, this invention provides a method, apparatus, and medium for aligning social network users based on correlation information.
[0009] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:
[0010] A social network user alignment method based on correlation information includes the following steps:
[0011] Step 1. Obtain user identity-related information from publicly available information sources;
[0012] Step 2. Search the social network based on the user's identity and related information, and save the retrieved web page snapshots;
[0013] Step 3. Filter the webpage snapshots to obtain target websites that may contain social networking application account information;
[0014] Step 4. Collect webpage information corresponding to target websites that are suspected of containing social network application account information, verify the collection results, and obtain the user's account information on social networks;
[0015] Step 5. Treat social network account information corresponding to the same related information as belonging to the same user, thereby achieving social network user alignment.
[0016] Furthermore, in step 1, the user identity association information refers to information that is associated with and corresponds to the identity of the social network account holder.
[0017] Furthermore, in step 3, a webpage parsing neural network model is constructed, and the trained webpage parsing neural network model is used to filter the webpage snapshots to obtain target websites that are suspected of containing social network account information.
[0018] Furthermore, step 3, before constructing the webpage parsing neural network model, also includes:
[0019] Determine the coordinates of the text content area and the image content area in the webpage snapshot;
[0020] Reconstruct the layout structure of the text and image content of the webpage snapshot, that is, to build an image of the same size with pixels filled with zero values for the webpage snapshot;
[0021] Based on the coordinates of the text content area and the coordinates of the image content area, the text content area and the image content area of each webpage snapshot are marked in the zero-value-filled image, and the layout image matrix of the text and image content of each webpage snapshot is obtained.
[0022] Furthermore, in step 3, the input layer of the webpage parsing neural network model is input with the layout image matrix of the text and images of each webpage snapshot, and the output layer uses softmax for binary classification.
[0023] Furthermore, in step 3, obtaining the trained webpage parsing neural network model includes: training the webpage parsing neural network model using a supervised training method.
[0024] Furthermore, the system checks whether each attribute field of the obtained results contains user association information. If it does, the user's account information on the social network is obtained.
[0025] Furthermore, step 5 includes: combining the verified social network account information with the currently retrieved user association information into a single data entry.
[0026] A social network user alignment device based on correlation information includes: a processor, a memory, and a program; the program is stored in the memory, and the processor calls the program stored in the memory to execute the social network user alignment method based on correlation information.
[0027] A computer-readable storage medium configured to store a program configured to execute the social network user alignment method based on association information.
[0028] Compared with the prior art, the present invention has the following technical effects:
[0029] This method extracts relevant information from publicly available sources, specifically information that is related to and corresponds to the identity of virtual account holders on social networking applications. Based on this, it retrieves the aforementioned relevant information and automatically filters the search results to identify potential target websites containing social networking application account information. By automatically collecting and verifying the details of these potential target websites, it obtains details of one or more social networking application accounts related to the aforementioned relevant information. Multiple social networking accounts strongly associated with the same person are considered to belong to the same user, ultimately achieving the goal of aligning social networking users. By building a webpage parsing neural network model, it can automatically filter potential target websites that may contain social networking application account details without requiring users to pre-specify the social networking applications to be matched. This method is more practical and scalable than existing technologies. Attached Figure Description
[0030] Figure 1 This is a flowchart of the social network user alignment method based on correlation information according to the present invention;
[0031] Figure 2This is a schematic diagram of the suspected target website screening process based on the webpage parsing neural network model of the present invention;
[0032] Figure 3 This is a schematic diagram of error data in this invention. Detailed Implementation
[0033] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.
[0034] Figure 1 This is a flowchart of a social network user alignment method based on correlation information according to the present invention.
[0035] Reference Figure 1 A social network user alignment method based on correlation information includes the following steps:
[0036] Step 1. Obtain user identity-related information from publicly available information sources.
[0037] We collect publicly available information source data and extract relevant information from it using regular expressions and other methods. A regular expression is a logical formula for string manipulation; it uses predefined specific characters and combinations of these characters to form a "rule string," which expresses a filtering logic for the string.
[0038] Related information refers to information that is associated with or corresponds to the identity of the social network account holder, such as the user's name and email address.
[0039] Step 2. Search the social network based on the user's identity and related information, and save the retrieved web page snapshots.
[0040] The system uses search engines to automatically retrieve related information from social networks and then takes screenshots of the retrieved web pages one by one.
[0041] Step 3. Filter the webpage snapshots to obtain target websites that may contain social networking application account information.
[0042] A webpage parsing neural network model is constructed and trained. The trained model is then used to filter webpage snapshots to obtain target webpage snapshots that are suspected of containing social network account information, thereby identifying the corresponding suspected target websites.
[0043] This step refers to Figure 2 , Figure 2This is a schematic diagram of the suspected target website screening process based on a webpage parsing neural network model, which includes the following steps:
[0044] Step 3.1 Determine the coordinates of the text content area and the image content area in the webpage snapshot.
[0045] To determine the coordinates of text and image content regions in a saved webpage snapshot, image text recognition and object detection techniques can be used. For example, an Attention OCR (Attention Optical Character Recognition, Natural Scene Text Detection and Recognition) model can be built to identify the coordinates of text content regions in the webpage snapshot. A YOLO (You Only Look Once) model can be built to identify the coordinates of image content regions in the webpage snapshot. The YOLO model is an object detection model used to find specific objects in an image. Object detection requires not only identifying the types of objects but also marking their locations.
[0046] Step 3.2 Reconstruct the layout structure of the text and image content of the webpage snapshot, that is, construct an image of the same size with pixels filled with zero values for the webpage snapshot; and mark the text content area and image content area of each webpage snapshot in the zero-value filled image according to the coordinates of the text content area and the coordinates of the image content area.
[0047] Since an image can be represented as a matrix, after marking the text content area and image content area of each webpage snapshot in the zero-filled image, the image is represented as a matrix, thus obtaining the image matrix of the text and image content layout of each webpage snapshot.
[0048] Step 3.3 Construct a webpage parsing neural network model and use the trained webpage parsing neural network model to determine whether the original webpage corresponding to the webpage snapshot to be parsed belongs to a suspected target website containing social network application account information.
[0049] In some embodiments, mainstream image processing models can be used to construct the web page parsing neural network model, such as Convolutional Neural Network (CNN). The structure of CNN can be divided into: input layer, convolutional layer, max pooling layer, fully connected layer, and output layer.
[0050] The constructed webpage parsing neural network model is trained to obtain a trained model. Supervised training, also known as supervised learning, is a machine learning method that learns or builds a pattern from training data and uses this pattern to predict new instances. The training data consists of input objects and expected outputs. The function's output can be a continuous value or a predicted classification label.
[0051] In this embodiment, the training set data consists of the text and image content layout images of webpage snapshots and their category labels. Category label 1 indicates that the original webpage corresponding to the current webpage snapshot is a social network application account details page, and 0 indicates that the original webpage corresponding to the current webpage snapshot is not a social network application account details page. The webpage parsing neural network model constructed by the training pair is trained to obtain the trained webpage parsing neural network model.
[0052] A trained webpage parsing neural network model is used to extract features and classify the text content and image layout of the webpage snapshots. The model takes the text and image layout matrix of each webpage snapshot as input and uses softmax for binary classification. The result is divided into two categories: the original webpage corresponding to the current snapshot contains social network account information, and the original webpage corresponding to the current snapshot does not contain social network account information. Specifically, the original webpage corresponding to the current snapshot is either a social network application account detail page or not. Based on the classification results, it can be determined whether the current input original webpage is a suspected target website containing social network application account information.
[0053] In some embodiments, a webpage parsing neural network model constructed using a convolutional neural network is trained on a training set. The input layer of the trained convolutional neural network model takes as input the text and image layout matrix of each webpage snapshot. Features are extracted through convolutional layers, and the pooling layer downsamples the convolutional results to reduce dimensionality. A fully connected layer further extracts high-level features from the features extracted by the convolutional layers, identifying discriminative features for classification. Softmax is then used for binary classification, categorizing the original webpage corresponding to the current webpage snapshot as containing social network account information or not containing such information. Specifically, the original webpage corresponding to the current webpage snapshot is either a social network application account detail page or not. Based on the classification results, it can be determined whether the original webpage corresponding to the current input is a suspected target website containing social network application account information.
[0054] Step 4. Collect the webpage information corresponding to the suspected target website, verify the collection results, and obtain the user's account information on social networks.
[0055] Automated data collection techniques are used to collect webpage details from suspected target websites. These techniques can employ existing technologies or tools, such as those based on webpage HTML (…).
[0056] Information extraction techniques based on Hyper Text Markup Language (HTL) or automated web scraping tools such as Octoparse can be used to collect details from target websites.
[0057] Figure 3 This is a schematic diagram of erroneous data in the present invention. To avoid such errors... Figure 3 The error scenario is as follows: Suppose a search engine retrieves related information M1, where M1 specifically refers to related information in the form of a pure numeric string. The URL (Uniform Resource Locator) of search result 1 contains M1, and the snapshot content of the webpage corresponding to search result 1 is determined to be a suspected target website. However, search result 1 only contains M1 in its URL; the social network account U1 it points to has no connection with the related information M1 itself. Therefore, this record is erroneous data and cannot be added to the database after verification.
[0058] Therefore, the results obtained from the detection need to be verified. It is necessary to determine whether each attribute field of the obtained results contains user association information. If it does, the user's account information on the social network is obtained. The verified social network account information and the retrieved association information are combined into a single data record and stored in the database. If it does not contain the information, it is considered erroneous data and cannot be saved to the database.
[0059] Step 5. Treat social network account information corresponding to the same related information as belonging to the same user, thereby achieving social network user alignment.
[0060] The verified social network account information is combined with the currently retrieved user association information into a single data entry. A search for a specific association information is then performed in the database; the search results represent the set of social network accounts associated with that association information. In other words, multiple social network accounts associated with the same association information are considered to belong to the same user. This completes the social network user alignment task.
[0061] This embodiment extracts information from publicly available information sources that is related to and corresponds to the user identity of the virtual account on a social network. Based on this, it retrieves the aforementioned related information and, by building a neural network model, automatically filters suspected target websites containing details of social network application accounts in the search results. It also automatically collects and verifies the details of these suspected target websites to obtain social network application account details related to the aforementioned related information. Compared to existing technologies, which often utilize user-created, highly repetitive virtual network identity information, this invention focuses on related information. This type of information is often relatively real and reliable, and has a strong correspondence with the user to whom the social network account belongs, making the social network user alignment method described in this embodiment more accurate and efficient. Furthermore, this embodiment, by building a webpage parsing neural network model, can automatically filter suspected target websites that may contain details of social network application accounts, without requiring users to pre-specify the social network application to be matched. This method is more practical and scalable than existing technologies.
[0062] In an embodiment of the present invention, a social network user alignment device based on association information is also provided, comprising: a processor, a memory, and a program; the program is stored in the memory, and the processor calls the program stored in the memory to execute the aforementioned social network user alignment method based on association information.
[0063] In the aforementioned implementation of a social network user alignment device based on correlation information, the memory and processor are electrically connected directly or indirectly to enable data transmission or interaction. For example, these components can be electrically connected to each other via one or more communication buses or signal lines, such as a bus connection. The memory stores computer-executable instructions that implement data access control methods, including at least one software functional module that can be stored in the memory in the form of software or firmware. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory.
[0064] The memory can be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), and Erasable Read-Only Memory (Erasable).
[0065] Programmable Read-Only Memory (EPROM), Electrically Erasable Read-Only Memory (EPROM)
[0066] Erasable Programmable Read-Only Memory (EEPROM), etc. The memory stores the program, and the processor executes the program after receiving the execution instruction.
[0067] A processor can be an integrated circuit chip with signal processing capabilities. The aforementioned processor can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor.
[0068] In an embodiment of the present invention, a computer-readable storage medium is also provided, which is configured to store a program configured to perform the above-described method for social network user alignment based on association information.
[0069] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, embodiments of the present invention can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, embodiments of the present invention can take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0070] The embodiments of the present invention are described with reference to flowchart illustrations and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the present invention. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, create means for implementing the functions specified in the flowcharts.
[0071] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in the flowchart.
[0072] These computer program instructions may also be loaded onto a computer or other programmable data processing terminal equipment to cause a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, such that the instructions that execute on the computer or other programmable terminal equipment provide steps for implementing the function specified in the flowchart.
[0073] The foregoing has provided a detailed description of the social network user alignment method, the social network user alignment device, and the application of a computer-readable storage medium based on correlation information provided by the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.
Claims
1. A method for aligning social network users based on relevance information, characterized in that, Includes the following steps: Step 1. Obtain user identity-related information from publicly available information sources; Step 2. Search the social network based on the user's identity and related information, and save the retrieved web page snapshots; Step 3. Filter the webpage snapshots to obtain target websites that may contain social networking application account information; In step 3, a webpage parsing neural network model is constructed, and the webpage snapshots are filtered through the trained webpage parsing neural network model to obtain target websites that may contain social network account information. Before constructing the webpage parsing neural network model, the following steps are also included: determining the coordinates of the text content region and the image content region in the webpage snapshot; reconstructing the layout structure of the text content and image content of the webpage snapshot, that is, constructing an image of the same size with pixels filled with zero values for the webpage snapshot; based on the coordinates of the text content region and the image content region, marking the text content region and image content region of each webpage snapshot in the zero-value-filled image, and obtaining the layout image matrix of the text and image content of each webpage snapshot. The input layer of the webpage parsing neural network model takes as input the image matrix of the text and image content layout of each webpage snapshot, and the output layer uses softmax for binary classification. Obtaining a well-trained neural network model for webpage parsing includes: training the neural network model for webpage parsing using supervised training methods; Step 4. Collect webpage information corresponding to target websites that are suspected of containing social network application account information, verify the collection results, and obtain the user's account information on social networks; Step 5. Treat social network account information corresponding to the same related information as belonging to the same user, thereby achieving social network user alignment.
2. The social network user alignment method based on correlation information according to claim 1, characterized in that, In step 1, the user identity association information refers to information that is associated with and corresponds to the identity of the social network account holder.
3. The social network user alignment method based on correlation information according to claim 1, characterized in that, In step 4, it is detected whether each attribute field of the obtained result contains user association information. If it does, the user's account information in the social network is obtained.
4. The social network user alignment method based on correlation information according to claim 1, characterized in that, Step 5 includes: combining the verified social network account information with the currently retrieved user association information into a single data entry.
5. A social network user alignment device based on correlation information, characterized in that, It includes: a processor, a memory, and a program; the program is stored in the memory, and the processor calls the program stored in the memory to execute the social network user alignment method based on association information as described in any one of claims 1-4.
6. A computer-readable storage medium, characterized in that: The computer-readable storage medium is configured to store a program configured to perform a social network user alignment method based on association information as described in any one of claims 1-4.