Determination method of vertical polarization index and related device

Abstract

The invention discloses a vertical polarization index determination method and a related device, and relates to the technical field of data processing, and the method comprises the steps: obtaining topic interaction related data of each participating user of a target topic, the topic interaction related data of each participating user comprises at least one piece of comment data and at least one piece of interaction feedback data aiming at the target topic, determining the standing site of each participating user, the number of people of each standing site camp and the number of sub viewpoints of each standing site camp according to the comment data of each participating user, and according to the interaction feedback data of each participating user, the standing of each participating user, the number of people of each standing camp and the number of sub viewpoints of each standing camp, determining a standing polarization index of the target topic. According to the method and the device, the standing site of each participating user can be determined based on the comment data of each participating user, the number of people and the number of sub viewpoints of each standing site camp are further counted, and on the basis, participation of interactive feedback data and multi-dimensional data is also referred, so that the accuracy of the standing site polarization index is improved.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method and apparatus for determining a position polarization index. Background Technology

[0002] In discussions, such as news, elections, academic discussions, and trending events, different users often hold different viewpoints. As the discussion escalates, two or even more opposing camps may emerge.

[0003] Therefore, there is an urgent need for a method to determine a position polarization index that measures the degree of divergence in group opinions, in order to help analyze the current state of a topic, optimize decision-making, and predict the trend of the topic. Summary of the Invention

[0004] In view of the above problems, this application provides a method and related apparatus for determining a position polarization index, so as to determine the position polarization index of a target topic and thus measure the degree of divergence in group opinions on the target topic. The specific solution is as follows:

[0005] The first aspect of this application provides a method for determining a position polarization index, comprising:

[0006] Obtain topic interaction-related data for each participating user of the target topic, wherein the topic interaction-related data for each participating user includes at least one comment and at least one interaction feedback for the target topic;

[0007] Based on the comment data of each participating user, determine the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp;

[0008] The polarization index of the target topic is determined based on the interaction feedback data of each participating user, the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp.

[0009] In one possible implementation, determining the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp based on the comment data of each participating user includes:

[0010] The stance of each participating user is determined based on the comment data of each participating user.

[0011] Based on the positions of the participating users, determine the number of people in each of the aforementioned positions / camps;

[0012] The comment data of each participating user is clustered by stance to obtain the comment data of each stance camp.

[0013] The comment data of each of the aforementioned positions is clustered into sub-viewpoints based on semantic similarity to obtain sub-viewpoint clusters for each of the aforementioned positions; the number of sub-viewpoint clusters for each of the aforementioned positions is taken as the number of sub-viewpoints for each of the aforementioned positions.

[0014] In one possible implementation, determining the stance of each participating user based on their comment data includes:

[0015] The comment data of each participating user is fed into a pre-configured stance classifier to obtain the stance label and confidence score corresponding to the comment data of each participating user.

[0016] Based on the stance label and confidence score corresponding to the comment data of each participating user, the reliability stance corresponding to the comment data of each participating user is determined;

[0017] The stance of each participating user is determined based on the reliability stance corresponding to the comment data of each participating user.

[0018] In one possible implementation, determining the polarization index of the target topic based on the interaction feedback data of each participating user, the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp includes:

[0019] Based on the interaction feedback data of each participating user and the position of each participating user, the intensity of cross-faction conflict is determined. The intensity of cross-faction conflict is used to measure the proportion of negative feedback with conflict characteristics in cross-faction interactions.

[0020] The camp balance is determined based on the number of people in each of the aforementioned camps. The camp balance is used to measure whether the number of people in each of the aforementioned camps is balanced.

[0021] The degree of fragmentation of sub-viewpoints is determined based on the number of people in each of the aforementioned positions and the number of sub-viewpoints in each of the aforementioned positions. The degree of fragmentation of sub-viewpoints is used to measure the stability of each of the aforementioned positions.

[0022] The position polarization index of the target topic is determined based on the intensity of cross-faction conflict, the degree of faction balance, and the degree of fragmentation of sub-viewpoints.

[0023] In one possible implementation, determining the intensity of cross-faction conflict based on the interaction feedback data of each participating user and the stance of each participating user includes:

[0024] Determine the weight of each interactive feedback data point from each participating user;

[0025] Based on the position of each participating user and the weight of each interactive feedback data of each participating user, determine whether each interactive feedback data of each participating user is conflicting interactive data; to obtain all conflicting interactive data;

[0026] Based on the respective weights of all the conflict interaction data, determine the cross-faction negative interaction weights;

[0027] The total weight of the interaction is determined based on the weight of each user's interaction feedback data.

[0028] The intensity of the cross-faction conflict is determined based on the cross-faction negative interaction weight and the total interaction weight.

[0029] In one possible implementation, determining whether each piece of interactive feedback data from each participating user is conflicting interactive data, based on the position of each participating user and the weight of each piece of interactive feedback data from each participating user, includes:

[0030] For each interaction feedback data of each participating user:

[0031] Determine whether the stance of the parent user and the participating user corresponding to the interactive feedback data belong to different stance camps;

[0032] If so, determine whether the interaction type of the interactive feedback data belongs to the preset set of negative feedback types;

[0033] If so, then the interactive feedback data is identified as conflicting interactive data.

[0034] In one possible implementation, determining the faction balance based on the number of members in each of the stated factions includes:

[0035] Based on the number of people in each of the stated positions, determine the absolute difference in the number of people between any two stated positions and the total number of people;

[0036] The balance of the factions is determined by the absolute difference in the number of people between each pair of the stated factions and the total number of people.

[0037] In one possible implementation, determining the fragmentation degree of sub-viewpoints based on the number of members in each of the aforementioned factions and the number of sub-viewpoints in each of the aforementioned factions includes:

[0038] The internal fragmentation degree of each of the aforementioned positions is determined based on the number of sub-viewpoints in each position camp.

[0039] The overall fragmentation level is determined based on the internal fragmentation level of each stated faction and the number of people in each stated faction.

[0040] The fragmentation degree of a sub-viewpoint is determined based on the overall fragmentation degree and the preset upper limit of the number of sub-viewpoints.

[0041] A second aspect of this application provides an apparatus for determining a position polarization index, comprising:

[0042] The data acquisition module is used to acquire topic interaction-related data of each participating user of the target topic. The topic interaction-related data of each participating user includes at least one comment data and at least one interaction feedback data for the target topic.

[0043] The stance statistics module is used to determine the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp based on the comment data of each participating user.

[0044] The polarization index determination module is used to determine the polarization index of the target topic based on the interaction feedback data of each participating user, the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp.

[0045] A third aspect of this application provides a computer program product including computer-readable instructions that, when executed on an electronic device, cause the electronic device to implement the method for determining the position polarization index of the first aspect or any implementation thereof.

[0046] A fourth aspect of this application provides an electronic device, comprising at least one processor and a memory connected to the processor, wherein:

[0047] The memory is used to store computer programs;

[0048] The processor is used to execute the computer program to enable the electronic device to implement the method for determining the position polarization index of the first aspect or any implementation thereof.

[0049] The fifth aspect of this application provides a computer storage medium carrying one or more computer programs, which, when executed by an electronic device, enable the electronic device to implement the method for determining the field polarization index of the first aspect or any implementation thereof.

[0050] By employing the aforementioned technical solution, the method for determining the position polarization index provided in this application obtains topic interaction-related data from each participating user on the target topic. The topic interaction-related data for each participating user includes at least one comment and at least one interactive feedback data related to the target topic. Based on the comment data of each participating user, the position of each participating user, the number of people in each position camp, and the number of sub-viewpoints in each position camp are determined. Based on the interactive feedback data of each participating user, the position of each participating user, the number of people in each position camp, and the number of sub-viewpoints in each position camp, the position polarization index of the target topic is determined. Therefore, this application can determine the position of each participating user based on their comment data, and then statistically analyze the number of people in each position camp and the number of sub-viewpoints. In addition to the position statistical results, this application also refers to the interactive feedback data of each participating user to determine the position polarization data of the target topic. The participation of multi-dimensional data improves the accuracy of the position polarization index. Attached Figure Description

[0051] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0052] Figure 1 A schematic diagram of a system architecture provided in this application;

[0053] Figure 2 A flowchart illustrating a method for determining a position polarization index provided in this application;

[0054] Figure 3 A schematic diagram of the structure of a device for determining the field polarization index provided in this application;

[0055] Figure 4 This is a schematic diagram of the structure of an electronic device provided in this application. Detailed Implementation

[0056] The embodiments of this application are described below with reference to the accompanying drawings. The terminology used in the implementation section of this application is for explaining specific embodiments only and is not intended to limit the scope of this application.

[0057] The embodiments of this application will now be described with reference to the accompanying drawings. Those skilled in the art will recognize that, with technological advancements and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are equally applicable to similar technical problems.

[0058] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms are interchangeable where appropriate; this is merely a way of distinguishing objects with the same attributes in the embodiments of this application. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, so that a process, method, system, product, or apparatus that comprises a series of elements is not necessarily limited to those elements, but may include other elements not explicitly listed or inherent to those processes, methods, products, or apparatuses.

[0059] Understandably, with the widespread use of social media and news platforms, comment sections have become an important channel for users to express their opinions. In comment sections for breaking news or trending events, user opinions often quickly diverge into different camps, and even fierce confrontations arise. Therefore, platforms urgently need a system that can identify, quantify, and provide early warnings of "polarization risks" in real time, helping administrators to promptly detect escalating confrontations, information bubbles, and potential risks of loss of control, and to effectively manage these situations.

[0060] Based on this, this application provides a method for determining the position polarization index, which can be applied to, for example... Figure 1 The system architecture shown includes a terminal 100 and a server 200. The server 200 may include one or more servers (…). Figure 1 (This example uses a server as an illustration).

[0061] Either terminal 100 or server 200 can be used independently to execute the method for determining the position polarization index provided in the embodiments of this application. Alternatively, terminal 100 and server 200 can also be used collaboratively to execute the method for determining the position polarization index provided in the embodiments of this application.

[0062] The following description Figure 1 The product form of the mid-terminal 100;

[0063] The terminal 100 in this application embodiment can be a mobile phone, tablet computer, wearable device, vehicle device, augmented reality (AR) / virtual reality (VR) device, laptop computer, ultra-mobile personal computer (UMPC), netbook, personal digital assistant (PDA), etc., and this application embodiment does not impose any restrictions on it.

[0064] To enable those skilled in the art to better understand this application, the method for determining the field polarization index of embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0065] Reference Figure 2 , Figure 2 A flowchart illustrating a method for determining a position polarization index provided in this application embodiment is shown below. Figure 2 As shown, the method for determining the position polarization index may include:

[0066] Step S201: Obtain topic interaction-related data for each participating user of the target topic. The topic interaction-related data for each participating user includes at least one comment and at least one interaction feedback for the target topic.

[0067] In this embodiment, the target topic includes, but is not limited to, the following topics: news topics, elections, academic topics, hot topic topics, etc.

[0068] Optionally, participating users can refer to users who have participated in discussions on the target topic in the past, or users who have participated in discussions on the target topic in the past and are still active. Optionally, whether a user is an active user can be determined by the timestamp of the user's last participation in the target topic.

[0069] In this embodiment, the ways in which users participate in the target topic include, but are not limited to: commenting, liking, disliking, and replying.

[0070] In order to determine the discussion status of each participating user on the target topic and to promptly detect escalation of confrontation, information bubbles, and potential risks of loss of control, this embodiment can obtain topic interaction-related data of each participating user on the target topic.

[0071] Specifically, for each participating user, the topic-related interaction data acquired in this embodiment includes: at least one comment and at least one interactive feedback data related to the target topic. Here, comment data includes data directly commented in the comment section, as well as data replied to based on other people's comments; interactive feedback data refers to various interactive feedback data such as liking, disliking, and replying based on other people's comments, as well as various interactive feedback data such as liking and disliking on the original text.

[0072] It is worth noting that in this embodiment, each comment data that is replied to based on other people's comments is both a comment data that participates in the processing of step S202 below, and an interactive feedback data that participates in the processing of step S203 below.

[0073] In one possible implementation, considering that the data such as comments, replies, likes, and dislikes from participating users may be insufficient to accurately determine the participating users' stance, optionally, each comment and each interaction feedback data may include the current data itself and the context data of the current data in order to accurately determine the user's stance.

[0074] For example, in one alternative approach, the following two types of key information for each participating user can be obtained through a data monitoring mechanism:

[0075] Category 1: Comment data, including but not limited to the following: comment text, target user identifier (IdentityDocument, ID), comment timestamp, parent comment identifier, and parent comment content.

[0076] For example, user 1 commented on the topic "Should I choose city A or city B for summer travel?" and said, "City A has an average summer temperature of 25-28℃, making it a natural summer resort, while city B has a midday temperature as high as 35℃, making it easy to get heatstroke during outdoor activities. Therefore, I recommend choosing city A." Based on this, user 2 replied, "I agree with the above opinion."

[0077] In the example above, a comment from user 2 includes: the comment text is "Agree with the above opinion", the target user identifier is "ID2" pointing to user 2, the comment timestamp is "2025-8-25 14:03", the parent comment identifier is "ID1" pointing to user 1, and the parent comment content is "Should I choose city A or city B for summer vacation?" The comment data is "City A has an average summer temperature of 25-28℃, making it a natural summer resort, while city B has a midday temperature as high as 35℃, making outdoor activities prone to heatstroke, so it is recommended to choose city A."

[0078] The second category is interactive feedback data, including but not limited to the following: initiator identifier (i.e., the identifier of other users who commented in the comment section), target user identifier (i.e., the identifier of the current user who interacted with the comments of others), behavior type (e.g., like, dislike (i.e., disagree), reply, etc.), behavior timestamp, and initiator's comment text.

[0079] For example, user 1 commented on the topic "Should I choose city A or city B for summer travel?" and said, "City A has an average summer temperature of 25-28℃, making it a natural summer resort, while city B has a midday temperature as high as 35℃, making it easy to get heatstroke during outdoor activities. Therefore, I recommend choosing city A." User 3 disliked this comment.

[0080] In the example above, one piece of interaction feedback data from participating user 3 includes: the initiator is identified as "ID1", the target user is identified as "ID3" pointing to participating user 3, the behavior type is "dislike", the behavior timestamp is "2025-8-25 15:22", and the initiator's comment text is "Should I choose city A or city B for summer travel?" The comment data is "City A has an average summer temperature of 25-28℃, making it a natural summer resort, while city B has a midday temperature as high as 35℃, making outdoor activities prone to heatstroke, so it is recommended to choose city A."

[0081] Of course, the specific data content included in the above two types of data can be other, and no specific limitations are made here.

[0082] Optionally, after obtaining the above data, it can be cleaned to remove invalid characters, redundant symbols, advertising information, duplicate content, etc. For example, the cleaned data can be standardized and output into the following two structures:

[0083] The first type of structure is the comment record, which includes the target user identifier, text content, timestamp, and parent comment identifier (if any).

[0084] The second type of structure is behavior records, which include the initiator's identifier, the initiator's comment text, the target user's identifier, the behavior type, and the timestamp.

[0085] It should also be noted that if a comment is a direct reply to the target topic, then the comment data does not contain contextual data. If the comment is not a direct reply to the target topic, then the comment data is formed by concatenating the current reply data and the commented data in chronological order.

[0086] Step S202: Based on the comment data of each participating user, determine the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp.

[0087] Considering that user comments, compared to "zero-cost" user behaviors such as likes and dislikes, are typically data reflecting user opinions, arguments, personal experiences, and other details, and are rich in emotion, they are usually the result of users investing time in thinking and organizing their thoughts. Therefore, this embodiment determines the stance of each participating user based on their comment data, resulting in a more accurate understanding of user stances.

[0088] As mentioned above, each participating user has at least one comment. In this embodiment, for each participating user, the stance can be determined based on one or more comment data from all the participating user's comment data. Then, based on the stances and comment data of all participating users, the number of people in each stance camp and the number of sub-viewpoints in each stance camp can be obtained.

[0089] Step S203: Determine the polarization index of the target topic based on the interaction feedback data of each participating user, the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp.

[0090] Here, the position polarization index is a quantitative indicator that measures the degree of differentiation in group opinions, attitudes, or behaviors. By quantifying the proportion of extreme positions or the differences between opposing camps, it can help analyze social phenomena, optimize decision-making, and predict trends.

[0091] The embodiments of this application can analyze the interactive feedback data of each participating user, the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp, so as to measure the stance split of each participating user on the target topic from multiple perspectives, and thus accurately measure the stance polarization index of the target topic.

[0092] The method for determining the position polarization index provided in this application obtains topic interaction-related data from each participating user on a target topic. The topic interaction-related data for each participating user includes at least one comment and at least one interaction feedback on the target topic. Based on the comment data of each participating user, the position of each participating user, the number of people in each position camp, and the number of sub-viewpoints in each position camp are determined. Based on the interaction feedback data of each participating user, the position of each participating user, the number of people in each position camp, and the number of sub-viewpoints in each position camp, the position polarization index of the target topic is determined. Therefore, this application can determine the position of each participating user based on their comment data, and then statistically analyze the number of people in each position camp and the number of sub-viewpoints. In addition to the position statistical results, this application also refers to the interaction feedback data of each participating user to determine the position polarization data of the target topic. The participation of multi-dimensional data improves the accuracy of the position polarization index.

[0093] In some embodiments of this application, the process of step S202, "determining the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp based on the comment data of each participating user," is described.

[0094] As described above, this embodiment first determines the stance of each participating user based on the comment data of each participating user.

[0095] Here, "determining the stance of each participating user based on their comment data" can be implemented in multiple ways, including but not limited to the following.

[0096] The first method involves analyzing the stance of each participating user based on all their comment data to determine their stance.

[0097] It is understandable that users' stances may change. For example, user 4 initially tended to support the target topic, but later shifted their opinion and became more opposed to it. Based on this, the following second approach is also provided.

[0098] The second method involves analyzing the stance of each participating user based on their most recent comment data to determine their stance.

[0099] Considering that not all comment data corresponds to a 100% certain user stance, and the above two methods directly analyze user stance to obtain user stance, there may be issues with inaccurate user stance. Therefore, the following third method is also provided.

[0100] The third method involves feeding the comment data of each participating user into a pre-configured stance classifier to obtain the stance label and confidence score corresponding to the comment data of each participating user. Based on the stance label and confidence score corresponding to the comment data of each participating user, the reliability stance corresponding to the comment data of each participating user is determined, and the stance of each participating user is determined based on the reliability stance corresponding to the comment data of each participating user.

[0101] Optionally, the aforementioned stance classifier can be trained based on a deep language model, wherein the training data includes: training comment data and corresponding user stances and confidence scores. Optionally, the confidence scores are in the range of 0-1.

[0102] Taking the position tags as an example, including support, opposition, and neutrality, this embodiment can optionally send each comment data of each participating user to the position classifier to obtain the position tag and confidence score corresponding to each comment data of each participating user. For example, if participating user 4 has 10 comment data, then the 10 comment data can be sent to the position classifier respectively to obtain the position tag and confidence score corresponding to the 10 comment data respectively.

[0103] For each comment from each participating user, if the confidence score is less than a preset threshold τ, the comment is determined to be neutral; if the confidence score is greater than or equal to the threshold τ, the comment is determined to be in the position indicated by the position label. This allows us to obtain the reliability of each participating user's comments.

[0104] Finally, for each participating user, the reliability stance corresponding to the most recent comment data of that participating user with a confidence score greater than or equal to the score threshold τ is determined as the participating user's stance.

[0105] Optionally, the score threshold τ can be determined by the validation set or the policy objective. For example, the score threshold τ can be in the range of 0.5-0.8. Of course, the score threshold τ can also be other values, which are not limited here.

[0106] It should also be noted that the above method of determining user stance is only an example, and there are other methods. For example, the reliability stance corresponding to the comment data with the highest confidence score among the comment data of the participating user that is greater than or equal to the score threshold τ can be determined as the participating user's stance.

[0107] By using confidence scores, the reliability of each comment data being identified as a stance indicated by a stance label can be accurately identified, thus making the user stance determined in this application more reliable.

[0108] After obtaining the positions of each participating user, this embodiment can determine the number of people in each position camp based on the positions of each participating user. For example, if there are 10,000 participating users in the target topic, of which 500 hold a supporting position, 400 hold an opposing position, and 100 hold a neutral position, then the number of people in each position camp includes 500 people in the supporting camp, 400 people in the opposing camp, and 100 people in the neutral camp.

[0109] Furthermore, the comment data of each participating user is clustered according to their stance to obtain comment data for each stance camp. For example, the comment data of each participating user is clustered according to the reliability stance corresponding to their comment data to obtain comment data for each stance camp.

[0110] Then, the comment data of each stance camp is clustered into sub-viewpoints based on semantic similarity to obtain sub-viewpoint clusters for each stance camp; the number of sub-viewpoint clusters obtained for each stance camp is taken as the number of sub-viewpoints for each stance camp.

[0111] In other words, taking the supporting and opposing camps as examples, this embodiment can cluster the comment data with a supporting or opposing stance within their respective camps according to semantic similarity to obtain the number of sub-viewpoints for each stance camp.

[0112] It is worth noting that since the polarization index is an indicator used to quantify the degree of divergence in opinions among user groups on a target topic, reflecting the trend of positions shifting from moderate to extreme, this embodiment can only analyze the opposing camps with conflicting opinions. For ease of description, the following uses the supporting and opposing camps as examples (this does not mean that this application is limited to the supporting and opposing camps). In this embodiment, the number of people in the supporting camp is denoted as N_pos, the number of people in the opposing camp is denoted as N_neg, the number of sub-opinions in the supporting camp is denoted as k_pos, and the number of sub-opinions in the opposing camp is denoted as k_neg.

[0113] In summary, this embodiment provides a "soft" approach to stance identification that combines comment data with confidence scores. This approach is closer to how humans naturally perceive complex stances, and the resulting stances are more accurate and reliable.

[0114] In some other embodiments of this application, the process of step S203 described above, "determining the position polarization index of the target topic based on the interactive feedback data of each participating user, the position of each participating user, the number of people in each position camp, and the number of sub-viewpoints in each position camp," will be introduced.

[0115] Research has found that the higher the intensity of conflict between opposing camps, the more easily the antagonistic perceptions between them become entrenched. Consequently, when users from different camps interact frequently and refuse to yield, user stances evolve from "difference" to "opposition," accelerating the shift from moderate to extreme positions. Simultaneously, high-intensity conflict can evoke strong negative emotions (such as anger and fear), pushing stances from rational discussion to emotional confrontation, thus leading to emotional polarization.

[0116] To this end, this embodiment provides a cross-faction conflict intensity index to measure the position polarization index from the perspective of conflict intensity. Specifically, this embodiment can determine the cross-faction conflict intensity based on the interaction feedback data of each participating user and the position of each participating user. The cross-faction conflict intensity is used to measure the proportion of negative feedback with conflict characteristics in cross-faction interactions.

[0117] The following details one possible process for "determining the intensity of cross-faction conflict based on the interactive feedback data and the positions of each participating user".

[0118] First, this embodiment can determine the weight of each piece of interactive feedback data for each participating user.

[0119] Optionally, for each piece of interactive feedback data, an initial weight can be preset based on the interaction type. This is used to reflect the basic impact of the interactive feedback data on the intensity of the conflict. For example, the initial weight of a negative response is greater than or equal to the initial weight of a dislike, which is greater than or equal to the initial weight of a like.

[0120] Optionally, the initial weights can be normalized to improve computational efficiency.

[0121] To reflect timeliness, this embodiment can also use the time difference between the generation time of each interactive feedback data and the current time t. (Optional, time unit is hours) Perform weight decay to obtain the weight of each interactive feedback data at the current time t, where the current time t refers to the time when the position polarization index is calculated.

[0122] Formula (1);

[0123] in, This represents the time decay factor, which can be set by the "half-life" (e.g., decays to half in 24 hours). This indicates the weight of the interactive feedback data at the current time t. This indicates the initial weight of the interactive feedback data.

[0124] Next, based on the position of each participating user and the weight of each interactive feedback data of each participating user, it is determined whether each interactive feedback data of each participating user is conflicting interactive data; in order to obtain all conflicting interactive data.

[0125] Optionally, the process of "determining whether each piece of interactive feedback data from each participating user is conflicting interactive data based on the position of each participating user and the weight of each piece of interactive feedback data from each participating user" may include: for each piece of interactive feedback data from each participating user, determining whether the position of the parent participating user corresponding to the interactive feedback data and the position of the participating user belong to different position camps; if so, determining whether the interaction type of the interactive feedback data belongs to a preset set of negative feedback types; if so, determining the interactive feedback data as conflicting interactive data. Here, the set of negative feedback types includes at least one negative feedback type, which refers to a type that explicitly or implicitly contains opposing semantics (such as opposition, replies with negative semantics).

[0126] In other words, this embodiment can perform the following two-step detection for each interactive feedback data of each participating user:

[0127] Cross-faction detection: Compare the positions of the two users at the time of the interaction. If they belong to the supporting camp and the opposing camp respectively, it is recorded as a cross-faction.

[0128] Negative type detection: Determines whether the interaction type of the interactive feedback data is a preset negative feedback type. Optionally, the negative type detection process can be implemented using preset rules or a pre-trained model.

[0129] An interactive feedback is marked as conflicting only if it simultaneously satisfies both cross-faction and negative feedback types.

[0130] For example, user 1's comment on the topic "Should I choose city A or city B for summer travel?" is: "City A has an average summer temperature of 25-28℃, making it a natural summer resort, while city B's midday temperature can reach 35℃, making outdoor activities risky for heatstroke. Therefore, I recommend choosing city A." User 5 replies: "No, city A has less greenery, and it feels very sunny when walking on the street. In comparison, most roads in city B are completely covered by trees on both sides, so it's not sunny and even quite cool. In addition, city B has a variety of tourist attractions, making it more suitable for tourism (interaction feedback data)." User 1 (the parent user) belongs to the camp that supports city A, while user 5 belongs to the camp that opposes city A (i.e., supports city B). Furthermore, user 5's reply contains a negative statement. Therefore, the interaction feedback data in the example is determined to be conflicting interaction data.

[0131] Through the above detection, all conflicting interaction data can be identified from all interactive feedback data of all participating users. In this embodiment, the cross-faction negative interaction weight can be determined according to the weight of each of the conflicting interaction data, and the total interaction weight can be determined according to the weight of each of the interactive feedback data of each participating user.

[0132] For example, the weights of all conflict interaction data are summed, and the sum is used as the cross-faction negative interaction weight. The weights of all interactive feedback data from each participating user are summed, and the sum is used as the total weight of the interaction. .

[0133] Finally, the intensity of cross-faction conflict is determined based on the negative interaction weights and total interaction weights across factions.

[0134] Optionally, the formula for calculating the intensity of cross-faction conflict is: ,in, Indicates the intensity of cross-faction conflict at the current moment (if If not greater than 0, then (0).

[0135] This embodiment provides a cross-faction conflict intensity index, which can reflect the proportion of negative behaviors with conflict nature in the current cross-faction interactions. The higher the index value, the more the conflict is concentrated between participating users with different positions.

[0136] Furthermore, the difference in the number of people in different camps will affect the way users in each camp express themselves and the degree of solidarity. For example, compared to camps with more people, camps with fewer people are more likely to attract attention through extreme behaviors (such as protests, cyberattacks, etc.) and are more likely to stick together in comments and replies, making it easier for antagonistic emotions to erupt.

[0137] To this end, this embodiment also provides a camp balance index to measure the position polarization index from the perspective of camp balance. Specifically, this embodiment can determine the camp balance based on the number of people in each camp. The camp balance is used to measure whether the number of people in each camp is balanced.

[0138] Optionally, the process for determining the balance of factions is as follows: based on the number of people in each faction, determine the absolute difference in the number of people between any two factions and the total number of people; based on the absolute difference in the number of people between any two factions and the total number of people, determine the balance of factions.

[0139] Taking the supporting and opposing camps as an example, the optional formula for calculating the camp balance is: ,in, This indicates the current balance of power between factions. This indicates the number of people supporting the current faction. This indicates the number of people in the opposition camp at the current moment.

[0140] In the above formula, the molecule The denominator represents the absolute difference in the number of people in the two camps, indicating the degree of imbalance. This represents the total number of people in both camps, used to normalize to the [0,1] interval.

[0141] This embodiment provides a camp balance index, which measures whether the number of users in the current supporting and opposing camps is balanced. The smaller the difference in the number of users, the more balanced the public opinion structure; the larger the difference, the more obvious the advantage one side has in the public opinion field.

[0142] Understandably, when there are multiple separate subgroups of viewpoints within the same camp, the stability of its stance will decrease, and the conflict of opinion between the camps will be more likely to recur, split, or even reorganize, thereby increasing the uncertainty of public opinion.

[0143] To this end, this embodiment also provides a sub-viewpoint fragmentation index to reveal the potential risk of "internal division" within camps, making the position polarization index more comprehensively reflect the true state of public opinion structure. Specifically, this embodiment can determine the sub-viewpoint fragmentation degree based on the number of people in each position camp and the number of sub-viewpoints in each position camp. This sub-viewpoint fragmentation degree is used to measure the stability of each position camp.

[0144] Optionally, in this embodiment, the internal fragmentation degree of each position camp can be determined first based on the number of sub-viewpoints of each position camp.

[0145] Optionally, the formula for calculating the internal fragmentation degree of a supporting faction is as follows: ,in, This indicates the degree of internal fragmentation within the supporting camp at the current moment. This represents the number of sub-viewpoints in the supporting camp at the current moment; similarly, the formula for calculating the internal fragmentation degree of the opposing camp is: ,in, This indicates the degree of internal fragmentation within the opposition camp at the current moment. This represents the number of sub-viewpoints in the opposing camp at the current moment. A logarithmic transformation is used here to reflect the non-linear impact of the increase in the number of sub-viewpoints on the camp's stability: an increase in the number of smaller sub-viewpoints significantly increases internal divisions, while a continued increase in the number of larger sub-viewpoints gradually weakens the effect.

[0146] Next, the overall fragmentation level is determined based on the internal fragmentation level of each faction and the number of people in each faction.

[0147] Optionally, the formula for calculating the overall fragmentation degree is: ,in, This indicates the overall fragmentation level at the current moment.

[0148] Next, the fragmentation degree of sub-viewpoints is determined based on the overall fragmentation degree and the preset upper limit of the number of sub-viewpoints.

[0149] Optionally, the formula for calculating the fragmentation degree of sub-viewpoints is: ,in, This indicates the degree of fragmentation of the sub-viewpoint at the current moment. This represents the upper limit of the preset number of sub-viewpoints, which can be obtained based on human experience or historical observations.

[0150] In this embodiment, a weighted average of the number of supporters and opponents at the current moment can more accurately reflect the influence of the faction with the larger number of supporters on the overall public opinion. The resulting sub-viewpoint fragmentation can more accurately measure the diversity of sub-viewpoints within a faction.

[0151] In this embodiment, the position polarization index of the target topic can be determined based on the intensity of cross-faction conflict, the degree of faction balance, and the degree of fragmentation of sub-viewpoints.

[0152] In one possible implementation, the intensity of cross-factional conflict, the degree of factional balance, and the degree of fragmentation of sub-viewpoints can be weighted and summed to obtain the position polarization index of the target topic, i.e. ,in, This represents the polarization index of the target topic at the current moment. The non-negative weights representing the balance between factions The non-negative weights representing the intensity of cross-faction conflict The non-negative weights representing the degree of fragmentation of sub-viewpoints , and The specific values ​​can be set according to business priorities or historical calibration. This application does not limit the specific values.

[0153] visible, Depend on , , It is formed by merging according to a predetermined weight, and is used to uniformly depict the degree of polarization in the public opinion field. When the size of the camps is relatively balanced ( Highly competitive, strong in cross-faction combat ( (Higher risk), and accompanied by a certain risk of internal tearing ( When (not low), It will rise significantly; conversely, it will remain at a low level.

[0154] In summary, this application provides a method for calculating the position polarization index based on the intensity of cross-faction conflict, the degree of faction balance, and the degree of fragmentation of sub-viewpoints. It can perform a real-time quantitative assessment of position polarization from three complementary dimensions, and the determined position polarization index is more accurate. By calculating the position polarization index at various times, it is convenient to compare and track across periods and events.

[0155] The above describes a method for determining the field polarization index provided by the embodiments of this application. The following will describe the apparatus for performing the above method for determining the field polarization index.

[0156] Please see Figure 3 , Figure 3 This is a schematic diagram of a device for determining the field polarization index provided in an embodiment of this application. Figure 3 As shown, the apparatus for determining the position polarization index may include:

[0157] The data acquisition module 301 is used to acquire topic interaction-related data of each participating user of the target topic. The topic interaction-related data of each participating user includes at least one comment data and at least one interaction feedback data for the target topic.

[0158] The position statistics module 302 is used to determine the position of each participating user, the number of people in each position camp, and the number of sub-viewpoints in each position camp based on the comment data of each participating user.

[0159] The polarization index determination module 303 is used to determine the polarization index of the target topic based on the interactive feedback data of each participating user, the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp.

[0160] In one possible implementation, the stance statistics module, when determining the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp based on the comment data of each participating user, can specifically be used for:

[0161] Determine each participating user's stance based on their comment data;

[0162] The number of people in each faction is determined based on the stances of the participating users.

[0163] The comment data of each participating user is clustered by stance to obtain the comment data of each stance camp;

[0164] The comment data for each stance camp is clustered into sub-viewpoints based on semantic similarity to obtain sub-viewpoint clusters for each stance camp; the number of sub-viewpoint clusters obtained for each stance camp is taken as the number of sub-viewpoints for each stance camp.

[0165] In one possible implementation, the stance statistics module, when determining the stance of each participating user based on their comment data, can specifically be used for:

[0166] Each participating user's comment data is fed into a pre-configured stance classifier to obtain the stance label and confidence score corresponding to each participating user's comment data;

[0167] Based on the stance label and confidence score corresponding to the comment data of each participating user, the reliability stance corresponding to the comment data of each participating user is determined;

[0168] The stance of each participating user is determined based on the reliability stance corresponding to the comment data of each participating user.

[0169] In one possible implementation, the polarization index determination module, when determining the polarization index of a target topic based on the interactive feedback data of each participating user, the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp, can specifically be used for:

[0170] Based on the interaction feedback data of each participating user and the position of each participating user, the intensity of cross-faction conflict is determined. The intensity of cross-faction conflict is used to measure the proportion of negative feedback with conflict characteristics in cross-faction interactions.

[0171] The faction balance is determined based on the number of people in each faction. The faction balance is used to measure whether the number of people in each faction is balanced.

[0172] The degree of fragmentation of sub-viewpoints is determined based on the number of people in each faction and the number of sub-viewpoints in each faction. The degree of fragmentation of sub-viewpoints is used to measure the stability of each faction.

[0173] The polarization index of the target topic is determined based on the intensity of cross-factional conflict, the degree of factional balance, and the degree of fragmentation of sub-viewpoints.

[0174] In one possible implementation, the polarization index determination module, when determining the intensity of cross-faction conflict based on the interactive feedback data and the stances of each participating user, can specifically be used for:

[0175] Determine the weight of each interactive feedback data from each participating user;

[0176] Based on the position of each participating user and the weight of each interactive feedback data of each participating user, determine whether each interactive feedback data of each participating user is conflicting interactive data; in order to obtain all conflicting interactive data;

[0177] The weight of negative cross-faction interactions is determined based on the weight of each of the conflict interaction data.

[0178] The total weight of the interaction is determined based on the weight of each participating user's interaction feedback data.

[0179] The intensity of cross-faction conflict is determined based on the negative interaction weights and total interaction weights across factions.

[0180] In one possible implementation, the polarization index determination module, when determining whether each interactive feedback data of each participating user is conflicting based on the position of each participating user and the weight of each interactive feedback data of each participating user, can specifically be used for:

[0181] For each interactive feedback data point from each participating user:

[0182] Determine whether the stance of the parent user and the participating user corresponding to the interactive feedback data belong to different stance camps;

[0183] If so, determine whether the interaction type of the interactive feedback data belongs to the preset set of negative feedback types;

[0184] If so, then the interactive feedback data is identified as conflicting interactive data.

[0185] In one possible implementation, the polarization index determination module, when determining the balance of power based on the number of members in each faction, can specifically be used for:

[0186] Based on the number of people in each faction, determine the absolute difference in the number of people between any two factions and the total number of people;

[0187] The degree of factional balance is determined by the absolute difference in the number of people between each two factions and the total number of people.

[0188] In one possible implementation, the polarization index determination module, when determining the fragmentation degree of sub-viewpoints based on the number of members in each faction and the number of sub-viewpoints in each faction, can specifically be used for:

[0189] The degree of internal fragmentation of each faction is determined based on the number of sub-viewpoints within each faction.

[0190] The overall fragmentation level is determined based on the internal fragmentation level of each faction and the number of people in each faction.

[0191] The fragmentation degree of sub-viewpoints is determined based on the overall fragmentation degree and the preset upper limit of the number of sub-viewpoints.

[0192] Each module in the aforementioned device for determining the position polarization index can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0193] This application also provides an electronic device, which may include at least one processor and a memory connected to the processor, wherein:

[0194] Memory is used to store computer programs;

[0195] The processor is used to execute computer programs to enable the electronic device to implement any of the methods for determining the position polarization index provided in the embodiments of this application.

[0196] refer to Figure 4 The diagram illustrates a structural schematic suitable for implementing the electronic device in the embodiments of this application. The electronic device in the embodiments of this application may include, but is not limited to, fixed terminals such as mobile phones, laptops, PDAs (personal digital assistants), PADs (tablet computers), desktop computers, etc. Figure 4 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0197] like Figure 4 As shown, the electronic device may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 into a random access memory (RAM) 603. When the electronic device is powered on, the RAM 603 also stores various programs and data required for the operation of the electronic device. The processing unit 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0198] Typically, the following devices can be connected to I / O interface 605: input devices 606 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 607 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 608 including, for example, memory cards, hard drives, etc.; and communication devices 609. Communication device 609 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 4 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown. More or fewer devices may be implemented or have alternatively.

[0199] This application also provides a computer program product including computer-readable instructions, which, when executed on an electronic device, cause the electronic device to implement any of the position polarization index determination methods provided in this application.

[0200] This application also provides a computer-readable storage medium carrying one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device is able to implement any of the position polarization index determination methods provided in this application.

[0201] It should also be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. In addition, in the device embodiment drawings provided in this application, the connection relationship between modules indicates that they have a communication connection, which can be implemented as one or more communication buses or signal lines.

[0202] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware, or it can be implemented by special-purpose hardware including application-specific integrated circuits, special-purpose CPUs, special-purpose memory, special-purpose components, etc. Generally, any function performed by a computer program can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be diverse, such as analog circuits, digital circuits, or special-purpose circuits. However, for this application, software program implementation is more often the preferred implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a computer floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk, or optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, training equipment, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0203] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product.

[0204] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, training device, or data center to another website, computer, training device, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

Claims

1. A method for determining a position polarization index, characterized in that, include: Obtain topic interaction-related data for each participating user of the target topic, wherein the topic interaction-related data for each participating user includes at least one comment and at least one interaction feedback for the target topic; Based on the comment data of each participating user, determine the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp; The polarization index of the target topic is determined based on the interaction feedback data of each participating user, the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp.

2. The method for determining the position polarization index according to claim 1, characterized in that, The step of determining the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp based on the comment data of each participating user includes: The stance of each participating user is determined based on the comment data of each participating user. Based on the positions of the participating users, determine the number of people in each of the aforementioned positions / camps; The comment data of each participating user is clustered by stance to obtain the comment data of each stance camp. The comment data of each of the aforementioned positions is clustered into sub-viewpoints based on semantic similarity to obtain sub-viewpoint clusters for each of the aforementioned positions; the number of sub-viewpoint clusters for each of the aforementioned positions is taken as the number of sub-viewpoints for each of the aforementioned positions.

3. The method for determining the position polarization index according to claim 2, characterized in that, The step of determining the stance of each participating user based on their comment data includes: The comment data of each participating user is fed into a pre-configured stance classifier to obtain the stance label and confidence score corresponding to the comment data of each participating user. Based on the stance label and confidence score corresponding to the comment data of each participating user, the reliability stance corresponding to the comment data of each participating user is determined; The stance of each participating user is determined based on the reliability stance corresponding to the comment data of each participating user.

4. The method for determining the position polarization index according to any one of claims 1-3, characterized in that, The step of determining the polarization index of the target topic based on the interaction feedback data of each participating user, the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp includes: Based on the interaction feedback data of each participating user and the position of each participating user, the intensity of cross-faction conflict is determined. The intensity of cross-faction conflict is used to measure the proportion of negative feedback with conflict characteristics in cross-faction interactions. The camp balance is determined based on the number of people in each of the aforementioned camps. The camp balance is used to measure whether the number of people in each of the aforementioned camps is balanced. The degree of fragmentation of sub-viewpoints is determined based on the number of people in each of the aforementioned positions and the number of sub-viewpoints in each of the aforementioned positions. The degree of fragmentation of sub-viewpoints is used to measure the stability of each of the aforementioned positions. The position polarization index of the target topic is determined based on the intensity of cross-faction conflict, the degree of faction balance, and the degree of fragmentation of sub-viewpoints.

5. The method for determining the position polarization index according to claim 4, characterized in that, The determination of the cross-faction conflict intensity based on the interaction feedback data and the stance of each participating user includes: Determine the weight of each interactive feedback data point from each participating user; Based on the position of each participating user and the weight of each interactive feedback data of each participating user, determine whether each interactive feedback data of each participating user is conflicting interactive data; to obtain all conflicting interactive data; Based on the respective weights of all the conflict interaction data, determine the cross-faction negative interaction weights; The total weight of the interaction is determined based on the weight of each user's interaction feedback data. The intensity of the cross-faction conflict is determined based on the cross-faction negative interaction weight and the total interaction weight.

6. The method for determining the position polarization index according to claim 5, characterized in that, The step of determining whether each piece of interactive feedback data from each participating user is conflicting interactive data based on the position of each participating user and the weight of each piece of interactive feedback data from each participating user includes: For each interaction feedback data of each participating user: Determine whether the stance of the parent user and the participating user corresponding to the interactive feedback data belong to different stance camps; If so, determine whether the interaction type of the interactive feedback data belongs to the preset set of negative feedback types; If so, then the interactive feedback data is identified as conflicting interactive data.

7. The method for determining the position polarization index according to claim 4, characterized in that, The determination of factional balance based on the number of members in each of the aforementioned factions includes: Based on the number of people in each of the stated positions, determine the absolute difference in the number of people between any two stated positions and the total number of people; The balance of the factions is determined by the absolute difference in the number of people between each pair of the stated factions and the total number of people.

8. The method for determining the position polarization index according to claim 4, characterized in that, The determination of sub-viewpoint fragmentation based on the number of members in each of the aforementioned positions and the number of sub-viewpoints in each of the aforementioned positions includes: The internal fragmentation degree of each of the aforementioned positions is determined based on the number of sub-viewpoints in each position camp. The overall fragmentation level is determined based on the internal fragmentation level of each stated faction and the number of people in each stated faction. The fragmentation degree of a sub-viewpoint is determined based on the overall fragmentation degree and the preset upper limit of the number of sub-viewpoints.

9. A device for determining the field polarization index, characterized in that, include: The data acquisition module is used to acquire topic interaction-related data of each participating user of the target topic. The topic interaction-related data of each participating user includes at least one comment data and at least one interaction feedback data for the target topic. The stance statistics module is used to determine the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp based on the comment data of each participating user. The polarization index determination module is used to determine the polarization index of the target topic based on the interaction feedback data of each participating user, the stance of each participating user, the number of people in each stance camp, and the number of sub-viewpoints in each stance camp.

10. An electronic device, characterized in that, It includes at least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs; The processor is configured to execute the computer program to enable the electronic device to implement the method for determining the position polarization index as described in any one of claims 1 to 8.

Images