Method for determining opinion leader

By obtaining online evaluation sets, determining the optimal number of topics and identifying topic leaders, and combining user activity information, the problem of difficult to identify opinion leaders in the existing technology is solved, and fast and accurate opinion leader identification and future prediction are achieved, and public sentiment is enhanced Monitoring accuracy and predictive capability.

CN119003766AActive Publication Date: 2024-11-22QINGDAO UNIV OF TECH +1
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202410989512.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-23
Publication Date
2024-11-22
Estimated Expiration
2044-07-23

AI Technical Summary

Technical Problem

It is difficult for prior art to quickly and accurately identify opinion leaders in cyber events, especially when there are differences in opinions among a large number of user messages.

Method used

By obtaining the network evaluation set, determining the optimal number of topics, dividing them into topic sets, and using text similarity and emotional similarity matrix to identify topic leaders groups and opinion leaders, and predicting future opinion leaders based on user activity information.

Benefits of technology

It has achieved rapid and accurate identification of opinion leaders in emergencies, improved the accuracy and prediction capabilities of public opinion monitoring, and can promptly detect and supervise before public opinion breaks out.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119003766A_ABST
    Figure CN119003766A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of public opinion monitoring, and particularly relates to an opinion leader determination method. The invention provides an opinion leader determination method. Obtaining a network evaluation set of the target event, wherein the network evaluation set comprises a plurality of network comment texts; determining an optimal theme number of the target event according to the network evaluation set; dividing the network evaluation set into a plurality of topic sets according to the optimal topic number; and determining a current opinion leader of each theme set according to each network comment text in the theme set. Through analysis of a large amount of message texts, opinion leaders can be rapidly and accurately discriminated.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of public opinion monitoring, and specifically relates to a method for determining an opinion leader. Background Art

[0002] With the help of modern information technology, the Internet has become the main battlefield for netizens to express their views and opinions. Based on the current situation, the influence of opinion leaders is becoming increasingly prominent. Opinion leaders refer to people who are active in interpersonal networks and have high representativeness and prestige. The influence of opinion leaders can influence social public opinion and public opinion in a more direct and rapid way through various communication channels. Therefore, it is very necessary to accurately identify opinion leaders in sudden public opinion events. Because there are a large number of user messages in network events, and there are differences in the views of different users, it is difficult for existing technologies to quickly and accurately find opinion leaders among many users. Summary of the invention

[0003] In view of the above technical problems, the present invention proposes a method for determining opinion leaders. This application provides a method for determining opinion leaders, including: obtaining a network evaluation set of a target event, wherein the network evaluation set includes multiple network comment texts; determining the optimal number of topics for the target event based on the network evaluation set; dividing the network evaluation set into multiple topic sets based on the optimal number of topics; and determining the current opinion leader of each topic set based on each network comment text in the topic set. By analyzing a large number of message texts, opinion leaders can be quickly and accurately identified.

[0004] In a first aspect, the present application provides a method for determining an opinion leader, comprising: obtaining a network evaluation set of a target event, wherein the network evaluation set includes multiple network comment texts; determining an optimal number of topics for the target event based on the network evaluation set; dividing the network evaluation set into multiple topic sets based on the optimal number of topics; and determining a current opinion leader for each of the topic sets based on each of the network comment texts in the topic set.

[0005] In some embodiments, determining the number of topics for the target event based on the network evaluation set includes: determining a perplexity index under different numbers of topics based on the network evaluation set; and determining the optimal number of topics based on the perplexity index.

[0006] In some embodiments, dividing the network evaluation set into multiple topic sets according to the optimal number of topics includes: using an LDA model to determine the topic probability that each network review text in the network evaluation set belongs to each topic according to the optimal number of subjects; determining the topic to which each network review text belongs according to the topic probability; and dividing the network evaluation set into multiple topic sets according to the topic of each network review text.

[0007] In some embodiments, determining the current opinion leader of each topic set based on the topic set includes: performing the following operations for any topic set: determining a text similarity matrix between each first comment text in the topic set based on the first comment text in the topic set; determining a topic leader group based on the text similarity matrix; determining a sentiment similarity matrix between each second comment text corresponding to the topic leader group based on the second comment text; and determining the current opinion leader in the topic leader group based on the sentiment similarity matrix.

[0008] In some embodiments, determining the topic leader group based on the text similarity matrix includes: obtaining a preset first percentage threshold and a first PageRank value threshold; converting the text similarity matrix into a one-dimensional text similarity vector, wherein the one-dimensional text similarity vector includes multiple text similarity factors; arranging the text similarity factors from large to small to form a text similarity factor sequence; determining a text similarity factor threshold based on the text similarity factor sequence and the first percentage threshold; determining text similarity factors in the one-dimensional text similarity vector that are greater than or equal to the text similarity factor threshold as important text factors; establishing a text undirected graph based on the important text factors; calculating the text PageRank value of each node in the text undirected graph; and determining the topic leader group of the topic based on the text PageRank value of each node and the first PageRank value threshold.

[0009] In some embodiments, determining the current opinion leader in the topic leader group according to the emotion similarity matrix includes: obtaining a preset second percentage threshold and a second PageRank value threshold; converting the emotion similarity matrix into a one-dimensional emotion similarity vector, wherein the one-dimensional emotion similarity vector includes a plurality of emotion similarity factors; arranging the emotion similarity factors from large to small to form an emotion similarity factor sequence; determining an emotion similarity factor threshold according to the emotion similarity factor sequence and the second percentage threshold; determining an emotion similarity factor in the one-dimensional emotion similarity vector that is greater than or equal to the emotion similarity factor threshold as an important emotion factor; establishing an emotion undirected graph according to the important emotion factors; calculating the emotion PageRank value of each node of the emotion undirected graph; and determining the current opinion leader of the topic according to the emotion PageRank value and the second PageRank value threshold of each node.

[0010] In some embodiments, determining the text similarity matrix between each first comment text in the topic set based on the first comment text in the topic set includes: calculating the word frequency of each word in each of the first comment texts; calculating the inverse document frequency of each word in the topic set; determining the weight of each word in the first comment text based on the word frequency and the inverse document frequency; determining the text similarity between each first comment text based on the weight of each word, thereby obtaining the text similarity matrix.

[0011] In some embodiments, determining the sentiment similarity matrix between each second comment text based on the second comment text corresponding to the topic leader group includes: obtaining a preset sentiment dictionary; determining the sentiment words in each second comment text and the sentiment polarity corresponding to the sentiment words based on the sentiment dictionary; determining the sentiment tendency score of each second comment text based on the sentiment words and the sentiment polarity; determining the sentiment similarity between each second comment text based on the sentiment tendency score, and thereby obtaining the sentiment similarity matrix.

[0012] In some embodiments, the method further includes: obtaining user activity information of each of the current opinion leaders; determining a leader feature importance sequence based on the user activity information; determining a weight of each feature in the user activity information based on the leader feature importance sequence; and predicting future opinion leaders based on the weight of each feature.

[0013] In some embodiments, predicting future opinion leaders based on the weights of various features includes: obtaining target activity information of a target user; determining a leader score of the target user based on the target activity information and the weights of various features; and predicting whether the target user can become an opinion leader in the future based on the leader score.

[0014] Beneficial effects of the invention: The present application provides a method for determining an opinion leader, comprising: obtaining a network evaluation set of a target event, wherein the network evaluation set includes a plurality of network comment texts; determining the optimal number of topics for the target event according to the network evaluation set; dividing the network evaluation set into a plurality of topic sets according to the optimal number of topics; and determining the current opinion leader of each of the topic sets according to each of the network comment texts in the topic set. By analyzing a large number of message texts, the opinion leader can be identified quickly and accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The scope of the present disclosure may be better understood by reading the following detailed description of exemplary embodiments in conjunction with the accompanying drawings. The drawings included are:

[0016] Figure 1 An overall flow chart of a method for determining an opinion leader provided in an embodiment of the present application;

[0017] Figure 2 An overall flow chart of a method for predicting opinion leaders provided in an embodiment of the present application. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0019] In the following description, reference is made to “some embodiments”, which describe a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0020] If similar descriptions of "first\second\third" appear in the application documents, the following instructions will be added. In the following description, the terms "first\second\third" involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It can be understood that "first\second\third" can be interchanged in a specific order or sequence where permitted, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0022] With the help of modern information technology, the Internet has become the main battlefield for netizens to express their views and opinions. Based on the current situation, the influence of opinion leaders is becoming increasingly prominent. Opinion leaders refer to people who are active in interpersonal networks and have high representativeness and prestige. The influence of opinion leaders can influence social public opinion and public opinion in a more direct and rapid way through various communication channels. Therefore, it is very necessary to accurately identify opinion leaders in sudden public opinion events. Because there are a large number of user messages in network events, and there are differences in the views of different users, it is difficult for existing technologies to quickly and accurately find opinion leaders among many users.

[0023] In view of the problems existing in the prior art, such as Figure 1 As shown, the present application provides a method for determining an opinion leader, and the method is applied to an electronic device, and the electronic device can be a server, a mobile terminal, a computer, a cloud platform, etc. The function implemented by the device data processing provided in the embodiment of the present application can be implemented by calling a program code by a processor of the electronic device, wherein the program code can be stored in a computer storage medium, and the method for determining an opinion leader includes:

[0024] Step S1: obtaining a network evaluation set of a target event, wherein the network evaluation set includes a plurality of network comment texts.

[0025] When the public opinion on a public event begins to ferment and heat up, the web crawler software python is used to collect the microblog posts, comments and other comment text content and related data of the users of the topic to be monitored from the microblog platform in real time. The Chinese word segmentation tool jieba is used to process the comment text, and the meaningless words in the text are removed according to the stop word list, and irrelevant content is filtered out, thereby obtaining the various network comment texts in the network evaluation data set.

[0026] Step S2: Determine the optimal number of topics for the target event based on the network evaluation set.

[0027] In some embodiments, step S2 of "determining the optimal number of topics for the target event based on the network evaluation set" includes:

[0028] Step S21: determining the perplexity index under different numbers of topics according to the network evaluation set.

[0029] Step S22: Determine the optimal number of topics according to the perplexity index.

[0030] Step S3: Divide the network evaluation set into multiple topic sets according to the optimal number of topics.

[0031] In some embodiments, step S3 "dividing the network evaluation set into multiple topic sets according to the optimal number of topics" includes:

[0032] Step S31: using the LDA model to determine the topic probability of each of the online review texts in the online evaluation set belonging to each topic according to the optimal number of subjects.

[0033] Step S32: Determine the topic to which each online comment text belongs according to the topic probability.

[0034] Step S33: dividing the network review set into a plurality of theme sets according to the theme of each network review text.

[0035] In the process of public opinion diffusion, all kinds of true and false opinions emerge in an endless stream, and facts and rumors are dazzling. Previous methods for identifying opinion leaders did not classify and select them, and the identification results sometimes did not hit the mark, but were a mixture of positive and negative remarks. Due to the endless stream of opinions in various comments, the discussion of an event can be divided into multiple opinions, which are the topics mentioned in this application. However, sometimes the topics of some comments are difficult to clarify, or new topics will be generated, but not all topics are worth detecting.

[0036] Therefore, in this application, before classifying each comment text by topic, it is necessary to determine the optimal number of topics. Perplexity is an indicator to measure the ability of the model to predict new documents. The lower the perplexity, the better the prediction ability of the model. Therefore, in this application, the optimal number of topics is determined by the perplexity indicator. When the perplexity decreases gradually or is at an inflection point, the corresponding number of topics is the optimal number of topics.

[0037] When determining the optimal number of topics, first use different numbers of topics to train the LDA model, then use the network evaluation set to calculate the perplexity of the LDA model with different numbers of topics, and establish a relationship table between perplexity and model number. In the relationship table, find the point where the rate of perplexity decreases slowly or the turning point. At this time, the number of topics corresponding to such a point is the optimal number of topics. Since the interpretability of each topic also needs to be considered, the maximum number of topics cannot be selected as the optimal number of topics, which will cause the topics to be overly subdivided and difficult to explain each topic.

[0038] After determining the optimal number of topics, the LDA model can be used to classify the online review texts in the online evaluation set according to the optimal number of topics, so as to form data sets with multiple different topics, namely, topic sets.

[0039] The LDA model is a method for clustering and extracting topic information from text data, and presents the article topic in the form of probability distribution. It can quickly identify topic types from large-scale text data, and thus becomes an important basis for determining the similarity of text topics.

[0040] Therefore, in this application, after LDA is used to cluster topics and obtain the topic probability of the text, the cosine similarity method is used to calculate the similarity between topics to verify whether the number of topics and the division between each topic are reasonable.

[0041] Step S4: determining the current opinion leader of each of the topic sets according to each of the network comment texts in the topic set.

[0042] In the discussion of events, each topic will have its corresponding topic leaders and opinion leaders, so for each topic set, you can follow the steps below to determine the current opinion leaders corresponding to each topic.

[0043] In the past, the diffusion of public opinion was usually studied from the perspective of social networks, with the interaction between users as a reference. However, the mutual influence between comments is also important in the process of information dissemination and opinion formation. Therefore, this application realizes the identification of opinion leaders by calculating text similarity and sentiment similarity, which expands the traditional method of evaluating influence or identifying user activity based on interactive behavior characteristics.

[0044] In some embodiments, step S4 of "determining the current opinion leaders of each of the topic sets according to each of the online comment texts in the topic set" includes:

[0045] Step S41: determining a text similarity matrix between each first comment text in the topic set according to the first comment text in the topic set.

[0046] In some embodiments, step S41 "determining a text similarity matrix between first comment texts in the topic set according to the first comment text in the topic set" includes:

[0047] Step S411: Calculate the frequency of each word in each of the first comment texts.

[0048] Step S412: Calculate the inverse document frequency of each word in the topic set.

[0049] Step S413: Determine the weight of each word in the first comment text according to the word frequency and the inverse document frequency.

[0050] Step S414: Determine the text similarity between each of the first comment texts according to the weight of each word, and then obtain the text similarity matrix.

[0051] Although this application divides multiple network rating texts into multiple topic sets based on clustering, everyone has their own ideas, so it is difficult to ensure that the content expressed by each first comment text in each topic set is completely consistent. It can only be said that each first comment text is consistent to a certain extent. In addition, because of the mutual influence between comments, if the similarity between some comments is high enough, it means that there are relatively consistent ideas between these comments, and then the first comment text with a wider influence can be determined through text similarity, and then the topic leader can be determined.

[0052] Since text similarity is the relationship between two texts, in this application, the text similarity of each first comment file in a subject set is expressed in the form of a matrix.

[0053] The text similarity between two comments is mainly determined by the expression of various words. Therefore, when calculating the text similarity between two first comment texts, the weight of each word in each first comment text can be determined first, and then the text similarity can be calculated based on the weight of each word.

[0054] Therefore, in this application, TF-IDF can be used to measure the weight of each word in each first comment text, and then the cosine similarity between the weights of each word in the two first comment texts can be calculated to obtain the text similarity between the two first comment texts.

[0055] Among them, TF-IDF is a method used to measure the importance of words in a text. Its basic principle is to assign a weight to each word in the text and consider the frequency of the word in the entire text collection. TF (term frequency) represents the frequency of occurrence of a word in a text, that is: IDF (Inverse Document Frequency) reflects the importance of a word in the entire text collection, namely:

[0056]

[0057] TF-IDF is the product of TF and IDF, which is used to measure the importance of a word in a text and takes into account the frequency in the entire text collection: TF-IDF(t,d,D)=TF(t,d)×IDF(t,D).

[0058] Once the TF-IDF weights of each word in each first comment text are calculated, these weights can be used to calculate the similarity between texts. A common method is cosine similarity:

[0059]

[0060] Step S42: determining a topic leader group according to the text similarity matrix.

[0061] In some embodiments, step S42 "determining a topic leader group according to the text similarity matrix" includes:

[0062] Step S421: Obtain a preset first percentage threshold and a first PageRank value threshold.

[0063] Step S422: converting the text similarity matrix into a one-dimensional text similarity vector, wherein the one-dimensional text similarity vector includes a plurality of text similarity factors.

[0064] Step S423: Arrange the text similarity factors from large to small to form a text similarity factor sequence.

[0065] Step S424: determining a text similarity factor threshold according to the text similarity factor sequence and the first percentage threshold.

[0066] Step S425: Determine the text similarity factors in the one-dimensional text similarity vector that are greater than or equal to the text similarity factor threshold as important text factors.

[0067] Step S426: Establishing a text undirected graph according to the important text factors.

[0068] Step S427: Calculate the text PageRank value of each node in the text undirected graph.

[0069] Step S428: Determine the topic leader group of the topic according to the text PageRank value of each node and the first PageRank value threshold.

[0070] Text similarity can express the similarity relationship between each first comment text, but it is not possible to directly find the topic leader among many first comment texts through the similarity relationship. Therefore, in this application, the text similarity matrix needs to be processed.

[0071] Since the text similarity matrix can only reflect the relationship between the first comment texts of each pair, and cannot reflect the relationship between the entire group, this application converts the text similarity matrix into a one-dimensional vector. In order to distinguish it from the calculation of the current opinion leader below, the one-dimensional vector is called a one-dimensional text similarity vector. The one-dimensional vector is composed of multiple factors. Similarly, for the purpose of distinction, each factor in the one-dimensional vector is called a text similarity factor.

[0072] Because the higher the similarity, the more similar the ideas between the texts, it can also be considered that the mutual influence between the two texts is more serious. Therefore, in this application, each first comment text is filtered by a first percentage threshold, and the text with higher text similarity is retained. The embodiment in the one-dimensional vector is to retain the text similarity factor with a larger value, that is, the important text factor. Since there is a corresponding relationship between the similarity factor and the text similarity and the first comment text, the text undirected graph can also be constructed by the important text factor, wherein each node in the undirected graph of this article is each first comment text, and the edge of the undirected graph represents the text similarity between the two first comment texts, wherein only the text similarity corresponding to the important text factor can be called an edge. After obtaining the undirected graph, the text PageRank value of each node of the text undirected graph is calculated according to the text undirected graph. The influence of each first comment text in the entire topic is determined by the text PageRank value, and the first comment text whose influence or text PageRank value meets the threshold is determined as the topic leader. Since there is more than one text with high influence in a topic, and more than one user, there is more than one topic leader, which constitutes a group of topic leaders.

[0073] Step S43: determining a sentiment similarity matrix between each of the second comment texts according to the second comment texts corresponding to the topic leader group.

[0074] In some embodiments, step S43 "determining the sentiment similarity matrix between each second comment text according to the second comment text corresponding to the topic leader group" includes:

[0075] Step S431: Obtain a preset sentiment dictionary.

[0076] Step S432: determining the sentiment words in each of the second comment texts and the sentiment polarities corresponding to the sentiment words according to the sentiment dictionary.

[0077] Step S433: determining the sentiment tendency score of each second comment text according to the sentiment words and the sentiment polarity.

[0078] Step S434: determining the sentiment similarity between each of the second comment texts according to the sentiment tendency score, and then obtaining the sentiment similarity matrix.

[0079] Emotional similarity can indicate the position between each comment text. If you want to find the current opinion leader, you need to analyze the emotional similarity between each second comment text in the topic leader group and clarify the emotional impact between each second comment text. Since each comment contains multiple emotional words, these emotional words have positive emotions and negative emotions. Therefore, before determining the emotional similarity between each second comment text in this application, it is necessary to first obtain a preset emotional dictionary, which is determined based on the emotional word classification generally recognized by group owners such as Li Jun Chinese Praise and Derogatory Dictionary of Tsinghua University, NTUSD, Hownet Emotional Dictionary, etc. Then, text processing is performed on each second comment text, including steps such as word segmentation and removal of stop words. The second comment text after text processing is matched with the preset emotional dictionary, and the various emotional words in each second comment text and the emotional tendency of each emotional word are counted. Then, the emotional tendency score of each second comment text is determined based on the emotional words and emotional tendency, and finally the similarity of the emotional tendency scores of the two second comment texts is calculated by cosine similarity, so as to obtain the emotional similarity matrix between the two second comment texts.

[0080] Among them, any second comment text Different sentiment words in the sentiment dictionary are assigned different sentiment scores.

[0081] Step S44: determining a current opinion leader in the topic leader group according to the sentiment similarity matrix.

[0082] In some embodiments, step S44 “determining the current opinion leader in the topic leader group according to the sentiment similarity matrix” includes:

[0083] Step S441: Obtain a preset second percentage threshold and a second PageRank value threshold.

[0084] Step S442: converting the emotion similarity matrix into a one-dimensional emotion similarity vector, wherein the one-dimensional emotion similarity vector includes a plurality of emotion similarity factors.

[0085] Step S443: Arrange the emotion similarity factors from large to small to form an emotion similarity factor sequence.

[0086] Step S444: determining an emotion similarity factor threshold according to the emotion similarity factor sequence and the second percentage threshold.

[0087] Step S445: Determine the emotion similarity factors in the one-dimensional emotion similarity vector that are greater than or equal to the emotion similarity factor threshold as important emotion factors.

[0088] Step S446: Establish an undirected emotional graph based on the important emotional factors.

[0089] Step S447: Calculate the sentiment PageRank value of each node in the sentiment undirected graph.

[0090] Step S448: Determine the current opinion leader of the topic according to the sentiment PageRank value and the second PageRank value threshold of each node.

[0091] Sentiment similarity can express the similarity relationship between the second comment texts, but it is not possible to directly find opinion leaders among the many second comment texts through the similarity relationship. Therefore, in this application, the sentiment similarity matrix needs to be processed.

[0092] Since the sentiment similarity matrix can only reflect the relationship between two comment texts, and cannot reflect the relationship between the entire group, this application first uses cosine similarity to convert the sentiment similarity matrix into a one-dimensional vector, which is referred to as a one-dimensional sentiment similarity vector. The one-dimensional vector is composed of multiple factors. Similarly, for the purpose of distinction, each factor in the one-dimensional vector is referred to as a sentiment similarity factor.

[0093] Because the higher the similarity, the more similar the emotional tendency between the texts, it can also be considered that the mutual emotional influence between the two texts is more serious. Therefore, in this application, each second comment text is filtered by a second percentage threshold, and the texts with higher emotional similarity are retained. The embodiment in the one-dimensional vector is to retain the emotional similarity factor with a larger value, that is, the important emotional factor. Since there is a corresponding relationship between the similarity factor and the emotional similarity and the second comment text, an emotional undirected graph can also be constructed by important emotional factors, wherein each node in the undirected graph of this article is each second comment text, and the edge of the undirected graph represents the emotional similarity between the two second comment texts, wherein only the emotional similarity corresponding to the important emotional factor can be called an edge. After obtaining the undirected graph, the emotional PageRank value of each node of the emotional undirected graph is calculated according to the emotional undirected graph. The influence of each second comment text in the entire topic is determined by the emotional PageRank value, and the second comment text whose influence or emotional PageRank value meets the threshold is determined as an opinion leader.

[0094] In addition, since the existing technologies mostly start from the characteristics of Weibo users themselves, such as the number of fans, reposts, likes, comments, and the number of followers, they ignore the importance of the text features of the post or comment itself, such as text length, posting and comment time, etc.; at the same time, since the diffusion of public opinion presents certain spatial characteristics, the IP of Weibo users cannot be ignored in the research, which is often overlooked in previous studies; in addition, in the recognition algorithm of opinion leaders, different weights are not given to different features, resulting in the current opinion leaders determined are not very accurate, and it is impossible to predict future opinion leaders.

[0095] In response to the above-mentioned problems, the present application can better determine the important feature sequence of becoming an opinion leader by using the current opinion leaders determined in the above steps, and predict the future opinion leaders of the event based on the important feature sequence, and then focus on verifying their published content, so as to nip it in the bud and conduct timely detection and supervision before a large-scale outbreak of public opinion; in addition, the news officials can imitate the important feature sequences of opinion leaders, compete with the opinion leaders who spread false news in terms of traffic, and give full play to the role of official opinion leaders.

[0096] Therefore, in some embodiments, the method further comprises:

[0097] Step S5: Obtain user activity information of each current opinion leader.

[0098] Step S6: Determine the leader feature importance sequence according to the user activity information.

[0099] Step S7: Determine the weight of each feature in the user activity information according to the leader feature importance sequence.

[0100] Step S8: predicting future opinion leaders based on the weights of various features.

[0101] In some embodiments, step S8 “predicting future opinion leaders based on weights of various features” includes:

[0102] Step S81: Obtain target activity information of the target user.

[0103] Step S82: Determine the leader score of the target user according to the target activity information and the weights of various features.

[0104] Step S83: predicting whether the target user can become an opinion leader in the future according to the leader score.

[0105] Since the second comment text corresponding to the current opinion leader has been determined, we can determine the user activity information of the current opinion leader based on the second comment text. The user activity information includes: the user's IP address, the time the comment was published, the length of the text, and the number of the user's fans, the degree of attention, and the degree of attention paid to others. These attribute data of all opinion leaders are combined to form a data set to be analyzed, and the trained neural network model for feature extraction is used to determine the important features from these data sets to be analyzed, and at the same time determine the importance of these important features in the current opinion leader sample group, thereby obtaining a feature importance sequence.

[0106] After obtaining the feature importance sequence, we can determine the weight of each important feature based on the feature importance feature sequence, clarify the important features of opinion leaders, and the weight of each important feature. At this time, we can use the weights of these important features to predict the future opinion leaders of existing users under this event, and at the same time, we can also allow the official to borrow these important features for imitation, so that the official account can become a future opinion leader and compete with the opinion leaders who publish false information.

[0107] When predicting future opinion leaders, the target activity information of the target user is obtained, and the leader score of the target user is calculated according to the attributes in the target activity information and the weights of various features. Then, a preset score threshold is obtained, and when the leader score of the target user is greater than or equal to the score threshold, it is determined that the target user is likely to become an opinion leader in the future.

[0108] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0109] It should be understood that "one embodiment" or "an embodiment" mentioned throughout the specification means that specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present application, the size of the sequence number of the above-mentioned processes does not mean the order of execution, and the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application. The above-mentioned sequence numbers of the embodiments of the present application are only for description and do not represent the advantages and disadvantages of the embodiments.

[0110] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0111] Those skilled in the art can understand that all or part of the steps of the above method embodiments can be completed by hardware related to program instructions, and the above program can be stored in a computer-readable storage medium, and when the program is executed, the steps of the above method embodiments are executed. The above storage medium includes: a mobile storage device, a read-only memory (ROM), a magnetic disk or an optical disk, and other media that can store program codes.

[0112] Alternatively, if the above-mentioned integrated unit of the present application is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application can essentially or in other words, the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for a controller to execute all or part of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0113] The above is only an implementation method of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for identifying opinion leaders. It is characterized in that include; Acquire a network evaluation set of a target event, wherein the network evaluation set includes a plurality of network comment texts; Determining the optimal number of topics for the target event based on the network evaluation set; Dividing the network evaluation set into a plurality of topic sets according to the optimal number of topics; The current opinion leaders of each of the topic sets are determined according to each of the network comment texts in the topic set.

2. The method according to claim 1, It is characterized in that Determining the number of topics of the target event according to the network evaluation set includes: Determining perplexity indicators for different numbers of topics based on the network evaluation set; The optimal number of topics is determined according to the perplexity indicator.

3. The method according to claim 1, It is characterized in that The network evaluation set is divided into a plurality of topic sets according to the optimal number of topics, including: Determine the topic probability of each of the network review texts in the network evaluation set belonging to each topic according to the optimal number of subjects using the LDA model; Determine the topic to which each online comment text belongs according to the topic probability; The network review set is divided into a plurality of theme sets according to the themes to which each network review text belongs.

4. The method according to claim 1, It is characterized in that Determining the current opinion leaders of each of the topic sets according to the topic sets includes: For each theme set, do the following: Determine a text similarity matrix between first comment texts in the topic set according to the first comment text in the topic set; Determine a topic leader group according to the text similarity matrix; Determine a sentiment similarity matrix between each of the second comment texts according to the second comment texts corresponding to the topic leader group; A current opinion leader is determined in the topic leader group according to the sentiment similarity matrix.

5. The method according to claim 4, It is characterized in that Determining a topic leader group according to the text similarity matrix includes: Obtaining a preset first percentage threshold and a first PageRank value threshold; Converting the text similarity matrix into a one-dimensional text similarity vector, wherein the one-dimensional text similarity vector includes a plurality of text similarity factors; Arrange the text similarity factors from large to small to form a text similarity factor sequence; Determining a text similarity factor threshold according to the text similarity factor sequence and the first percentage threshold; Determine the text similarity factor in the one-dimensional text similarity vector that is greater than or equal to the text similarity factor threshold as an important text factor; Establishing a text undirected graph according to the important text factors; Calculating the text PageRank value of each node in the text undirected graph; The topic leader group of the topic is determined according to the text PageRank value of each node and the first PageRank value threshold.

6. The method according to claim 4, It is characterized in that Determining the current opinion leader in the topic leader group according to the emotion similarity matrix includes: Obtain a preset second percentage threshold and a second PageRank value threshold; Converting the emotion similarity matrix into a one-dimensional emotion similarity vector, wherein the one-dimensional emotion similarity vector includes a plurality of emotion similarity factors; Arrange the emotion similarity factors from large to small to form an emotion similarity factor sequence; Determine an emotion similarity factor threshold according to the emotion similarity factor sequence and the second percentage threshold; Determine the emotion similarity factor in the one-dimensional emotion similarity vector that is greater than or equal to the emotion similarity factor threshold as an important emotion factor; Establishing an undirected emotional graph according to the important emotional factors; Calculating the sentiment PageRank value of each node of the sentiment undirected graph; The current opinion leader of the topic is determined according to the sentiment PageRank value of each node and the second PageRank value threshold.

7. The method according to claim 4, It is characterized in that The determining of a text similarity matrix between first comment texts in the topic set according to the first comment text in the topic set comprises: Calculate the frequency of each word in each of the first comment texts; Calculate the inverse document frequency of each word in the topic set; Determine the weight of each word in the first review text according to the word frequency and the inverse document frequency; The text similarity between each of the first comment texts is determined according to the weight of each word, thereby obtaining the text similarity matrix.

8. The method according to claim 4, It is characterized in that The step of determining the sentiment similarity matrix between the second comment texts according to the second comment texts corresponding to the topic leader groups includes: Get the preset sentiment dictionary; Determine the sentiment words in each of the second comment texts and the sentiment polarity corresponding to the sentiment words according to the sentiment dictionary; Determine the sentiment tendency score of each second comment text according to the sentiment words and the sentiment polarity; The sentiment similarity between each of the second comment texts is determined according to the sentiment tendency score, and then the sentiment similarity matrix is ​​obtained.

9. The method according to any one of claims 1 to 9, It is characterized in that The method further comprises: Obtaining user activity information of each current opinion leader; Determining a leader feature importance sequence based on the user activity information; Determining the weight of each feature in the user activity information according to the leader feature importance sequence; Predict future opinion leaders based on the weights of each feature.

10. The method according to claim 9, It is characterized in that The prediction of future opinion leaders based on the weights of various features includes: Obtain target activity information of target users; Determine the leader score of the target user according to the target activity information and the weights of various features; Whether the target user can become an opinion leader in the future is predicted based on the leader score.

Citation Information

Patent Citations

  • Internet forum-oriented opinion leader mining method

    CN102890702A

  • Network opinion leader identification method and device

    CN111177526A

  • Method and device for identifying opinion leader

    CN111460317A

  • Social network opinion leader mining method based on whole and part of graph structure

    CN114492455A

  • Hot event venation generation method and device based on incremental clustering and theme division

    CN116992029A