Cross-social-platform message propagation visual analysis method

Abstract

The invention discloses a message propagation visual analysis method across social platforms. The method comprises the following steps: summarizing cross-platform propagation modes and back mechanisms thereof, and judging whether each cross-platform propagation mode has related attributes or behaviors; determining important factors influencing the cross-platform propagation probability, constructing a cross-platform message propagation model, calculating the cross-platform propagation probability of the topic level and the cross-platform propagation probability of the post level according to the important factors, and judging the type of a cross-platform propagation mode; and designing a cross-platform propagation visual analysis system comprising a topic level visual analysis interface and a post level visual analysis interface to display the cross-platform propagation relationship and realize user interaction analysis. According to the method, the problems of complex mechanism, difficulty in reasoning and high expandability requirement of existing cross-social-platform propagation analysis can be solved, interactive analysis of cross-social-platform message propagation conditions is realized through a visual interface, and meaningful data insight is found.

Description

Technical Field

[0001] The present invention belongs to the technical field of social media communication analysis, and in particular relates to a method for visually analyzing message propagation across social platforms. Background Art

[0002] Social media has been integrated into people's daily lives and has become an important channel for people to share life moments, express personal emotions, and discuss public events. The diversity of social media platforms has greatly facilitated the easy spread of information between these platforms, which has led to a hot research issue. In-depth analysis and understanding of the cross-platform propagation of information can help analysts gain insights into complex user behaviors, design effective marketing strategies, and prevent the spread of harmful content. Although many studies have been devoted to the visualization of social media information propagation, few studies have fully considered the cross-platform aspect due to the inherent complexity and hidden nature of the cross-platform propagation mechanism.

[0003] Existing information diffusion visualization methods mainly use metaphor-based designs to display the diffusion process of spatiotemporal information on social media platforms (e.g., Weibo or Twitter). However, due to the lack of in-depth understanding of the cross-platform propagation mechanism, these visualization methods often fail to reveal the underlying data patterns. In addition, although a few studies have attempted to integrate information from multiple social media platforms, such as the technical solution disclosed in the literature (S. Liu, X. Wang, J. Chen, J. Zhu, and B. Guo. Topic panorama: A full picture of relevant topics. IEEE Transactions on Visualization and Computer Graphics, 22(12): 2508–2521, 2016.), they only studied how to merge topics extracted from different platforms into a unified graph for presentation. Some other studies, such as the literature (Y. Lu, H. Wang, S. Landis, and R. Maciejewski. A visual analytics framework for identifying topic drivers in media events. IEEE Transactions on Visualization and Computer Graphics, 24(9): 2501–2515, 2018.), only use information from multiple platforms to obtain comprehensive insights related to their research topics. Using cross-platform propagation mechanisms to intuitively visualize and reveal the propagation patterns between social media platforms is still an urgent problem to be solved.

[0004] There are three major challenges in achieving visual analysis of message propagation across social platforms:

[0005] (1) The cross-platform communication mechanism is complex. The driving factors behind cross-platform communication may vary greatly in different situations, so there are diverse and complex communication patterns. Identifying and analyzing these patterns requires a lot of effort and extensive expertise, which is a challenging and labor-intensive task.

[0006] (2) Implicit communication is difficult to infer. Information dissemination within the platform is mostly visible and visible, mainly through functions such as forwarding, commenting, and liking. However, unlike the dissemination within the platform, cross-platform dissemination is mostly not convenient to obtain directly, and the inference of the probability of implicit dissemination is relatively complex and difficult.

[0007] (3) Multifaceted information visualization requires high scalability. Given the large number of posts and the complex dynamics of information dissemination, a scalable visualization design is needed to effectively display the cross-platform dissemination of information. In addition, given the complexity of social media data, the visualization design must be comprehensive enough to support the display and analysis of multifaceted information. Summary of the invention

[0008] In view of the above, the purpose of the present invention is to provide a visual analysis method for message propagation across social platforms to solve the problems faced by existing cross-social platform propagation analysis, such as complex mechanisms, difficult reasoning, and high scalability requirements, so that experts who study cross-platform propagation of social media can interactively analyze the message propagation across social platforms through a visual interface and discover meaningful data insights.

[0009] In order to achieve the above-mentioned invention object, the technical solution provided by the present invention is as follows:

[0010] An embodiment of the present invention provides a method for visually analyzing message propagation across social platforms, comprising the following steps:

[0011] Based on the cross-platform communication examples between different social media platforms, we summarize various cross-platform communication patterns, the mechanisms behind each cross-platform communication pattern, and the relevant attributes or behaviors used to determine whether each cross-platform communication pattern exists;

[0012] Based on the cross-platform communication mode and its underlying mechanism and related attributes or behaviors, the important factors that affect the cross-platform communication probability are determined, and a cross-platform message communication model is constructed to calculate the cross-platform communication probability at the topic level according to the important factors, calculate the cross-platform communication probability at the post level according to the important factors, and determine the type of cross-platform communication mode between posts;

[0013] A cross-platform communication visual analysis system is designed, which includes a topic-level visual analysis interface and a post-level visual analysis interface. In the topic-level visual analysis interface, the cross-platform communication probability at the topic level is visualized as the width of the message propagation path between topics; in the post-level visual analysis interface, the cross-platform communication probability at the post level is visualized as the width of the message propagation path between posts and the type of cross-platform communication pattern between posts is displayed. The cross-platform communication visual analysis system is used to display the cross-platform communication relationship and realize user interaction analysis.

[0014] Preferably, the cross-platform communication mode includes cross-platform communication between accounts held by the same entity, cross-platform communication driven by account positioning, and cross-platform communication driven by content. Among them, the mechanism behind the cross-platform communication between accounts held by the same entity includes the needs of operating multiple accounts and testing needs, the mechanism behind the cross-platform communication driven by account positioning includes transfer accounts or news accounts or media accounts, fan accounts, and functional accounts, and the mechanism behind the content-driven cross-platform communication includes the desire to share or express stimulated by the source content, the promotion task for the source content, and content expansion.

[0015] Preferably, important factors affecting the probability of cross-platform dissemination include information salience, influence of key opinion leaders, recency effect, content relevance, user relevance and historical information relevance.

[0016] Preferably, in the cross-platform message propagation model, the cross-platform propagation probability of the topic level is calculated based on important factors including information significance, influence of key opinion leaders and content relevance, and the calculation formula is as follows:

[0017]

[0018] in, It represents the probability of topic h spreading from platform i to platform j at time t, S represents the information significance, K represents the influence of key opinion leaders, and E represents the content relevance.

[0019] Preferably, in the cross-platform message propagation model, the cross-platform propagation probability at the post level is calculated based on important factors including recency effect, content relevance, user relevance, and historical information relevance, and the calculation formula is as follows:

[0020]

[0021] Among them, P T→W represents the cross-platform propagation probability from post T to post W, max(·) represents the maximum value function, Direct(T postURL ,W text ) means checking whether the text of W contains the URL of T. postURL represents the URL of T, Wtext represents the text of W, CR represents the content correlation between T and W, UR represents the user correlation between the account that publishes T and the account that publishes W, HR represents the historical information correlation between the account that publishes T and the account that publishes W, t(W) represents the publishing time of W, t(T) represents the publishing time of T, α represents the weight, and t(W)-t(T) quantifies the recency effect.

[0022] Preferably, in the cross-platform message propagation model, when determining the type of cross-platform propagation mode between posts, if the user information U of post T userinfo (T) and user information U of post W userinfo (W) similarity value Sim(U userinfo (T),U userinfo (W)) is greater than or equal to the first similarity threshold, it is a cross-platform communication between accounts held by the same entity; if the similarity value Sim(U userinfo (T),U userinfo (W)) is less than the first similarity threshold and greater than or equal to the second similarity threshold, or the user information is not an ordinary account, or words related to the source platform appear in the user information, it is cross-platform communication driven by account positioning; if it does not meet the judgment conditions of cross-platform communication between accounts held by the same entity and cross-platform communication driven by account positioning, it is content-driven cross-platform communication.

[0023] Preferably, the topic-level visual analysis interface in the cross-platform communication visual analysis system includes a topic-level communication view, a timeline view and a topic-level detail view; wherein, in the topic-level communication view, the cross-platform communication relationship of topics is displayed in a metaphor-based visualization method and the cross-platform communication probability of the topic level is visualized as the message propagation path between topics; in the timeline view, the number of event-related posts on each platform is displayed over time, and time range selection and animation control are supported; in the topic-level detail view, detailed information of posts and posting accounts are listed according to different topic clusters.

[0024] Preferably, in the topic-level propagation view, a flower-shaped visualization image character adapted from the Nightingale Rose diagram is used to represent the overall situation of the collection of posts on a specific topic posted on the platform within a specific time range; the color of the petals, i.e., the sectors, of each flower-shaped visualization image character corresponds to the color of the platform, the number of petals is equal to the number of days on which posts related to the topic have been posted within the selected time range, and a vertical line is used to separate the petals representing the earliest date and the petals representing the latest date to indicate the start and end dates, so that the user can understand the changes in the posting situation over time in a clockwise direction, the length of each petal encodes the daily posting volume of the current topic on the current platform, and influential posts are highlighted by overlaying dark colors on each petal. Number of posts; the radius of the inner circle of the flower-shaped visual graphic character encodes the total number of posts in this collection, and the angles of the two semicircles around it encode the total number of influential posts in the collection and the total number of influential posting accounts. A point in the pistil represents a post, and the color of each point reflects the sentiment of the post; clicking a petal will display the representative post of the current date, and clicking the pistil will display the representative word cloud of the current post collection; a hexagonal grid is used to place the flower-shaped visual graphic character, and each flower-shaped visual graphic character is placed in the center of a hexagon. Flower-shaped visual graphic characters from different platforms are arranged in different directions, and flower-shaped visual graphic characters on the same platform are organized together in a honeycomb cluster style;

[0025] At the same time, an algorithm based on a hexagonal grid is used to arrange the message propagation path from the flower-shaped visual image characters of the source topic to the flower-shaped visual image characters of the target topic, and the width of the message propagation path is used to encode the size of the cross-platform propagation probability at the topic level, and arrows are added to the message propagation path to indicate the propagation direction. The color of the message propagation path is encoded using the color of the source platform. It supports hovering the mouse over the message propagation path to illuminate the current path. After left-clicking the path, three seeds will appear on the path. Each seed represents a class of significant cross-platform cases, including the source post and the target post containing the same URL in the content, the target post containing the URL of the source post, and the target post containing certain specific references to the source post. It supports users to right-click on any path to jump to the post-level visual analysis interface, and further explore the cross-platform message propagation relationship between the source post set and the target post set connected by the path at the post level.

[0026] Preferably, the post-level visual analysis interface in the cross-platform communication visual analysis system includes a post-level communication view and a post-level detail view; wherein, in the post-level communication view, the communication relationship from post to post between different collections on the same topic on different platforms is displayed, and the cross-platform communication probability at the post level is visualized as the width of the message propagation path between posts, and the type of cross-platform communication pattern between posts is displayed based on the message propagation path between posts clicked by users; in the post-level detail view, the historical information of the posting account is displayed through the historical information relevance module, and the historical information includes a historical post list and interest distribution.

[0027] Preferably, in the post-level propagation view, each collection is visualized as a circle, and the posts in the circle are laid out using a bubble chart. The coding of the posts is based on their influence indicators. If the sum of the number of reposts, comments and likes of the post is greater than or equal to the first set threshold or the number of followers of the posting account is greater than or equal to the second set threshold, the post is visualized as a flower, and the size of the flower is proportional to the influence indicator. The color of the inner circle of the flower represents the emotion of the post. Otherwise, the post is visualized as a point, and its color represents the emotion of the post. Users are supported to adjust the first set threshold and the second set threshold through the control bar; between collections, the cross-platform message propagation path extends from one post to another, and the width of the message propagation path is encoded by the cross-platform probability at the post level. The interface supports adjusting the number of displayed paths according to user needs, and uses edge bundling technology for path layout. When the user adjusts the first set threshold and / or the second set threshold, unimportant paths will be filtered out. When the control bar of the first set threshold and the second set threshold is dragged to the far right, that is, the maximum value, only the most important paths are retained on the interface; clicking on a flower or a dot representing a post will display the detailed content of the post.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] The visual analysis method of message propagation across social platforms proposed in this invention, based on the systematic summary of cross-platform propagation patterns and the mechanisms behind them, determines the important factors that affect the probability of propagation, and proposes a cross-platform message propagation model based on the principle of infectious disease model. Based on the cross-platform message propagation model, the important factors are used to calculate the probability of propagation and judge the type of cross-platform propagation patterns, thereby solving the problem of complex cross-platform propagation mechanisms and difficult reasoning. At the same time, through the hierarchical design of the visual interface from the overall (topic level) to the details (post level) and the novel visual metaphor, the various aspects of cross-platform propagation information are clearly and intuitively presented, effectively revealing the laws and patterns of cross-platform propagation, and can help experts understand, explore and analyze the cross-platform propagation of events. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art 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 creative work.

[0031] Figure 1 It is a flowchart of a method for visually analyzing message propagation across social platforms provided by an embodiment of the present invention;

[0032] Figure 2 Schematic diagram of the architecture of a visual analysis method for message propagation across social platforms provided by an embodiment of the present invention;

[0033] Figure 3 is a schematic diagram of a topic level visual analysis interface provided by an embodiment of the present invention;

[0034] Figure 4 It is a schematic diagram of a post level visual analysis interface provided by an embodiment of the present invention;

[0035] Figure 5 It is a schematic diagram of a flower-shaped visualized image character design provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0036] To make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific implementation methods described herein are only used to explain the present invention and do not limit the scope of protection of the present invention.

[0037] The inventive concept of the present invention is: in response to the problems of complex mechanism, difficult reasoning and high scalability requirements faced by cross-social platform communication analysis in the prior art, an embodiment of the present invention provides a visual analysis method for message propagation across social platforms. First, various modes of cross-platform communication are systematically summarized to understand the complex mechanisms behind cross-platform communication; then, important factors affecting cross-platform communication are determined and a cross-platform message propagation model is constructed on this basis to calculate the possibility of cross-platform communication of topics or posts to achieve reasoning about cross-platform communication; finally, a multi-view cross-platform communication visual analysis system is developed, which uses a novel visual metaphor to compare the cross-platform communication of information to the communication between multiple gardens in an ecosystem, so as to assist experts in exploring and analyzing the cross-platform communication of an event through multi-faceted information visualization with good scalability.

[0038] Figure 1 is a flow chart of a method for visually analyzing message propagation across social platforms provided by an embodiment of the present invention, Figure 2Schematic diagram of the architecture of the visual analysis method for message propagation across social platforms provided by an embodiment of the present invention. Figure 1 and Figure 2 As shown, the embodiment provides a method for visually analyzing message propagation across social platforms, including the following steps:

[0039] S1, based on the cross-platform communication examples between different social media platforms, summarizes a variety of cross-platform communication patterns, the mechanisms behind each cross-platform communication pattern, and the relevant attributes or behaviors used to determine whether each cross-platform communication pattern exists.

[0040] In the embodiment, a large number of cross-platform communication examples were collected from mainstream social media platforms at home and abroad, such as Weibo, Douban, Twitter, Facebook, and Reddit. Based on these examples, their characteristics, potential motivations for their occurrence, and judgment criteria were discussed. Finally, the conclusions shown in Table 1 were obtained about the cross-platform communication mechanism and the mechanism behind it. There are three types of cross-platform communication modes: cross-platform communication between accounts held by the same entity, cross-platform communication driven by account positioning, and cross-platform communication driven by content.

[0041] Table 1 Cross-platform communication patterns and their underlying mechanisms

[0042]

[0043] The first type of cross-platform communication model is cross-platform communication between accounts held by the same entity, which is further divided into two points according to the different mechanisms behind it: the first point is the need to operate multiple accounts at the same time. In order to take care of the audiences on different platforms, the same entity sometimes adopts different expression strategies on different platforms; the second point is the need for testing. Sometimes the communication effect is tested on one platform before being released to the real target platform. In view of these characteristics, when judging whether such a model of cross-platform communication between accounts held by the same entity exists, the relevance of user information and the influence of the account should be focused on.

[0044] The second type of cross-platform communication model is account positioning-driven communication, which is further divided into three points according to the different mechanisms behind it: transfer account / news account / media account, fan account, and functional account. When judging whether such a communication model driven by account positioning exists, the account positioning should be locked from the account name, personal profile and authentication information. At the same time, attention should be paid to whether the relevant information mentions the information source platform or user. In addition, generally, the transfer account and fan account reflect that the account's focus content has an audience on the platform, which can be seen from the number of fans and the data of post forwarding, commenting and liking.

[0045] The third type of cross-platform communication is content-driven cross-platform communication, which can be further divided into three points according to the different mechanisms behind it: the first point is that the source content arouses the desire to share or express, and generally some personal opinions and attitudes are mixed in when reposting the message. In particular, when holding an attitude opposite to the source content, people tend to express themselves on platforms that can get more support. However, when holding the same attitude, people will also quote it to support their own views; the second point is to promote the source content; the third point is content expansion, which stems from the differences in posting styles and review mechanisms on different platforms. For example, tweets are generally short and concise, while forums such as Reddit often share more detailed content. Therefore, it is not uncommon to simply introduce in tweets and jump to links for details. When judging whether such a model of content-driven cross-platform communication exists, attention should be paid to specific URLs or indicative words in the post content. In addition, attention should be paid to the publisher's activity on their platform and their interest in the original content, which can be inferred from their historical posting information, which may also include cross-platform content sharing. In addition, it should be taken into account that original content that can catalyze cross-platform dissemination usually already has a certain audience base on the target platform, which means that on the target platform, the relevant topics not only have a certain degree of discussion heat, but also show relatively clear attitude tendencies.

[0046] S2, based on the cross-platform propagation pattern and its underlying mechanisms and related attributes or behaviors, determine the important factors that affect the probability of cross-platform propagation, and build a cross-platform message propagation model to calculate the cross-platform propagation probability at the topic level according to important factors, calculate the cross-platform propagation probability at the post level according to important factors, and determine the type of cross-platform propagation pattern between posts.

[0047] In the embodiment, according to the systematic summary in step S1, the important factors for estimating the probability of cross-platform communication are determined: information salience, key opinion leader influence, recency effect, content relevance, user relevance and historical information relevance. Information salience means that the ability of information to spread is proportional to the total amount of related posts, and a hot topic is more likely to spread across platforms. Key opinion leader influence means that influential accounts or posts are more likely to be spread across platforms. Recency effect means that new information has a greater impact on the audience than old information. Content relevance means that when there is a certain degree of relevance between the content of the posts, the possibility of cross-platform communication between them will increase. User relevance means that whether it is the communication between accounts held by the same entity or the communication driven by account positioning, there is a strong correlation between the source account (the account that initially posted the content) and the target account (the account that spread the content on the new platform). Historical information relevance means that historical behavior information can reflect account habits. When two accounts have similar interests, continue to discuss certain topics, or have established a cross-platform communication relationship, the probability of continuous cross-platform interaction between them will increase.

[0048] Then, we use these important factors to build a cross-platform message propagation model. Specifically, the entire cross-platform message propagation model is divided into two parts. The first part of the model is used to calculate the cross-platform propagation probability at the topic level, and the second part of the model is used to calculate the cross-platform propagation probability at the post level.

[0049] Part I Topic Level Model: Considering the content relevance in communication, cross-platform communication should occur between clusters of the same topic on different platforms. In order to calculate the possibility of communication between clusters of the same topic on different platforms, the cross-platform communication probability at the topic level is calculated based on important factors including information significance, influence of key opinion leaders and content relevance. The calculation formula is as follows:

[0050]

[0051] in, represents the probability that topic h spreads from platform i to platform j at time t. This formula is inspired by the principle of the infectious disease model. S corresponds to the basic transmission rate, which indicates the significance of information and is reflected by the number of posts; K corresponds to the source strength, which indicates the influence of key opinion leaders and is reflected by the participation index and the number of fans; E corresponds to the contact rate, which indicates the content relevance and is reflected by the content similarity and the content relevance that implies the spread. These important factors are calculated based on various variables (as shown in Table 2). The variables in Table 2 are arranged and combined according to their positive and negative correlation with the factors. The weights of the variables are determined based on empirical observations of small-scale data. The meaning of the symbols of the variables in Table 2 is explained in Table 3. Considering the recency effect and the time cost of transmission, the posts published in period t are spread by the posts published during [t-Δt, t], which is abbreviated as t-1. The specific calculation of each variable in the topic level model in the first part is shown in Table 4.

[0052] Table 2 Calculation method of important factors

[0053]

[0054] Table 3 Meaning of symbols

[0055]

[0056]

[0057] Table 4 Variable calculation method in topic level model

[0058]

[0059]

[0060] Part II Post-level model: The cross-platform propagation probability at the topic level provides an overview of the cross-platform diffusion of a topic. Subsequently, within the same topic, the propagation probability between two posts on different platforms is further calculated. In order to estimate the possibility of cross-platform propagation from post T to another post W within the same topic, the cross-platform propagation probability at the post level is calculated based on important factors including recency effect, content relevance, user relevance, and historical information relevance. The calculation formula is as follows:

[0061]

[0062] Among them, P T→W represents the cross-platform propagation probability from post T to post W, and max(·) represents the maximum value function. Direct(T postURL ,W text) means checking whether the text of W contains the URL of T, which is regarded as a direct sign of spread. The combination of other important factors is inspired by the principles in the infectious disease model. CR represents the content correlation between T and W, UR represents the user correlation between the account that publishes T and the account that publishes W, HR represents the historical information correlation between the account that publishes T and the account that publishes W, t(W) represents the publishing time of W, t(T) represents the publishing time of T, α represents the weight (α is 0.2 in the embodiment), and t(W)-t(T) represents the quantification of the recency effect. These important factors are calculated based on various variables (as shown in Table 2), which are arranged and combined according to their positive and negative correlation with the factors. The weights of the variables are determined based on empirical observations of small-scale data. The specific calculations of each variable in the post-level model in Part II are shown in Table 5.

[0063] In addition, the post-level model in the second part can also determine the type of cross-platform communication mode through the following logic: If the user information U of post T userinfo (T) and user information U of post W userinfo (W) similarity value Sim(U userinfo (T),U userinfo (W)) is greater than or equal to the first similarity threshold (in the embodiment, the first similarity threshold is 0.5), it is a cross-platform communication between accounts held by the same entity; if the similarity value Sim(U userinfo (T),U userinfo (W)) is less than the first similarity threshold and greater than or equal to the second similarity threshold (the second similarity threshold in the embodiment is 0.3), or the user information is not an ordinary account, or words related to the source platform appear in the user information, then it is cross-platform communication driven by account positioning; if both the judgment conditions of cross-platform communication between accounts held by the same entity and cross-platform communication driven by account positioning are not met, then it is content-driven cross-platform communication.

[0064] Table 5 Calculation method of variables in post-level model

[0065]

[0066]

[0067] S3. Design a cross-platform communication visual analysis system including a topic-level visual analysis interface and a post-level visual analysis interface. In the topic-level visual analysis interface, the cross-platform communication probability at the topic level is visualized as the width of the message propagation path between topics. In the post-level visual analysis interface, the cross-platform communication probability at the post level is visualized as the width of the message propagation path between posts and the type of cross-platform communication pattern between posts is displayed. The cross-platform communication visual analysis system is used to display the cross-platform communication relationship and realize user interaction analysis.

[0068] In the embodiment, a visualization technology based on visual metaphor is used to design a cross-platform communication visual analysis system including a topic-level visual analysis interface and a post-level visual analysis interface. The cross-platform communication of information is likened to communication between gardens, assisting experts in conducting communication analysis.

[0069] Specifically, Figure 3 As shown in the topic-level visual analysis interface, in order to facilitate the analysis of the cross-platform dissemination of an event, the entire interface includes a topic-level dissemination view ( Figure 3 a), Timeline view ( Figure 3 b) and the topic level detail view ( Figure 3 c in the figure). The user first selects the events and platforms to be analyzed, and then the timeline view ( Figure 3 b) in the figure shows how the number of event-related posts on each platform changes over time. This view also supports the selection of a time range ( Figure 3 b1) and animation control ( Figure 3 After selecting a time range, the topic level detail view ( Figure 3 c) and the topic-level propagation view ( Figure 3 a) in the figure shows the topic clustering results and the cross-platform communication capabilities of the topic level. Figure 3 c) Lists posts and posting accounts by topic cluster. Topic-level propagation view ( Figure 3 a) in the figure shows the cross-platform spread of topics in a metaphor-based visualization. After understanding the overall spread of each topic across different platforms, users can right-click on the message propagation path between a topic pair to jump to the post-level visual analysis interface ( Figure 4 ) to explore the cross-platform spread of the topic pairs connected by the path at the post level. Once switched to the post level, the post level detail view ( Figure 4 b) in will adjust the content to match the post level spread view ( Figure 4In addition, the entire cross-platform communication visual analysis system interface uses a unified color coding, and information about the same platform is coded in the same color system: in the embodiment, platform 1 (such as Twitter) is represented by blue, platform 2 (such as Weibo) is represented by orange, and platform 3 (such as Facebook) is represented by green.

[0070] In the topic-level diffusion view, social platforms, posts, cross-platform diffusion paths, and typical diffusion instances are visualized as elements in a virtual ecosystem, including gardens, flowers, wind, and seeds. Specifically, in the topic-level analysis, a flower-shaped visualization glyph represents a collection of posts about a specific topic published on the platform within a specific time frame. The topic-level flower-shaped glyph is designed to summarize the overall situation of the collection (e.g. Figure 3 ), and encode the factors that affect its propagation potential in a visual way. Specifically, Figure 5 As shown in the figure, the topic-level flower-shaped glyphs are adapted from the Nightingale Rose diagram, and the outer radius of these glyphs is uniform. The color of the petals (i.e., sectors) of the flower-shaped glyph corresponds to the color of the platform, and the number of petals is equal to the number of days related to the topic was published in the selected time range. To indicate the start and end dates, a vertical black line is used between the petals representing the earliest date and the petals representing the latest date (as shown in Figure 1). Figure 3 The length of each petal encodes the number of posts on the current topic on the current platform on that day, and is represented by a dark layer on each petal (as shown in a4 in the figure). Figure 5 The radius of the pistil (i.e., the inner circle) of the flower-shaped glyph encodes the total number of posts in this set, and the angles of the two semicircles around it (i.e., Figure 5 The two gray areas in the inner circle (shown in the figure) encode the total number of influential posts and the total number of influential posting accounts in the collection. Each dot in the pistil represents a post. The color of each dot reflects the sentiment of the post, and the corresponding post's sentiment is displayed in the upper left corner of the topic-level propagation view. Clicking a petal will display the representative posts of the current date (such as Figure 3 Clicking the flower core will display the representative word cloud of the current post collection. Use a hexagonal grid to place flower glyphs, with each flower glyph placed at the center of a hexagon. To enhance the proximity of flower glyphs within the same platform, flower visual glyphs from different platforms are arranged in different directions. Flower glyphs on the same platform are organized together in a honeycomb cluster style.

[0071] At the same time, in the topic-level propagation view, the cross-platform propagation path (such as Figure 3 A6 in the figure is used to show the calculation results of the cross-platform propagation probability at the topic level. A hexagonal grid-based algorithm is used to arrange the message propagation path from the flower-shaped glyph of the source topic to the flower-shaped glyph of the target topic. In order to determine the path between the flower-shaped glyphs under the same topic, an improved breadth-first search (BFS) algorithm is used to generate propagation edges along the points in the hexagonal grid. Starting from the source hexagon, the vertex closest to the target hexagon is first found. Considering the aesthetics and neatness, the subsequent path finding process is as follows: The algorithm aims to pursue the shortest path while maximizing the degree of shared line segments between paths. Therefore, for each calculated path, the algorithm will increase the incentive value of each hexagonal grid vertex it passes through, ensuring that vertices with higher incentive values ​​are given priority in subsequent path calculations. The width of the message propagation path is used to encode the size of the cross-platform propagation probability at the topic level (such as Figure 3 Because the propagation probability is directional, the interface displays the direction with a higher propagation probability. The color of the message propagation path is coded using the color of the source platform. Click the animation control button ( Figure 3 b2 in the topic level propagation view) can display the dynamic process of all paths in turn in the topic level propagation view. You can hover the mouse over the message propagation path to illuminate the current path and enhance the visibility of the path. By left-clicking the path, the significant propagation instance used to calculate the propagation probability will be revealed. After left-clicking the path, three seeds will appear on the path (such as Figure 3 Each seed represents a significant cross-platform case, including the source post and the target post containing the same URL, the target post containing the URL of the source post, and the target post containing certain specific references to the source post. After checking the spread of the topic level, the user can right-click any path to jump to the post level visual analysis interface to further explore the message propagation relationship between the source post set and the target post set connected by the path at the post level.

[0072] Specifically, Figure 4 As shown in FIG. 1 , in the post-level visual analysis interface, the focus is on the spread from post to post between two collections on the same topic on two different platforms. The post-level visual analysis interface includes a post-level spread view and a post-level detail view. In the post-level spread view (such as Figure 4 ), showing the propagation relationship from post to post between different collections of the same topic on different platforms. Each collection is visualized as a circle, and the posts in the circle are laid out using a bubble chart (as shown in Figure 4 The coding of posts is based on their influence indicators (such as Figure 4If the sum of the number of reposts, comments and likes of a post is greater than or equal to the first threshold (adjusted by the user) or the number of followers of the posting account is greater than or equal to the second threshold (adjusted by the user), the post is visualized as a flower, the size of the flower is proportional to the influence index, and the color of the inner circle of the flower represents the emotion of the post. Otherwise, the post is visualized as a dot, and its color represents the emotion of the post. The user can adjust the first and second thresholds through the control bar (as shown in Figure 1). Figure 4 ), and an example of a numerical color icon showing the sentiment of the corresponding post from negative to positive is displayed in the upper left corner of the post-level propagation view. Between sets, the cross-platform message propagation path extends from one post to another. The width of the message propagation path is encoded by the cross-platform probability at the post level. The interface supports adjusting the number of displayed paths according to user needs, and uses edge bundling technology for path layout. When the user adjusts the first set threshold and / or the second set threshold, unimportant paths are filtered out. When the control bar of the first set threshold and the second set threshold is dragged to the far right, that is, the maximum value, only the most important paths are retained on the interface. Clicking a flower or a dot representing a post will display the detailed content of the post (such as Figure 4 Click the animation control button ( Figure 3 b2) will also display the dynamic process of all paths in the post-level propagation view in sequence. When a user clicks on the message propagation path between posts in the post-level propagation view, the detailed information of the pair of posts will be displayed side by side (such as Figure 4 Specific spreading words and similar content such as tags and URLs will be highlighted, and the cross-platform spreading mode of this case identified by the cross-platform message propagation model will also be directly displayed. At the same time, in the post-level detail view (such as Figure 4 The historical information of these two posting accounts will be displayed in the "Historical Information Relevance" module of the detail view, including their historical post lists (such as Figure 4 ) and interest distribution (as shown in b1 in Figure 4 As shown in b2 in the figure, the interest distribution is the keywords calculated based on historical posts. Different circles in the interest distribution represent different keywords. The higher the frequency of the keyword, the larger the radius of the circle in the interest distribution.

[0073] The specific implementation methods described above provide a detailed description of the technical solutions and beneficial effects of the present invention. It should be understood that the above is only the most preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, supplements and equivalent substitutions made within the scope of the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A visual analysis method for message propagation across social platforms, characterized in that: The following steps are involved: Based on the cross-platform communication examples between different social media platforms, we summarize various cross-platform communication patterns, the mechanisms behind each cross-platform communication pattern, and the relevant attributes or behaviors used to determine whether each cross-platform communication pattern exists; Based on the cross-platform communication mode and its underlying mechanism and related attributes or behaviors, the important factors that affect the cross-platform communication probability are determined, and a cross-platform message communication model is constructed to calculate the cross-platform communication probability at the topic level according to the important factors, calculate the cross-platform communication probability at the post level according to the important factors, and determine the type of cross-platform communication mode between posts; A cross-platform communication visual analysis system is designed, which includes a topic-level visual analysis interface and a post-level visual analysis interface. In the topic-level visual analysis interface, the cross-platform communication probability at the topic level is visualized as the width of the message propagation path between topics; in the post-level visual analysis interface, the cross-platform communication probability at the post level is visualized as the width of the message propagation path between posts and the type of cross-platform communication pattern between posts is displayed. The cross-platform communication visual analysis system is used to display the cross-platform communication relationship and realize user interaction analysis.

2. The method for visual analysis of message propagation across social platforms according to claim 1, characterized in that: Cross-platform communication modes include cross-platform communication between accounts held by the same entity, cross-platform communication driven by account positioning, and cross-platform communication driven by content. Among them, the mechanism behind cross-platform communication between accounts held by the same entity includes the need to operate multiple accounts and the need to test; the mechanism behind cross-platform communication driven by account positioning includes transfer accounts, news accounts, media accounts, fan accounts, and functional accounts; the mechanism behind content-driven cross-platform communication includes the desire to share or express oneself stimulated by the source content, the promotion task for the source content, and content expansion.

3. The method for visual analysis of message propagation across social platforms according to claim 1, characterized in that: Important factors that affect the probability of cross-platform dissemination include information salience, influence of key opinion leaders, recency effect, content relevance, user relevance and historical information relevance.

4. The method for visual analysis of message propagation across social platforms according to claim 3, characterized in that: In the cross-platform message propagation model, the cross-platform propagation probability of the topic level is calculated based on important factors including information salience, influence of key opinion leaders, and content relevance. The calculation formula is as follows: in, It represents the probability of topic h spreading from platform i to platform j at time t, S represents the information significance, K represents the influence of key opinion leaders, and E represents the content relevance.

5. The method for visually analyzing message propagation across social platforms according to claim 3, characterized in that: In the cross-platform message propagation model, the cross-platform propagation probability at the post level is calculated based on important factors including recency effect, content relevance, user relevance, and historical information relevance. The calculation formula is as follows: Among them, P T→W represents the cross-platform propagation probability from post T to post W, max(·) represents the maximum value function, Direct(T postURL ,W text ) means checking whether the text of W contains the URL of T. postURL represents the URL of T, W text represents the text of W, CR represents the content correlation between T and W, UR represents the user correlation between the account that publishes T and the account that publishes W, HR represents the historical information correlation between the account that publishes T and the account that publishes W, t(W) represents the publishing time of W, t(T) represents the publishing time of T, α represents the weight, and t(W)-t(T) quantifies the recency effect.

6. The method for visually analyzing message propagation across social platforms according to claim 2, characterized in that: In the cross-platform message propagation model, when judging the type of cross-platform propagation mode between posts, if the user information U of post T userinfo (T) and user information U of post W userinfo (W) similarity value Sim(U userinfo (T),U userinfo (W)) is greater than or equal to the first similarity threshold, it is a cross-platform communication between accounts held by the same entity; if the similarity value Sim(U userinfo (T),U userinfo (W)) is less than the first similarity threshold and greater than or equal to the second similarity threshold, or the user information is not an ordinary account, or words related to the source platform appear in the user information, it is cross-platform communication driven by account positioning; if it does not meet the judgment conditions of cross-platform communication between accounts held by the same entity and cross-platform communication driven by account positioning, it is content-driven cross-platform communication.

7. The method for visually analyzing message propagation across social platforms according to claim 1, characterized in that: The topic-level visual analysis interface in the cross-platform communication visual analysis system includes a topic-level communication view, a timeline view, and a topic-level detail view. In the topic-level communication view, the cross-platform communication relationship of topics is displayed in a metaphor-based visualization method, and the cross-platform communication probability of the topic level is visualized as the message propagation path between topics. The timeline view shows how the number of event-related posts on each platform changes over time, and supports time range selection and animation control. In the topic-level detail view, detailed information on posts and posting accounts are listed according to different topic clusters.

8. The method for visually analyzing message propagation across social platforms according to claim 7, characterized in that: In the topic-level propagation view, a flower-shaped visualization graphic character adapted from the Nightingale Rose diagram is used to represent the overall situation of the collection of posts about a specific topic posted on the platform within a specific time range; the color of the petals, i.e., the sectors, of each flower-shaped visualization graphic character corresponds to the color of the platform, and the number of petals is equal to the number of days on which posts related to the topic were posted within the selected time range. A vertical line is used to separate the petals representing the earliest date and the latest date to indicate the start and end dates, allowing users to understand the changes in posting situation over time in a clockwise direction. The length of each petal encodes the daily posting volume of the current topic on the current platform, and influential posts are highlighted by overlaying dark colors on each petal. Quantity; the radius of the inner circle of the flower-shaped visual graphic character encodes the total number of posts in this collection, and the angles of the two semicircular rings around it encode the total number of influential posts and the total number of influential posting accounts in the collection. A point in the pistil represents a post, and the color of each point reflects the sentiment of the post; clicking a petal will display the representative post of the current date, and clicking the pistil will display the representative word cloud of the current post collection; a hexagonal grid is used to place the flower-shaped visual graphic character, and each flower-shaped visual graphic character is placed in the center of a hexagon. Flower-shaped visual graphic characters from different platforms are arranged in different directions, and flower-shaped visual graphic characters on the same platform are organized together in a honeycomb cluster style; At the same time, an algorithm based on a hexagonal grid is used to arrange the message propagation path from the flower-shaped visual image characters of the source topic to the flower-shaped visual image characters of the target topic, and the width of the message propagation path is used to encode the size of the cross-platform propagation probability at the topic level, and arrows are added to the message propagation path to indicate the propagation direction. The color of the message propagation path is encoded using the color of the source platform. It supports hovering the mouse over the message propagation path to illuminate the current path. After left-clicking the path, three seeds will appear on the path. Each seed represents a class of significant cross-platform cases, including the source post and the target post containing the same URL in the content, the target post containing the URL of the source post, and the target post containing certain specific references to the source post. It supports users to right-click on any path to jump to the post-level visual analysis interface, and further explore the cross-platform message propagation relationship between the source post set and the target post set connected by the path at the post level.

9. The method for visually analyzing message propagation across social platforms according to claim 1, characterized in that: The post-level visual analysis interface in the cross-platform communication visual analysis system includes a post-level communication view and a post-level detail view; wherein, in the post-level communication view, the communication relationship from post to post between different collections on the same topic on different platforms is displayed, and the cross-platform communication probability at the post level is visualized as the width of the message propagation path between posts, and the type of cross-platform communication pattern between posts is displayed based on the message propagation path between posts clicked by users; in the post-level detail view, the historical information of the posting account is displayed through the historical information relevance module, and the historical information includes the historical post list and interest distribution.

10. The method for visually analyzing message propagation across social platforms according to claim 9, characterized in that: In the post-level propagation view, each collection is visualized as a circle, and the posts in the circle are laid out using a bubble chart. The coding of the posts is based on their influence indicators. If the sum of the number of reposts, comments and likes of a post is greater than or equal to the first set threshold or the number of followers of the posting account is greater than or equal to the second set threshold, the post is visualized as a flower, and the size of the flower is proportional to the influence indicator. The color of the inner circle of the flower represents the emotion of the post. Otherwise, the post is visualized as a dot, and its color represents the emotion of the post. Users can adjust the first set threshold and the second set threshold through the control bar. Between collections, the cross-platform message propagation path extends from one post to another, and the width of the message propagation path is encoded by the cross-platform probability at the post level. The interface supports adjusting the number of displayed paths according to user needs, and uses edge bundling technology for path layout. When the user adjusts the first set threshold and / or the second set threshold, unimportant paths will be filtered out. When the control bar of the first set threshold and the second set threshold is dragged to the far right, that is, the maximum value, only the most important path will be retained on the interface. Clicking a flower or a dot representing a post will display the detailed content of the post.

Images