Social communication path identification method and device based on interactive behavior

Abstract

In the embodiment of the present invention, a method, device and electronic device for identifying social communication paths based on interactive behavior are provided, which belongs to the field of data processing technology. The method includes: collecting interactive behavior data of users through multiple social media platform API interfaces, and performing data cleaning and standardization processing on the interactive behavior; constructing an information communication network based on the user interactive behavior data, and calculating the weight of each contact surface in the network through a contact surface weight calculation function; constructing a communication feature vector based on the contact surface and the weight of the contact surface, and identifying and classifying the information communication path Ts based on the communication feature vector and the path calculation function Fl; analyzing the communication speed and key nodes of each social media platform through the cross-platform communication path feature vector G, and evaluating the contribution value of different social platforms in the process of information diffusion. The present invention can improve the accuracy and real-time performance of communication path identification, and at the same time has the ability to identify cross-platform communication paths.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method, device and electronic device for identifying a social communication path based on interactive behavior. Background Art

[0002] With the rapid development of social media, the speed and scale of information dissemination on the platform have increased significantly. Social media has become an important channel for information dissemination, especially in marketing, information management, and public event communication. Tracking and identifying the dissemination path of information is crucial for enterprises and institutions.

[0003] The following problems exist in the implementation of social communication path identification in the prior art:

[0004] 1. Lack of comprehensive consideration of multi-dimensional interactive behaviors: Traditional communication path identification methods mainly rely on simple communication chain analysis, which usually ignores the impact of users' interactive behaviors on information dissemination, resulting in inaccurate communication path identification.

[0005] 2. Difficulty in identifying propagation paths in complex networks: Social networks usually have complex nodes and connections. Existing methods find it difficult to quickly and accurately identify propagation paths in large-scale complex networks.

[0006] 3. Poor real-time performance: Existing propagation path identification methods often require post-analysis and cannot dynamically track the propagation path while information is being transmitted.

[0007] 4. Insufficient analysis capabilities for cross-platform communication: Information is often disseminated through multiple social media platforms, and existing methods make it difficult to track and identify dissemination paths across platforms.

[0008] The above problems become technical problems that need to be solved. Summary of the invention

[0009] In view of this, an embodiment of the present invention provides a method, device and electronic device for identifying a social communication path based on interactive behavior, which at least partially solves the problems existing in the prior art.

[0010] In a first aspect, an embodiment of the present invention provides a method for identifying a social communication path based on interactive behavior, comprising:

[0011] Collecting user interaction behavior data through multiple social media platform API interfaces, and performing data cleaning and standardization on the interaction behavior, wherein the interaction behavior data includes user ID, content ID, interaction type and timestamp;

[0012] An information propagation network is constructed based on user interaction behavior data, wherein the propagation network includes nodes and contact surfaces, and the weight of each contact surface in the network is calculated by a contact surface weight calculation function;

[0013] Based on the contact surface and the weight of the contact surface, a propagation feature vector Vs=[Lc, Ss, Dc, Iy, Ah, Hc] is constructed, and based on the propagation feature vector and the path calculation function Fl, the information propagation path Ts is identified and classified, where Lc represents the propagation path length, Ss represents the propagation speed, Dc represents the propagation level, Iy represents the node social influence, Ah represents the node activity, and Hc represents the frequency of interactive behavior;

[0014] When the newly generated user interaction behavior data exceeds the preset value θ, the propagation feature vector is updated to Vsd, and the updated information propagation path Tsd is calculated based on the updated propagation feature vector Vsd;

[0015] Analyze the information dissemination trajectory of the same user on multiple different social media platforms to form a cross-platform propagation path feature vector G=[Ts,Tsd]. Use the cross-platform propagation path feature vector G to analyze the propagation speed and key nodes of each social media platform and evaluate the contribution value of different social platforms in the information diffusion process.

[0016] According to a specific implementation of an embodiment of the present invention, the user's interactive behavior data is collected through multiple social media platform API interfaces, and the interactive behavior data is cleaned and standardized, including:

[0017] Collect interactive behavior data of social media users, including users’ forwarding, likes, comments and sharing data.

[0018] According to a specific implementation of an embodiment of the present invention, the method of collecting user interaction behavior data through multiple social media platform API interfaces and performing data cleaning and standardization processing on the interaction behavior also includes:

[0019] Clean the collected data to remove duplicate, invalid and noisy data;

[0020] Standardize the data, including unifying the time format, encoding the behavior type, and normalizing user identifiers.

[0021] According to a specific implementation of an embodiment of the present invention, the information propagation network is constructed based on user interaction behavior data, the propagation network includes nodes and contact surfaces, and the weight of each contact surface in the network is calculated by a contact surface weight calculation function, including:

[0022] Each user is represented as a node in the communication network. The attributes of the node include user ID, the type of behavior involved in the communication, and the interaction time.

[0023] The interactive behaviors between users are set as the contact surfaces of the propagation paths. Each contact surface connects two users, indicating that there is an information propagation relationship between the two users.

[0024] The weight W(i,j) represents the propagation weight between user i and user j. The weight W(i,j) of the contact surface is calculated by the following formula:

[0025]

[0026] f 转发 (i,j) represents the frequency of user i forwarding the content of user j, and f 评论 (i,j) represents the frequency of user i’s comments on user j’s content, f 点赞 (i,j) represents the frequency of user i’s likes on user j’s content, and α, β, and γ are weight parameters.

[0027] According to a specific implementation of an embodiment of the present invention, the propagation feature vector Vs=[Lc, Ss, Dc, Iy, Ah, Hc] is constructed based on the contact surface and the weight of the contact surface, and the information propagation path Ts is identified and classified based on the propagation feature vector and the path calculation function Fl, including:

[0028] Calculate the shortest path length from the information source node to the target node and obtain the propagation path length Lc:

[0029]

[0030] P(s,t) is the set of all paths from the source node s to the target node t, p is the propagation path, (u,v) is a contact surface on the path p, (u,v) is the weight of the contact surface (u,v), λ is a tuning parameter used to balance the influence of the contact surface weight and the node degree, and degree(u) is the degree of node u, indicating the number of contact surfaces connected to node u;

[0031] Based on the time interval for information to propagate from the source node to the target node, the propagation speed Ss is calculated:

[0032]

[0033] p is the shortest path from source node s to target node t, timestamp(v) is the timestamp of node v, and μ is a tuning parameter used to control the effect of contact surface weight on propagation speed;

[0034] Calculate the number of layers that the information passes through from the source node to the target node, and get the propagation level Dc:

[0035]

[0036] p is the shortest path from source node s to target node t, level(v) is the level number of node v, and η is a tuning parameter used to balance the influence of level difference and node degree;

[0037] Calculate the node social influence Iy:

[0038]

[0039] w1, w2, w3 are weight parameters, flr(i) is the number of users followed by node i, itf(i) is the interaction frequency of node i, hsd(i) is the historical communication record of node i, δ is the adjustment parameter, and ctt(i) is the centrality of node i.

[0040] Compute node activity Ah:

[0041]

[0042] w4, w5, w6 are weight parameters, itf(i) is the interaction frequency of node i, itd(i) is the interaction depth of node i, and atd(i) is the active duration of node i;

[0043] Based on the frequency of users performing interactive behaviors on the propagation path, the interactive behavior frequency Hc is calculated:

[0044]

[0045] f 转发 (i,j) is the frequency of user i forwarding the content of user j, f 评论 (i,j) is the frequency of user i’s comments on user j’s content, f 点赞 (i, j) is the frequency of user i's likes on user j's content, ρ is the adjustment parameter, and sts(i, j) is the sentiment score of user i on user j's content;

[0046] The calculated features are combined into a feature vector Vs=[Lc, Ss, Dc, Iy, Ah, Hc].

[0047] According to a specific implementation of an embodiment of the present invention, the method of constructing a propagation feature vector Vs=[Lc, Ss, Dc, Iy, Ah, Hc] based on the contact surface and the weight of the contact surface, identifying and classifying the information propagation path Ts based on the propagation feature vector and the path calculation function Fl, further includes:

[0048] Construct a training model that includes the first-layer model and the second-layer model. Use the stacking method to take the outputs of multiple base models as new features and input them into a meta-model for final classification. The first-layer meta-model is expressed as:

[0049]

[0050]

[0051]

[0052] P RF , P SVM and P XGB Represent the prediction results of random forest, support vector machine and XGBoost models respectively, and F RF 、F SVM and F XGB They represent random forest, support vector machine, and XGBoost models respectively;

[0053] The second-level metamodel is represented as:

[0054]

[0055]

[0056] Vm is the meta-feature vector, Fmeta is the meta-model, and LR is the binary classification function;

[0057] Using the prediction results of the base model P RF ,P SVM ,P XGB As the new meta-feature vector Vm, train the meta-model FMeta, and use the meta-model to perform the final classification Ts on the prediction results of the basic model:

[0058] .

[0059] According to a specific implementation method of an embodiment of the present invention, when the newly generated user interaction behavior data exceeds a preset value θ, the propagation feature vector is updated to Vsd, and an updated information propagation path Tsd is calculated based on the updated propagation feature vector Vsd, including:

[0060] When the amount of newly generated user interaction behavior data exceeds a preset value θ, the newly generated user interaction behavior data is collected and preprocessed;

[0061] Based on the updated data, the nodes and contact surfaces in the propagation network are reconstructed, and new contact surface weights are calculated;

[0062] According to the updated propagation network, the propagation feature vector is recalculated to obtain an updated propagation feature vector Vsd, and the updated propagation feature vector Vsd is input into the trained model to perform feature standardization and feature generation to generate a new feature vector;

[0063] The new feature vector is predicted using the basic model and meta-model to obtain the updated information propagation path Tsd.

[0064] According to a specific implementation of an embodiment of the present invention, analyzing the propagation speed and key nodes of each social media platform through the cross-platform propagation path feature vector G to evaluate the contribution value of different social platforms in the information diffusion process includes:

[0065] Analyze the information dissemination trajectory of the same user on multiple different social media platforms to form a cross-platform dissemination path feature vector G;

[0066] By analyzing the timestamp and propagation path, the time required for information to be transmitted from one node to another is determined, thereby evaluating the propagation speed Pts of each platform;

[0067] Calculate the activity value Phy of each node on different platforms during the propagation process;

[0068] Based on the propagation speed Pts and activity value Phy, the contribution value of different social platforms in the process of information diffusion is evaluated.

[0069] In a second aspect, an embodiment of the present invention provides a social communication path identification device based on interactive behavior, comprising:

[0070] A collection module collects user interaction behavior data through multiple social media platform API interfaces, and performs data cleaning and standardization on the interaction behavior. The interaction behavior data includes user ID, content ID, interaction type and timestamp;

[0071] A construction module, which constructs an information propagation network based on user interaction behavior data, wherein the propagation network includes nodes and contact surfaces, and the weight of each contact surface in the network is calculated by a contact surface weight calculation function;

[0072] The identification module constructs a propagation feature vector Vs=[Lc, Ss, Dc, Iy, Ah, Hc] based on the contact surface and the weight of the contact surface, and identifies and classifies the information propagation path Ts based on the propagation feature vector and the path calculation function Fl, where Lc represents the propagation path length, Ss represents the propagation speed, Dc represents the propagation level, Iy represents the node social influence, Ah represents the node activity, and Hc represents the frequency of interactive behavior;

[0073] An updating module, when the newly generated user interaction behavior data exceeds a preset value θ, updates the propagation feature vector to Vsd, and calculates an updated information propagation path Tsd based on the updated propagation feature vector Vsd;

[0074] The evaluation module analyzes the information dissemination trajectory of the same user on multiple different social media platforms to form a cross-platform propagation path feature vector G=[Ts,Tsd]. The cross-platform propagation path feature vector G is used to analyze the propagation speed and key nodes of each social media platform to evaluate the contribution value of different social platforms in the information diffusion process.

[0075] In a third aspect, an embodiment of the present invention further provides an electronic device, the electronic device comprising:

[0076] at least one processor; and,

[0077] a memory communicatively connected to the at least one processor; wherein,

[0078] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the social communication path identification method based on interactive behavior in any of the aforementioned first aspects or any of the implementations of the first aspect.

[0079] In a fourth aspect, an embodiment of the present invention further provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable the computer to execute the social communication path identification method based on interactive behavior in the aforementioned first aspect or any implementation method of the first aspect.

[0080] In the fifth aspect, an embodiment of the present invention further provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer executes the social communication path identification method based on interactive behavior in the aforementioned first aspect or any implementation method of the first aspect.

[0081] The social communication path identification scheme based on interactive behavior in the embodiment of the present invention includes: collecting user interactive behavior data through multiple social media platform API interfaces, and performing data cleaning and standardization processing on the interactive behavior, wherein the interactive behavior data includes user ID, content ID, interaction type and timestamp; constructing an information propagation network based on user interactive behavior data, wherein the propagation network includes nodes and contact surfaces, and calculating the weight of each contact surface in the network through a contact surface weight calculation function; constructing a propagation feature vector Vs=[Lc, Ss, Dc, Iy, Ah, Hc] based on the contact surfaces and the weights of the contact surfaces, and identifying and classifying information propagation based on the propagation feature vector and the path calculation function Fl. Path Ts, Lc represents the length of the propagation path, Ss represents the propagation speed, Dc represents the propagation level, Iy represents the node social influence, Ah represents the node activity, and Hc represents the frequency of interactive behavior; when the newly generated user interactive behavior data exceeds the preset value θ, the propagation feature vector is updated to Vsd, and the updated information propagation path Tsd is calculated based on the updated propagation feature vector Vsd; the information propagation trajectory of the same user on multiple different social media platforms is analyzed to form a cross-platform propagation path feature vector G=[Ts,Tsd], and the propagation speed and key nodes of each social media platform are analyzed through the cross-platform propagation path feature vector G, and the contribution value of different social platforms in the information diffusion process is evaluated. The present invention has the following beneficial effects:

[0082] aMulti-dimensional interactive behavior analysis:

[0083] By comprehensively analyzing users' multi-dimensional interactive behaviors such as forwarding, commenting, and likes, the system can more accurately identify social communication paths and improve the accuracy of communication path identification.

[0084] bReal-time propagation path tracking:

[0085] The present invention supports real-time monitoring and dynamic adjustment of the information propagation process, and can identify and track the propagation path while the information is propagating.

[0086] cCross-platform communication path integration:

[0087] The system can integrate data from multiple social media platforms and achieve unified identification of cross-platform communication paths, making up for the shortcomings of existing methods in cross-platform communication analysis capabilities.

[0088] d Propagation key node identification:

[0089] Through machine learning technology, the system can identify key node users for information dissemination and provide data support for information diffusion and crisis response.

[0090] Easy to deploy and expand:

[0091] The present invention is based on a general data processing and analysis framework, can flexibly adapt to social media platforms of different sizes, and is easy to deploy and expand. BRIEF DESCRIPTION OF THE DRAWINGS

[0092] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0093] Figure 1 A schematic diagram of a flow chart of a method for identifying a social communication path based on interactive behavior provided by an embodiment of the present invention;

[0094] Figure 2 A schematic flow chart of another method for identifying a social communication path based on interactive behavior provided by an embodiment of the present invention;

[0095] Figure 3 A schematic diagram of the structure of a social communication path identification device based on interactive behavior provided by an embodiment of the present invention;

[0096] Figure 4 A schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0097] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0098] The following describes the embodiments of the present disclosure through specific examples, and those skilled in the art can easily understand other advantages and effects of the present disclosure from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all of the embodiments. The present disclosure can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present disclosure. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present disclosure.

[0099] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein may be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present disclosure, it should be understood by those skilled in the art that an aspect described herein may be implemented independently of any other aspect, and two or more of these aspects may be combined in various ways. For example, any number of aspects described herein may be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein may be used to implement this device and / or practice this method.

[0100] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present disclosure. The drawings only show components related to the present disclosure rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0101] Additionally, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, it will be understood by those skilled in the art that the aspects described may be practiced without these specific details.

[0102] The disclosed embodiment provides a method for identifying a social communication path based on interactive behavior. The method for identifying a social communication path based on interactive behavior provided in the embodiment can be executed by a computing device, which can be implemented as software, or as a combination of software and hardware, and which can be integrated in a server, terminal device, etc.

[0103] See also Figure 1 and Figure 2 , the present disclosure embodiment provides a method for identifying a social communication path based on interactive behavior, including:

[0104] S101, collecting user interaction behavior data through multiple social media platform API interfaces, and performing data cleaning and standardization processing on the interaction behavior, wherein the interaction behavior data includes user ID, content ID, interaction type and timestamp.

[0105] Collect social media users’ interactive behavior data through social media platform API interfaces or web crawler technology, including users’ forwarding, liking, commenting, sharing and other behaviors. The collected information includes user ID, content ID, interaction type, timestamp, etc.

[0106] Clean the collected data to remove duplicate, invalid and noisy data to ensure the accuracy of propagation path identification. Noise data usually includes machine accounts, invalid interactions (such as meaningless comments), etc.

[0107] In order to unify user identification and behavior records on different platforms, the system standardizes the data, including time format unification, behavior type coding, user identifier normalization, etc., to ensure that cross-platform data can be consistently integrated and analyzed.

[0108] S102, constructing an information propagation network based on user interaction behavior data, wherein the propagation network includes nodes and contact surfaces, and calculating the weight of each contact surface in the network through a contact surface weight calculation function.

[0109] You can set up a propagation path construction module, which is responsible for building an information propagation network based on the user's interactive behavior data. The network contains nodes (users) and contact surfaces (interactive behaviors):

[0110] A node definition:

[0111] Each user is represented as a node in the communication network. The attributes of the node include user ID, type of behavior involved in the communication (such as forwarding, commenting, liking, etc.), and interaction time.

[0112] b Definition and weighting of contact surface:

[0113] The interaction between users constitutes the contact surface of the communication path. Each contact surface connects two users, indicating that there is an information communication relationship between the two users. The weight of the contact surface can be weighted according to the type and frequency of the interaction behavior:

[0114]

[0115] Among them, 𝑊(𝑖,𝑗) represents the communication weight between user 𝑖 and user 𝑗, 𝑓forward, 𝑓comment and 𝑓like are the frequencies of forwarding, commenting and liking of user 𝑖’s content on user 𝑗, 𝛼, 𝛽 and 𝛾 are the weight parameters of different behaviors, which can be adjusted according to the actual scenario.

[0116] c Timing propagation chain:

[0117] Based on the timestamps of the interactive behaviors, the system constructs the communication chain in chronological order to ensure that the communication path can reflect the real direction of information flow. Early interactive behaviors will be recorded first to construct the initial communication path of the information.

[0118] S103, based on the contact surface and the weight of the contact surface, construct a propagation feature vector Vs=[Lc, Ss, Dc, Iy, Ah, Hc], and identify and classify the information propagation path Ts based on the propagation feature vector and the path calculation function Fl, where Lc represents the propagation path length, Ss represents the propagation speed, Dc represents the propagation level, Iy represents the social influence of the node, Ah represents the node activity, and Hc represents the frequency of interactive behavior.

[0119] The propagation path identification module is one of the cores of the present invention. It uses machine learning technology to identify the propagation path of information and dynamically track the propagation process:

[0120] The identification of propagation paths can be achieved through algorithms such as the shortest path and maximum flow path in graph theory. The specific steps include:

[0121] Shortest propagation path: Use the shortest path algorithm (such as Dijkstra algorithm) to identify the shortest path for information to propagate from the originating user to other users. The shortest path can help identify the most efficient path for propagation and is often used to track viral transmission.

[0122] The most influential communication path: Use the maximum flow path algorithm (such as the Ford-Fulkerson algorithm) to identify the most influential path in the information dissemination process. This path often passes through the most influential users and is suitable for tracking the core disseminators of information.

[0123] Key node analysis: Identify key users of information dissemination by calculating key nodes in the network (such as betweenness centrality, degree centrality, etc.). Users with high betweenness centrality are usually bridges in the dissemination chain and play a key role in the dissemination of information across social networks.

[0124] In the present invention, the machine learning model is used to identify the propagation path of information on social media. By analyzing the interactive behavior characteristics of users, the system can effectively capture the characteristics of the propagation path, and then realize path classification and prediction. The specific implementation steps are as follows:

[0125] aFeature extraction:

[0126] In order to effectively identify the propagation path, it is first necessary to extract features from the data generated during the information propagation process. Features can be divided into the following categories:

[0127] Path structure characteristics:

[0128] Propagation path length: the shortest path or maximum propagation path length of information from the starting node to the end node. A longer path usually means that the information is spread more widely.

[0129] Propagation speed: The speed at which information propagates between nodes is calculated using timestamps. A path with a faster propagation speed may indicate that the path has a higher propagation potential.

[0130] Diffusion level: The number of diffusion levels that information passes through. A deeper diffusion level usually indicates that the information spreads on multiple levels.

[0131] Node features:

[0132] Social network influence of a node: The influence of a node in a social network is evaluated by the number of users it follows, the frequency of interactions, historical communication records, etc. Nodes with high influence usually significantly accelerate the spread of information.

[0133] Node activity: the frequency of a node’s participation in interactions (forwarding, commenting, liking, etc.), as well as the depth of its interaction with other nodes. Users with high activity levels may become key nodes for dissemination.

[0134] Interactive behavior characteristics:

[0135] Behavior frequency: the frequency with which users perform interactive behaviors (forwarding, liking, commenting) on ​​a certain communication path. This can help determine the main form of information dissemination.

[0136] Behavior type distribution: Different interactive behaviors have different effects on communication. Forwarding often directly spreads information, while likes and comments enhance users’ sense of participation and recognition, indirectly affecting communication effects.

[0137] b Feature vector construction

[0138] After extracting the above features, the system converts each propagation path into a feature vector. The feature vector is used to characterize the overall characteristics of the propagation path and serves as the input of the machine learning model.

[0139] Assume that there are 𝑛 propagation paths, and the characteristics of each path are expressed as:

[0140]

[0141] in:

[0142] 𝐿𝑖 represents the length of the propagation path 𝑖.

[0143] 𝑆𝑖 represents the propagation speed.

[0144] 𝐷𝑖 represents the propagation level.

[0145] 𝐼𝑖1,𝐼𝑖2,… represent the social influence characteristics of each node in the path.

[0146] 𝐴𝑖1,𝐴𝑖2,… represent the activity characteristics of each node in the path.

[0147] Through these feature vectors, the system can capture the structural and behavioral pattern characteristics of the propagation path, thereby providing training data for subsequent machine learning models.

[0148] c Propagation path classification

[0149] After obtaining the feature vector, the system uses a supervised learning algorithm to classify the propagation path. You can choose from a variety of classification algorithms, such as random forest, support vector machine (SVM) or XGBoost. Let's take random forest as an example, the specific classification process is as follows:

[0150] Random Forest Model Training:

[0151] Random forest is an ensemble learning method based on decision trees. It generates multiple decision trees and combines the output results of these trees for classification. The training process is as follows:

[0152] Dataset construction: The training set is constructed using the propagation path data after feature extraction and vectorization. The training set consists of 𝑋𝑖, and the label 𝑌𝑖 represents the category of the propagation path (such as high propagation path, low propagation path, malicious propagation path, etc.).

[0153] Decision tree generation: For each decision tree, Random Forest randomly extracts samples and feature subsets from the dataset and builds a decision tree. Each decision tree is trained on these samples to find the decision rule that can best distinguish between different categories of propagation paths.

[0154] Model integration: Random forest uses a voting mechanism of multiple decision trees to determine the final propagation path classification result. Specifically, the propagation path 𝑋𝑖 is voted through the judgment results of all decision trees to obtain the final classification.

[0155] Model training optimization: The hyperparameters of random forest (such as the number of trees, maximum depth, etc.) are optimized through methods such as cross-validation to ensure that the model has good generalization ability.

[0156] Propagation path classification results:

[0157] After the model training is completed, the random forest can classify the new propagation path. The system will input the feature vector of the new propagation path and output its propagation category through the random forest model. For example:

[0158] Highly propagable path: A path that passes through influential and active nodes with fast propagation speed and long propagation path.

[0159] Low propagation path: The node influence is weak, the propagation speed is slow, the path is short, and the information propagation is limited.

[0160] Malicious propagation paths: paths that may contain a large amount of false information, machine accounts, or abnormal interactive behaviors.

[0161] Through such classification, the system can identify the information most likely to be widely disseminated and the potential malicious dissemination paths, thereby helping platform managers or companies to intervene in a timely manner.

[0162] S104, when the newly generated user interaction behavior data exceeds the preset value θ, the propagation feature vector is updated to Vsd, and the updated information propagation path Tsd is calculated based on the updated propagation feature vector Vsd.

[0163] Since the dissemination behavior of social media changes dynamically in real time, the present invention introduces an incremental learning mechanism to ensure that the model can adjust the classification and recognition results of the propagation path at any time. The specific steps are as follows:

[0164] aIncremental data processing:

[0165] When new user interaction behaviors (such as new forwarding, commenting, liking, etc.) occur, for example, when the newly generated user interaction behavior data exceeds the preset value θ, the system will instantly update the feature vector of the propagation path and recalculate its propagation path.

[0166] bIncremental learning update model:

[0167] Using the newly generated propagation path data, the random forest model is incrementally learned to gradually update the model's weights and decision rules without retraining the entire model, ensuring the system's real-time responsiveness.

[0168] cReal-time transmission path adjustment:

[0169] The system dynamically updates classification results based on new propagation paths, identifies potential fast-spreading paths and malicious propagation paths, and issues alarms.

[0170] S105, analyze the information dissemination trajectory of the same user on multiple different social media platforms to form a cross-platform propagation path feature vector G=[Ts,Tsd]. Use the cross-platform propagation path feature vector G to analyze the propagation speed and key nodes of each social media platform and evaluate the contribution value of different social platforms in the information diffusion process.

[0171] In order to identify the information propagation path across platforms, the present invention realizes data fusion across multiple platforms. The specific steps are as follows:

[0172] aUser ID fusion:

[0173] Identify the same user across platforms by matching public user information (such as user name, email address, mobile phone number) and behavioral characteristics.

[0174] bCross-platform propagation path merging:

[0175] The propagation data collected on different platforms are integrated to generate cross-platform propagation path feature vectors, and the trained machine learning model is used for classification and path identification.

[0176] cPlatform influence analysis:

[0177] By analyzing the propagation speed, key nodes and other characteristics on each platform, the system can evaluate the role of different platforms in the information diffusion process and identify users and paths that play a key role in the dissemination of information on multiple platforms.

[0178] The result display and analysis module can be used to present the results of propagation path identification to users in an intuitive way and generate a detailed propagation analysis report:

[0179] The system uses chart tools (such as D3.js, Gephi, etc.) to visualize the information propagation path. In the propagation network diagram, nodes represent users, and contact surfaces represent the interactive behaviors between users. The size of the node can be adjusted according to the user's propagation influence, and the thickness of the contact surface can be adjusted according to the propagation frequency or weight.

[0180] The system generates a propagation analysis report based on the propagation path identification results. The report includes the starting node of the propagation, key propagation nodes, path length, cross-platform propagation status, etc., and provides trend analysis and diffusion prediction.

[0181] When a propagation path shows an abnormally high propagation speed or may cause malicious propagation, the system will issue an alarm to help users respond in time.

[0182] The present invention has the following beneficial effects:

[0183] a Multi-dimensional interactive behavior analysis:

[0184] By comprehensively analyzing users' multi-dimensional interactive behaviors such as forwarding, commenting, and likes, the system can more accurately identify social communication paths and improve the accuracy of communication path identification.

[0185] bReal-time propagation path tracking:

[0186] The present invention supports real-time monitoring and dynamic adjustment of the information propagation process, and can identify and track the propagation path while the information is propagating.

[0187] cCross-platform communication path integration:

[0188] The system can integrate data from multiple social media platforms and achieve unified identification of cross-platform communication paths, making up for the shortcomings of existing methods in cross-platform communication analysis capabilities.

[0189] d Propagation key node identification:

[0190] Through machine learning technology, the system can identify key node users for information dissemination and provide data support for information diffusion and crisis response.

[0191] Easy to deploy and expand:

[0192] The present invention is based on a general data processing and analysis framework, can flexibly adapt to social media platforms of different sizes, and is easy to deploy and expand.

[0193] According to a specific implementation of an embodiment of the present invention, the user's interactive behavior data is collected through multiple social media platform API interfaces, and the interactive behavior data is cleaned and standardized, including:

[0194] Collect interactive behavior data of social media users, including users’ forwarding, likes, comments and sharing data.

[0195] According to a specific implementation of an embodiment of the present invention, the method of collecting user interaction behavior data through multiple social media platform API interfaces and performing data cleaning and standardization processing on the interaction behavior also includes:

[0196] Clean the collected data to remove duplicate, invalid and noisy data;

[0197] Standardize the data, including unifying the time format, encoding the behavior type, and normalizing user identifiers.

[0198] According to a specific implementation of an embodiment of the present invention, the information propagation network is constructed based on user interaction behavior data, the propagation network includes nodes and contact surfaces, and the weight of each contact surface in the network is calculated by a contact surface weight calculation function, including:

[0199] Each user is represented as a node in the communication network. The attributes of the node include user ID, the type of behavior involved in the communication, and the interaction time.

[0200] The interactive behaviors between users are set as the contact surfaces of the propagation paths. Each contact surface connects two users, indicating that there is an information propagation relationship between the two users.

[0201] The weight W(i,j) represents the propagation weight between user i and user j. The weight W(i,j) of the contact surface is calculated by the following formula:

[0202]

[0203] f 转发 (i,j) represents the frequency of user i forwarding the content of user j, and f评论 (i,j) represents the frequency of user i’s comments on user j’s content, f 点赞 (i,j) represents the frequency of user i’s likes on user j’s content, and α, β, and γ are weight parameters.

[0204] According to a specific implementation of an embodiment of the present invention, the propagation feature vector Vs=[Lc, Ss, Dc, Iy, Ah, Hc] is constructed based on the contact surface and the weight of the contact surface, and the information propagation path Ts is identified and classified based on the propagation feature vector and the path calculation function Fl, including:

[0205] Calculate the shortest path length from the information source node to the target node and obtain the propagation path length Lc:

[0206]

[0207] P(s,t) is the set of all paths from the source node s to the target node t, p is the propagation path, (u,v) is a contact surface on the path p, (u,v) is the weight of the contact surface (u,v), λ is a tuning parameter used to balance the influence of the contact surface weight and the node degree, and degree(u) is the degree of node u, indicating the number of contact surfaces connected to node u;

[0208] Based on the time interval for information to propagate from the source node to the target node, the propagation speed Ss is calculated:

[0209]

[0210] p is the shortest path from source node s to target node t, timestamp(v) is the timestamp of node v, and μ is a tuning parameter used to control the effect of contact surface weight on propagation speed;

[0211] Calculate the number of layers that the information passes through from the source node to the target node, and get the propagation level Dc:

[0212]

[0213] p is the shortest path from source node s to target node t, level(v) is the level number of node v, and η is a tuning parameter used to balance the influence of level difference and node degree;

[0214] Calculate the node social influence Iy:

[0215]

[0216] w1, w2, w3 are weight parameters, flr(i) is the number of users followed by node i, itf(i) is the interaction frequency of node i, hsd(i) is the historical communication record of node i, δ is the adjustment parameter, and ctt(i) is the centrality of node i.

[0217] Compute node activity Ah:

[0218]

[0219] w4, w5, w6 are weight parameters, itf(i) is the interaction frequency of node i, itd(i) is the interaction depth of node i, and atd(i) is the active duration of node i;

[0220] Based on the frequency of users performing interactive behaviors on the propagation path, the interactive behavior frequency Hc is calculated:

[0221]

[0222] f 转发 (i,j) is the frequency of user i forwarding the content of user j, f 评论 (i,j) is the frequency of user i’s comments on user j’s content, f 点赞 (i, j) is the frequency of user i's likes on user j's content, ρ is the adjustment parameter, and sts(i, j) is the sentiment score of user i on user j's content;

[0223] The calculated features are combined into a feature vector Vs=[Lc, Ss, Dc, Iy, Ah, Hc].

[0224] According to a specific implementation of an embodiment of the present invention, the method of constructing a propagation feature vector Vs=[Lc, Ss, Dc, Iy, Ah, Hc] based on the contact surface and the weight of the contact surface, identifying and classifying the information propagation path Ts based on the propagation feature vector and the path calculation function Fl, further includes:

[0225] Construct a training model that includes the first-layer model and the second-layer model. Use the stacking method to take the outputs of multiple base models as new features and input them into a meta-model for final classification. The first-layer meta-model is expressed as:

[0226]

[0227]

[0228]

[0229] P RF , P SVM and P XGBRepresent the prediction results of random forest, support vector machine and XGBoost models respectively, and F RF 、F SVM and F XGB They represent random forest, support vector machine, and XGBoost models respectively;

[0230] The second-level metamodel is represented as:

[0231]

[0232]

[0233] Vm is the meta-feature vector, Fmeta is the meta-model, and LR is the binary classification function;

[0234] Using the prediction results of the base model P RF ,P SVM ,P XGB As the new meta-feature vector Vm, train the meta-model FMeta, and use the meta-model to perform the final classification Ts on the prediction results of the basic model:

[0235] .

[0236] According to a specific implementation method of an embodiment of the present invention, when the newly generated user interaction behavior data exceeds a preset value θ, the propagation feature vector is updated to Vsd, and an updated information propagation path Tsd is calculated based on the updated propagation feature vector Vsd, including:

[0237] When the amount of newly generated user interaction behavior data exceeds a preset value θ, the newly generated user interaction behavior data is collected and preprocessed;

[0238] Based on the updated data, the nodes and contact surfaces in the propagation network are reconstructed, and new contact surface weights are calculated;

[0239] According to the updated propagation network, the propagation feature vector is recalculated to obtain an updated propagation feature vector Vsd, and the updated propagation feature vector Vsd is input into the trained model to perform feature standardization and feature generation to generate a new feature vector;

[0240] The new feature vector is predicted using the basic model and meta-model to obtain the updated information propagation path Tsd.

[0241] According to a specific implementation of an embodiment of the present invention, analyzing the propagation speed and key nodes of each social media platform through the cross-platform propagation path feature vector G to evaluate the contribution value of different social platforms in the information diffusion process includes:

[0242] Analyze the information dissemination trajectory of the same user on multiple different social media platforms to form a cross-platform dissemination path feature vector G;

[0243] By analyzing the timestamp and propagation path, the time required for information to be transmitted from one node to another is determined, thereby evaluating the propagation speed Pts of each platform;

[0244] Calculate the activity value Phy of each node on different platforms during the propagation process;

[0245] Based on the propagation speed Pts and activity value Phy, the contribution value of different social platforms in the process of information diffusion is evaluated.

[0246] Corresponding to the above method embodiment, see Figure 3 The embodiment of the present invention further discloses a social communication path identification device 30 based on interactive behavior, comprising:

[0247] The collection module 301 collects the user's interactive behavior data through multiple social media platform API interfaces, and performs data cleaning and standardization processing on the interactive behavior, wherein the interactive behavior data includes user ID, content ID, interaction type and timestamp;

[0248] A construction module 302 is used to construct an information propagation network based on the user interaction behavior data, wherein the information propagation network includes nodes and contact surfaces, and the weight of each contact surface in the network is calculated by a contact surface weight calculation function;

[0249] The identification module 303 constructs a propagation feature vector Vs=[Lc, Ss, Dc, Iy, Ah, Hc] based on the contact surface and the weight of the contact surface, and identifies and classifies the information propagation path Ts based on the propagation feature vector and the path calculation function Fl, where Lc represents the propagation path length, Ss represents the propagation speed, Dc represents the propagation level, Iy represents the node social influence, Ah represents the node activity, and Hc represents the frequency of interactive behavior;

[0250] The updating module 304 updates the propagation feature vector to Vsd when the newly generated user interaction behavior data exceeds the preset value θ, and calculates the updated information propagation path Tsd based on the updated propagation feature vector Vsd.

[0251] Evaluation module 305 analyzes the information dissemination trajectory of the same user on multiple different social media platforms to form a cross-platform propagation path feature vector G=[Ts,Tsd]. Through the cross-platform propagation path feature vector G, the propagation speed and key nodes of each social media platform are analyzed to evaluate the contribution value of different social platforms in the information diffusion process.

[0252] See also Figure 4The embodiment of the present invention further provides an electronic device 60, the electronic device comprising:

[0253] at least one processor; and,

[0254] a memory communicatively connected to the at least one processor; wherein,

[0255] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the social communication path identification method based on interactive behavior in the aforementioned method embodiment.

[0256] An embodiment of the present invention further provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable the computer to execute the aforementioned method embodiment.

[0257] An embodiment of the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer executes the social communication path identification method based on interactive behavior in the aforementioned method embodiment.

[0258] Reference below Figure 4 , which shows a schematic diagram of the structure of an electronic device 60 suitable for implementing the embodiment of the present disclosure. The electronic device in the embodiment of the present disclosure may include, but is not limited to, mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle terminals (such as vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 4 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0259] like Figure 4 As shown, the electronic device 60 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the electronic device 60 are also stored. The processing device 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0260] Typically, the following devices may be connected to the I / O interface 605: an input device 606 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, a gage, a gyroscope, etc.; an output device 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 608 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 609. The communication device 609 may allow the electronic device 60 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 4 The electronic device 60 is shown with various devices, but it should be understood that it is not required to implement or have all the devices shown. More or fewer devices may be implemented or have instead.

[0261] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device 609, or installed from a storage device 608, or installed from a ROM 602. When the computer program is executed by the processing device 601, the above-mentioned functions defined in the method of the embodiment of the present disclosure are executed.

[0262] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A social communication path identification method based on interactive behavior, characterized in that: include: Collecting user interaction behavior data through multiple social media platform API interfaces, and performing data cleaning and standardization on the interaction behavior, wherein the interaction behavior data includes user ID, content ID, interaction type and timestamp; An information propagation network is constructed based on user interaction behavior data, wherein the propagation network includes nodes and contact surfaces, and the weight of each contact surface in the network is calculated by a contact surface weight calculation function; Based on the contact surface and the weight of the contact surface, a propagation feature vector Vs=[Lc, Ss, Dc, Iy, Ah, Hc] is constructed, and based on the propagation feature vector and the path calculation function Fl, the information propagation path Ts is identified and classified, where Lc represents the propagation path length, Ss represents the propagation speed, Dc represents the propagation level, Iy represents the node social influence, Ah represents the node activity, and Hc represents the frequency of interactive behavior; When the newly generated user interaction behavior data exceeds the preset value θ, the propagation feature vector is updated to Vsd, and the updated information propagation path Tsd is calculated based on the updated propagation feature vector Vsd; Analyze the information dissemination trajectory of the same user on multiple different social media platforms to form a cross-platform propagation path feature vector G=[Ts,Tsd]. Use the cross-platform propagation path feature vector G to analyze the propagation speed and key nodes of each social media platform and evaluate the contribution value of different social platforms in the information diffusion process. The method of constructing a propagation feature vector Vs=[Lc, Ss, Dc, Iy, Ah, Hc] based on the contact surface and the weight of the contact surface, and identifying and classifying the information propagation path Ts based on the propagation feature vector and the path calculation function Fl includes: Calculate the shortest path length from the information source node to the target node and get the propagation path length Lc: P(s,t) is the set of all paths from the source node s to the target node t, p is the propagation path, (u,v) is a contact surface on the path p, (u,v) is the weight of the contact surface (u,v), λ is a tuning parameter used to balance the influence of the contact surface weight and the node degree, and degree(u) is the degree of node u, indicating the number of contact surfaces connected to node u; Based on the time interval for information to propagate from the source node to the target node, the propagation speed Ss is calculated: p is the shortest path from source node s to target node t, timestamp(v) is the timestamp of node v, and μ is a tuning parameter used to control the effect of contact surface weight on propagation speed; Calculate the number of layers that the information passes through from the source node to the target node, and get the propagation level Dc: p is the shortest path from source node s to target node t, level(v) is the level number of node v, and η is a tuning parameter used to balance the influence of level difference and node degree; Calculate the node social influence Iy: w1, w2, w3 are weight parameters, flr(i) is the number of users followed by node i, itf(i) is the interaction frequency of node i, hsd(i) is the historical communication record of node i, δ is the adjustment parameter, and ctt(i) is the centrality of node i. Compute node activity Ah: w4, w5, w6 are weight parameters, itf(i) is the interaction frequency of node i, itd(i) is the interaction depth of node i, and atd(i) is the active duration of node i; Based on the frequency of users performing interactive behaviors on the propagation path, the interactive behavior frequency Hc is calculated: f 转发 (i,j) is the frequency of user i forwarding the content of user j, f 评论 (i,j) is the frequency of user i’s comments on user j’s content, f 点赞 (i, j) is the frequency of user i's likes on user j's content, ρ is the adjustment parameter, and sts(i, j) is the sentiment score of user i on user j's content; Combine the calculated features into a feature vector Vs=[Lc, Ss, Dc, Iy, Ah, Hc]; Construct a training model that includes the first-layer model and the second-layer model. Use the stacking method to take the outputs of multiple base models as new features and input them into a meta-model for final classification. The first-layer meta-model is expressed as: P RF , P SVM and P XGB Represent the prediction results of random forest, support vector machine and XGBoost models respectively, and F RF 、F SVM and F XGB They represent random forest, support vector machine, and XGBoost models respectively; The second-level metamodel is represented as: Vm is the meta-feature vector, Fmeta is the meta-model, and LR is the binary classification function; Using the prediction results of the base model P RF ,P SVM ,P XGB As the new meta-feature vector Vm, train the meta-model FMeta, and use the meta-model to perform the final classification Ts on the prediction results of the basic model: 。 2. The method according to claim 1, characterized in that The user's interactive behavior data is collected through multiple social media platform API interfaces, and the interactive behavior data is cleaned and standardized, including: Collect interactive behavior data of social media users, including users’ forwarding, likes, comments and sharing data.

3. The method according to claim 2, characterized in that The method of collecting user interaction behavior data through multiple social media platform API interfaces and performing data cleaning and standardization on the interaction behavior also includes: Clean the collected data to remove duplicate, invalid and noisy data; Standardize the data, including unifying the time format, encoding the behavior type, and normalizing user identifiers.

4. The method according to claim 3, characterized in that The information propagation network is constructed based on the user interaction behavior data, and the propagation network includes nodes and contact surfaces. The weight of each contact surface in the network is calculated by the contact surface weight calculation function, including: Each user is represented as a node in the communication network. The attributes of the node include user ID, the type of behavior involved in the communication, and the interaction time. The interactive behaviors between users are set as the contact surfaces of the propagation paths. Each contact surface connects two users, indicating that there is an information propagation relationship between the two users. The weight W(i,j) represents the propagation weight between user i and user j. The weight W(i,j) of the contact surface is calculated by the following formula: f 转发 (i,j) represents the frequency of user i forwarding the content of user j, and f 评论 (i,j) represents the frequency of user i’s comments on user j’s content, f 点赞 (i,j) represents the frequency of user i’s likes on user j’s content, and α, β, and γ are weight parameters.

5. The method according to claim 1, characterized in that When the newly generated user interaction behavior data exceeds the preset value θ, the propagation feature vector is updated to Vsd, and the updated information propagation path Tsd is calculated based on the updated propagation feature vector Vsd, including: When the amount of newly generated user interaction behavior data exceeds a preset value θ, the newly generated user interaction behavior data is collected and preprocessed; Based on the updated data, the nodes and contact surfaces in the propagation network are reconstructed, and new contact surface weights are calculated; According to the updated propagation network, the propagation feature vector is recalculated to obtain an updated propagation feature vector Vsd, and the updated propagation feature vector Vsd is input into the trained model to perform feature standardization and feature generation to generate a new feature vector; The new feature vector is predicted using the basic model and meta-model to obtain the updated information propagation path Tsd.

6. The method according to claim 5, characterized in that The method of analyzing the propagation speed and key nodes of each social media platform through the cross-platform propagation path feature vector G and evaluating the contribution value of different social platforms in the information diffusion process includes: Analyze the information dissemination trajectory of the same user on multiple different social media platforms to form a cross-platform dissemination path feature vector G; By analyzing the timestamp and propagation path, the time required for information to be transmitted from one node to another is determined, thereby evaluating the propagation speed Pts of each platform; Calculate the activity value Phy of each node on different platforms during the propagation process; Based on the propagation speed Pts and activity value Phy, the contribution value of different social platforms in the process of information diffusion is evaluated.

7. A social communication path identification device based on interactive behavior, characterized in that: include: A collection module collects user interaction behavior data through multiple social media platform API interfaces, and performs data cleaning and standardization on the interaction behavior. The interaction behavior data includes user ID, content ID, interaction type and timestamp; A construction module, which constructs an information propagation network based on user interaction behavior data, wherein the propagation network includes nodes and contact surfaces, and the weight of each contact surface in the network is calculated by a contact surface weight calculation function; The identification module constructs a propagation feature vector Vs=[Lc, Ss, Dc, Iy, Ah, Hc] based on the contact surface and the weight of the contact surface, and identifies and classifies the information propagation path Ts based on the propagation feature vector and the path calculation function Fl, where Lc represents the propagation path length, Ss represents the propagation speed, Dc represents the propagation level, Iy represents the node social influence, Ah represents the node activity, and Hc represents the frequency of interactive behavior; An updating module, when the newly generated user interaction behavior data exceeds a preset value θ, updates the propagation feature vector to Vsd, and calculates an updated information propagation path Tsd based on the updated propagation feature vector Vsd; The evaluation module analyzes the information dissemination trajectory of the same user on multiple different social media platforms, forms a cross-platform propagation path feature vector G=[Ts,Tsd], analyzes the propagation speed and key nodes of each social media platform through the cross-platform propagation path feature vector G, and evaluates the contribution value of different social platforms in the information diffusion process; The method of constructing a propagation feature vector Vs=[Lc, Ss, Dc, Iy, Ah, Hc] based on the contact surface and the weight of the contact surface, and identifying and classifying the information propagation path Ts based on the propagation feature vector and the path calculation function Fl includes: Calculate the shortest path length from the information source node to the target node and get the propagation path length Lc: P(s,t) is the set of all paths from the source node s to the target node t, p is the propagation path, (u,v) is a contact surface on the path p, (u,v) is the weight of the contact surface (u,v), λ is a tuning parameter used to balance the influence of the contact surface weight and the node degree, and degree(u) is the degree of node u, indicating the number of contact surfaces connected to node u; Based on the time interval for information to propagate from the source node to the target node, the propagation speed Ss is calculated: p is the shortest path from source node s to target node t, timestamp(v) is the timestamp of node v, and μ is a tuning parameter used to control the effect of contact surface weight on propagation speed; Calculate the number of layers that the information passes through from the source node to the target node, and get the propagation level Dc: p is the shortest path from source node s to target node t, level(v) is the level number of node v, and η is a tuning parameter used to balance the influence of level difference and node degree; Calculate the node social influence Iy: w1, w2, w3 are weight parameters, flr(i) is the number of users followed by node i, itf(i) is the interaction frequency of node i, hsd(i) is the historical communication record of node i, δ is the adjustment parameter, and ctt(i) is the centrality of node i. Compute node activity Ah: w4, w5, w6 are weight parameters, itf(i) is the interaction frequency of node i, itd(i) is the interaction depth of node i, and atd(i) is the active duration of node i; Based on the frequency of users performing interactive behaviors on the propagation path, the interactive behavior frequency Hc is calculated: f 转发 (i,j) is the frequency of user i forwarding the content of user j, f 评论 (i,j) is the frequency of user i’s comments on user j’s content, f 点赞 (i, j) is the frequency of user i's likes on user j's content, ρ is the adjustment parameter, and sts(i, j) is the sentiment score of user i on user j's content; Combine the calculated features into a feature vector Vs=[Lc, Ss, Dc, Iy, Ah, Hc]; Construct a training model that includes the first-layer model and the second-layer model. Use the stacking method to take the outputs of multiple base models as new features and input them into a meta-model for final classification. The first-layer meta-model is expressed as: P RF , P SVM and P XGB Represent the prediction results of random forest, support vector machine and XGBoost models respectively, and F RF 、F SVM and F XGB They represent random forest, support vector machine, and XGBoost models respectively; The second-level metamodel is represented as: Vm is the meta-feature vector, Fmeta is the meta-model, and LR is the binary classification function; Using the prediction results of the base model P RF ,P SVM ,P XGB As the new meta-feature vector Vm, train the meta-model FMeta, and use the meta-model to perform the final classification Ts on the prediction results of the basic model: 。 8. An electronic device, characterized in that: The electronic device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the social communication path identification method based on interactive behavior as described in any one of claims 1 to 6.

Images