Information pushing method for breaking information cocoon houses in information spreading process
By identifying and breaking the information cocoon during information dissemination, and using the mutual push of the diffusion extension belt and the introduced account, the problems of group polarization and information narrowing caused by the information cocoon are solved, and the diversity and breadth of information interaction are achieved.
Patent Information
- Application Number
- CN202510145756.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-10
- Publication Date
- 2025-05-30
AI Technical Summary
Information cocoons lead to group polarization, narrowing information and differentiated group separation. Especially in the context of user immersive application of artificial intelligence, the formation and impact of information cocoons are more hidden and profound.
By collecting and analyzing the account's participation information on multiple topics during the information dissemination process, an N-dimensional coordinate system is constructed, the information cocoon account is identified, and the information cocoon is broken through the mutual push of the diffusion extension belt and the introduced account.
Effectively break the information cocoon, weaken the polarization of group, narrowing of information and differentiated group separation, and promote the diversity and breadth of information interaction.
Smart Images
Figure CN120067444A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the fields of information technology and big data, and particularly relates to a method for breaking the information cocoon in the process of information dissemination. Background Art
[0002] With the development of the Internet, the personalized push function of information publishing software has become more and more precise, leading to the emergence of information cocoons.
[0003] Problems such as group polarization, information narrowing, and differential group segregation brought about by information cocoons pose potential threats to the virtual social public domain - the cyber space. Especially in the context of users' immersive application of artificial intelligence, the formation and influence of information cocoons will be more concealed and profound.
[0004] The view dissemination mode in the information cocoon not only conforms to the form and characteristics of the classical view dissemination model, but also has the unique view diffusion characteristics of homogeneous selection, homogeneous content, and homogeneous groups.
[0005] First of all, in the information cocoon, users usually only focus on the information they are interested in or pushed, resulting in the singularity and repetition of the consumed information content. The homogeneous selection of users can be abstracted as the growth mechanism of the user group in the information cocoon. Secondly, once users form specific information preferences, their information space will be relatively stable, and external information is difficult to penetrate. This kind of closure is not only reflected in the information content, but also in that users often rely on fixed channels and tools to obtain information. The stability and closure of the personal information space can be abstracted as the degree of homogeneity of views or interests among individuals in the information cocoon. Finally, due to the common or similar knowledge effect, users will form groups with common cognitive structures and values. Summary of the Invention
[0006] Therefore, on the premise of retaining personalized push, it is imperative to break the information cocoon in the process of information dissemination by certain means.
[0007] In view of the above problems existing in the prior art, the present invention proposes an information push method that can effectively identify and break the information cocoon.
[0008] Specifically, the present invention provides an information push method for breaking the information cocoon in the process of information dissemination, and the method includes:
[0009] (1) Taking a predetermined time interval as a basic time period, collecting the account information and content information participating in each relevant topic on the target network platform, and for each topic, extracting all active and passive participation situations of each account in the participating accounts regarding this topic;
[0010] (2) Construct an N-dimensional coordinate system. In the constructed N-dimensional coordinate system, allocate an N-dimensional space of a certain scale to each theme based on the participation activity of each theme.
[0011] (3) For the current period t 0 , for each account, retrieve all the active and passive participation records of the account in each relevant theme. For each relevant participation record, form a participation record point for the account. Each participation record point is located in the N-dimensional space of the corresponding theme, forming a set of participation information points for the account.
[0012] (4) Construct a minimum bounding box for the set of participation information points of the account, include a predetermined proportion of the points in the set of participation information points in the minimum bounding box, and determine the spatial size of the minimum bounding box. When the spatial size of the minimum bounding box is less than the first predetermined threshold, determine that the active or passive participation information of the corresponding account is in the information cocoon.
[0013] (5) Repeat step (4) to construct the set of participation information points of each account, and screen out all the account information in the information cocoon, which is called the cocoon account. Find the positions of all the active participation information points outside the minimum bounding box in the set of participation information points of the cocoon account. Connect each (if there are too many discrete points, then connect the participation points with a density exceeding a certain value per unit area) active participation information point to its nearest active participation information point to form a diffusion extension band. For each diffusion extension band, determine whether the diffusion extension band passes through the minimum bounding box of other accounts. If it passes through, call the other account the introduced account. Based on the themes participated in on the diffusion extension band of the cocoon account, find the active and / or passive participation information of the introduced account, screen out the participation information that is similar or opposite to the participation information of the cocoon account on the same theme, and push the participation information of the introduced account and the relevant theme content and the participation information of the cocoon account and the relevant theme content to each other.
[0014] (6) Monitor all the active or passive participation information of the cocoon account in the next time period t 1 , calculate the minimum bounding box of the cocoon account in the time period t 1 , determine whether the spatial size of the minimum bounding box is less than the second predetermined threshold. If it is less than, calculate its diffusion extension band according to step (5). For each diffusion extension band in the time period t 1 , connect it to the time period t 0Compare with the diffusion extension zone with the closest distance, and judge the change amount of the number of active participation point sets in this diffusion extension zone. If the number increases, re-determine the relevant cocoon room accounts within the scope of the theme covered by this diffusion extension zone, and increase the mutual push between the relevant accounts and the cocoon room accounts in the diffusion extension zone where the number increases. Otherwise, if the number of active participation point sets in all the diffusion extension zones of the cocoon room accounts does not increase, expand the diffusion extension zone by a predetermined space size, and re-determine the introduced accounts within the scope of the theme covered by each diffusion extension zone of this cocoon room account. Based on the theme participated in on the re-determined diffusion extension zone, search for the active and / or passive participation information of the introduced accounts, screen out the participation information that is semantically similar or opposite to the participation information of the cocoon room account on the same theme, and push the participation information and relevant theme content of the introduced accounts and the participation information and relevant theme content of the cocoon room account to each other.
[0015] In a preferred implementation manner, the N-dimensional space is a three-dimensional space, a four-dimensional space, and more-dimensional spaces. The step (2) includes taking any one theme as the central theme, calculating the semantic correlation between each theme and this theme, using the semantic correlation distance between the two as the distance between the two themes, setting several themes with the closest distance to the central theme as the first circle of peripheral themes, arranging them around the central theme according to the correlation distance with the central theme, determining one or more themes with the smallest correlation distance from each first circle of peripheral theme as the second circle of peripheral themes, and so on, setting multiple circles of outer themes based on the correlation with the inner circle themes. For each theme, divide a certain scale of N-dimensional space for it respectively.
[0016] In another preferred implementation manner, judge whether there are entanglement points between the diffusion extension zones of any two cocoon room accounts. If there are entanglement points, increase the push of the relevant theme content within the predetermined range of the entanglement points to the cocoon room accounts participating in the entanglement. The entanglement points mentioned here refer to that any two cocoon room accounts have actively participated in any relevant theme multiple times. The present invention does not collide the cocoon room account with a general account, but collides the cocoon room account with the cocoon room account, which can utilize the stubbornness in the thinking of the cocoon room account owner to pull each other, change the inherent thinking loop of the cocoon room account, and achieve twice the result with half the effort.
[0017] In another preferred implementation manner, determine the central theme based on the popularity of the theme, and take the theme with the largest participation volume as the central theme.
[0018] In another preferred implementation manner, in the step (3), for each participation, set it at a point with a distance of d from the center of the space in the space where it is located according to the correlation distance d between the participation content and the core keyword of the theme.
[0019] In another preferred implementation, the second predetermined threshold is greater than the first predetermined threshold, and the relevant accounts are those whose minimum bounding boxes overlap with the range covered by the diffusion band of the target cocoon account.
[0020] In another preferred implementation, three-dimensional spaces of different sizes are divided for different themes based on the heat weights of the themes.
[0021] In another preferred implementation, the theme to which the information actively participated by the cocoon account in the target cocoon account belongs is used as the central theme.
[0022] Beneficial effects
[0023] By using the method of the present invention, the information cocoon can be effectively broken, and group polarization, information narrowing, and differential group segregation in the information dissemination process can be weakened.
[0024] Through the method of the present invention, through the entangled push of relevant accounts in the extended diffusion band of the cocoon account, even after the push stops, the cocoon account can get rid of the cocoon bondage and form a larger-range information interaction in the subsequent activity process. Description of the drawings
[0025] Figure 1 It is a schematic flowchart of the method of the present invention.
[0026] Figure 2 It is a schematic diagram of the extended diffusion band of a cocoon account in an embodiment of the present invention. The cocoon account located in the center is drawn in the figure, and its three extended diffusion bands respectively pass through the core areas of three other cocoon account point sets; Detailed implementation manners
[0027] The present invention will be described in detail below with reference to the accompanying drawings and their embodiments, but the protection scope of the present invention is not limited to the scope described in the embodiments.
[0028] The method of the present invention includes the following steps:
[0029] (1) Collect the account information and content information of the discussions on multiple relevant themes on the target network platform. For each theme, extract all the participation contents of each account in the participating accounts for that theme. Preferably, extract keywords from the participation contents of each account for that theme, and determine the semantic direction of the account based on the extracted keywords; it should be noted that the "theme" mentioned in the present invention needs to be understood in a broad sense, which can be the relevant content of a certain type of event or the content of a certain classification section. For example, the theme can be stars, variety shows, real estate, vehicles, etc. If the number of cocoon rooms judged according to the currently constructed N-dimensional space is less than a certain value, the central theme is replaced and the space construction is restarted.
[0030] (2) Construct an N-dimensional coordinate system. In the constructed N-dimensional coordinate system, based on the participation activity of each theme (such as the total amount of active discussions within a certain period of time), allocate an N-dimensional space of a certain scale to each theme. In this embodiment, a three-dimensional example is used for description. In one implementation, the total amount of active discussions is calculated using the following formula: ∑a 1 *λ 1 +a 2 *λ 2 +a 3 *λ 3 , λ 1 , λ 2 , λ 3 are respectively the number of person-times of strong participation, moderate participation, and weak participation. For example, strong participation includes actively posting new information exceeding a predetermined number of words, moderate participation includes actively posting information with fewer words than the predetermined number or replying to others' information, and weak participation includes giving likes, mechanical replies, etc. a 1 , a 2 , a 3 are respectively the weight coefficients of strong participation, moderate participation, and weak participation.
[0031] Taking any one theme as the central theme, calculate the semantic relevance between each theme and this theme, use the semantic relevance distance between the two as the distance between the two themes, set several themes with the closest distance to the central theme as the first-layer peripheral themes, and set them around the central theme according to the relevance distance between them. And so on, calculate the semantic relevance between other themes and each peripheral theme, and determine one or more themes with the smallest relevance distance to each first-layer peripheral theme as the second-layer peripheral themes (when setting the spatial distribution of themes, it is also possible to calculate the relevance distance between each peripheral theme and multiple inner-layer themes for each peripheral theme, and set it at the layer position with the smallest comprehensive distance to multiple themes). And so on, set the outer-layer themes based on the relevance to the inner-layer themes. For each theme, allocate an N-dimensional space of a certain scale to it respectively. Here, a three-dimensional example is used for discussion. In one implementation, the same-sized three-dimensional space is allocated to each theme; in another implementation, different-sized three-dimensional spaces are allocated to different themes based on the heat weight of the theme.
[0032] (3) For each account, retrieve all the active and passive participation records of the account in each relevant topic. For each relevant participation record, form a participation record point for the account. Each participation record point is located in the three-dimensional space of the corresponding topic. A core position point (based on the core content of the topic) is set for each topic, and according to the relevance distance d from the core topic, it is set at a point in the three-dimensional space with a distance of d from the center of the three-dimensional space, forming a set of participation information points for the account. Here, the active participation mentioned refers to the participation forms that require subjective awareness intervention, such as the active information release and information reply of the account, and the passive participation refers to the participation forms of passive acceptance of information, such as the relevant information pushed and browsed passively.
[0033] (4) Construct a minimum bounding box for the account participation point set, and include more than a predetermined proportion, such as more than 90% of the points in the participation point set in this minimum bounding box. Determine the spatial size of the minimum bounding box. When the spatial size of the minimum bounding box is less than the first predetermined threshold (the concentration threshold of the participation point set), it is determined that the active or passive participation information of the corresponding account is in the information cocoon.
[0034] (5) Repeat step (4) to construct the participation point sets of each account, and screen out all the account information in the information cocoon, which is called the cocoon account. Locate the positions of all the active participation information points outside the minimum bounding box in the participation information of the cocoon account, and connect each active participation information point to its nearest active participation information point to form a diffusion extension band. For each diffusion extension band, determine whether the diffusion extension band passes through other accounts and point cloud dense areas, preferably the minimum bounding boxes of other cocoon accounts. If it passes through, the other cocoon account is called the introduced account. Based on the topics participated in on the diffusion extension band of the cocoon account, retrieve the active and / or passive participation information of the introduced account, screen out the participation information that is similar or opposite to the participation information of the cocoon account on the same topic, and push the participation information of the introduced account and the relevant topic content and the participation information of the cocoon account and the relevant topic content to each other.
[0035] Figure 2 Shown is a distribution diagram of an exemplary cocoon account participation point set, in which the cocoon account located in the center is drawn, and three extension diffusion bands extending from it pass through the core areas of the point sets of three other cocoon accounts respectively.
[0036] (6) Monitor all active or passive participation information of the cocoon room account in the next time period t1, calculate the minimum bounding box of the cocoon room account in period t1, and determine whether the spatial size of the minimum bounding box is less than the second predetermined threshold, where the second predetermined threshold is greater than the first predetermined threshold. If it is less, calculate its diffusion extension band according to step (5), compare the diffusion extension band with the diffusion extension band closest to the t0 period, and judge the change amount of the number of active participation information (set of active participation points) in this diffusion extension band. If the number increases, re-determine the relevant cocoon room accounts within the scope of the theme covered by this diffusion extension band. The relevant cocoon room accounts mentioned here refer to the cocoon room accounts whose minimum bounding boxes overlap with the scope covered by the diffusion band. If the number of active participation information (set of active participation points) in all the diffusion extension bands of the cocoon room account does not increase, expand the diffusion extension band by a predetermined spatial size (for example, increase the three-dimensional coordinate points at the boundary of the diffusion extension band by 1-2 basic coordinate units), and re-determine the introduced accounts (expanded interactive push account scope) within the scope of the theme covered by each diffusion extension band of this cocoon room account. Based on the themes participated in on the diffusion extension band of the cocoon room account, find the active and / or passive participation information of the introduced accounts, screen out the participation information that is semantically similar or opposite to the participation information of the cocoon room account on the same theme, and push the participation information and relevant theme content of the introduced accounts and the participation information and relevant theme content of the cocoon room account to each other.
[0037] Example
[0038] To verify the effect of this invention application, a simplified simulation experiment was conducted. A three-dimensional space containing 27 spatial theme blocks was constructed, and each theme had the same spatial area. From the cocoon room account data that had been mastered, 50 cocoon room accounts were selected and projected into the three-dimensional space (the side length of each theme was set to 1-3 grids), and the minimum bounding boxes and diffusion extension bands of the 50 cocoon room accounts were constructed respectively. 10 cocoon room accounts whose diffusion extension bands passed through the point cloud core area (minimum bounding box) of other cocoon room accounts were selected from the 50 cocoon room accounts for a data push experiment. After the push experiment was carried out for 3 cycles (with a cycle of 2 days), the push was stopped and a measurement experiment was conducted. The participation data in this cycle was projected into the three-dimensional space again, and the size of the minimum bounding box of the cocoon room account was determined again. It was found that the sizes of the minimum bounding boxes of 10 of the accounts exceeded the cocoon room account threshold, and the sizes of the minimum bounding boxes of 9 of the accounts exceeded the cocoon room account threshold by a large margin (more than 50%).
[0039] The specific experimental data is as follows:
[0040]
[0041]
[0042] Comparative Example
[0043] Similar to the above example, a three-dimensional space containing 27 spatial theme blocks was constructed, and each theme had the same spatial area. From the mastered cocoon room account data, 20 cocoon room accounts and 30 general accounts were selected and projected into the three-dimensional space. The minimum bounding boxes and diffusion extension zones of the 20 cocoon room accounts were constructed respectively. Ten cocoon room accounts whose diffusion extension zones passed through the core areas (minimum bounding boxes) of other account point clouds were selected from the 20 cocoon room accounts for data push experiments. After the push experiment was carried out for 3 cycles, the push was stopped and a measurement experiment was carried out. The data projected into the three-dimensional space in this cycle was extracted again, and the size of the minimum bounding box of the cocoon room account was judged again. It was found that the sizes of the minimum bounding boxes of 7 of these accounts exceeded the cocoon room account threshold, and the sizes of the minimum bounding boxes of 5 of these accounts greatly exceeded the cocoon room account threshold.
[0044] The specific experimental data is as follows:
[0045]
[0046] It can be seen from the data comparison that through the method of the present invention, by means of the entangled push of related accounts in the extended diffusion zone of the cocoon room account, even after the push is stopped, the cocoon room account can get rid of the cocoon room bondage and form a larger range of information interaction in the subsequent activity process.
[0047] Although the principle of the present invention has been described in detail above in connection with the preferred embodiments of the present invention, those skilled in the art should understand that the above embodiments are only explanations of the illustrative implementation manners of the present invention, and do not limit the scope of the present invention. The details in the embodiments do not constitute a limitation on the scope of the present invention. Without departing from the spirit and scope of the present invention, any obvious changes such as equivalent transformations and simple replacements based on the technical solution of the present invention fall within the protection scope of the present invention.
Claims
1. An information push method for breaking the information cocoon during information dissemination, characterized in that: The method comprises: (1) Using a predetermined time interval as a basic time period, collect account information and content information on the target network platform that participate in each relevant topic. For each topic, extract all active and passive participation information of each account in the participating accounts on the topic; (2) Constructing an N-dimensional coordinate system, in which an N-dimensional space of a certain scale is allocated to each topic based on the participation activity of each topic; (3) For the current period t0, for each account, retrieve all active and passive participation records of the account in each relevant topic, and for each relevant participation record, form a participation record point for the account. Each participation record point is located in the N-dimensional space of the corresponding topic, forming a participation information point set for the account; (4) constructing a minimum bounding box for the account's participation information point set, including a predetermined proportion of points in the participation information point set in the minimum bounding box, and determining the spatial size of the minimum bounding box. When the spatial size of the minimum bounding box is less than a first predetermined threshold, it is determined that the active or passive participation information of the corresponding account is in the information cocoon. (5) Repeat step (4) to construct a participation information point set for each account, and filter out all account information in the information cocoon, referred to as the cocoon account, find the positions of all active participation information points outside the minimum bounding box in the participation information point set of the cocoon account, connect the active participation information points whose density exceeds the threshold with their nearest active participation information points to form a diffusion extension belt, and for each diffusion extension belt, determine whether the diffusion extension belt passes through the minimum bounding box of other accounts. If so, the other account is referred to as the introduced account, and find the active and / or passive participation information of the introduced account based on the topic of participation in the diffusion extension belt of the cocoon account, filter out the participation information that is similar or opposite to the participation information of the cocoon account on the same topic, and push the participation information and related topic content of the introduced account to the participation information and related topic content of the cocoon account; (6) Monitor all active or passive participation information of the cocoon account in the next time period t1, calculate the minimum bounding box of the cocoon account in the time period t1, and determine whether the spatial size of the minimum bounding box is less than the second predetermined threshold. If so, calculate its diffusion extension belt according to step (5). For each diffusion extension belt in the time period t1, compare it with the diffusion extension belt closest to it in the time period t0, and determine the change in the number of active participation point sets in the diffusion extension belt. If the number increases, re-determine the relevant cocoon accounts within the subject scope covered by the diffusion extension belt, and increase the number of relevant cocoon accounts in the diffusion extension belt where the number increases. Mutual push between accounts and cocoon accounts. Otherwise, if the number of active participation point sets in all diffusion extension belts of the cocoon account has not increased, the diffusion extension belt will be expanded outward with a predetermined spatial size, and the introduced accounts within the subject range covered in each diffusion extension belt of the cocoon account will be redetermined. The active and / or passive participation information of the introduced accounts will be searched based on the subjects involved in the redetermined diffusion extension belts, and the participation information with similar or opposite semantics to the participation information of the cocoon account on the same subject will be screened out, and the participation information of the introduced accounts and related subject content will be pushed to each other with the participation information and related subject content of the cocoon account.
2. The information push method for breaking the information cocoon during information dissemination according to claim 1 is characterized in that: The N-dimensional space is a three-dimensional space, a four-dimensional space, or a higher-dimensional space. The step (2) includes taking any one topic as the central topic, calculating the semantic relevance between each topic and the topic, using the semantic relevance distance between the two topics as the distance between the two topics, setting several topics closest to the central topic as first-circle peripheral topics, setting them on the periphery of the central topic according to the relevance distance with the central topic, determining one or more topics with the smallest relevance distance with each first-circle peripheral topic as second-circle peripheral topics, and so on. Multiple circles of outer circle topics are set based on the relevance with the inner circle topics, and for each topic, an N-dimensional space of a certain scale is divided for it.
3. The information push method for breaking the information cocoon during information dissemination according to claim 2 is characterized in that: Determine whether there is an entanglement point between the diffusion extension zones of any two cocoon accounts. If there is an entanglement point, increase the push of relevant subject content within the predetermined range of the entanglement point to the cocoon accounts involved in the entanglement.
4. The information push method for breaking the information cocoon during information dissemination according to claim 3 is characterized in that: Determine the central topic based on the popularity of the topic, and take the topic with the largest number of participants as the central topic.
5. The information push method for breaking the information cocoon during information dissemination according to claim 1 is characterized in that: In the step (3), for each participation, the distance between the participating content and the core keywords of the theme is set at the point in the space where the distance from the center of the space is.
6. The information push method for breaking the information cocoon during information dissemination according to claim 1 is characterized in that: The second predetermined threshold is greater than the first predetermined threshold, and the related accounts are accounts whose minimum bounding box overlaps with the diffusion zone of the target cocoon account.
7. The information push method for breaking the information cocoon during information dissemination according to claim 2 is characterized in that: Based on the popularity weight of the topic, different topics are divided into three-dimensional spaces of different sizes.