Group speech viewpoint tracking method and device, computer equipment and storage medium

Abstract

The invention relates to a group speech viewpoint tracking method and device, computer equipment and a storage medium. The method comprises the following steps: acquiring comment data about a target object of each user in a group at a sampling time point and character attribute information of each user from a social platform; performing part-of-speech tagging on the comment data, and constructing and analyzing a dependency syntax tree for a part-of-speech tagging result to obtain viewpoint features; performing weighted summation according to the viewpoint polarity score and the polarity weight value corresponding to each viewpoint feature to obtain a personal speech viewpoint polarity score, and determining a personal speech viewpoint according to the personal speech viewpoint polarity score; calculating a viewpoint weight score according to the personal speech viewpoint polarity score of each user and the user weight value, and determining a group speech viewpoint according to the viewpoint weight score; and the group speech viewpoints corresponding to different sampling time points are connected in series according to a time sequence to obtain group speech viewpoint change veins. By adopting the method, the opinion opinions of the group can be dynamically tracked, and efficient and accurate support is provided for public opinion analysis.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, apparatus, computer equipment and storage medium for tracking group speech and opinions. Background Art

[0002] With the development of internet technology, the public has become accustomed to using the internet to disseminate biased opinions and views on certain hot-button issues in real life. Furthermore, due to the internet's divergent and informal nature, more and more people are willing to express their views online. Given the widespread dissemination of online information, it is necessary to accurately grasp the views of the public in real time to prevent unfavorable views from having a significant impact on society.

[0003] In order to grasp the changes in group opinions, the existing technology is to regard the opinions of the user who represents the group as the group's opinions, and regard the changes in the user's opinions as the changes in the group's opinions.

[0004] However, in situations where the group representative is unclear or there are multiple group representatives, the above method cannot accurately grasp the changes in the group's speech and opinions. Summary of the Invention

[0005] Based on this, it is necessary to provide a method, device, computer equipment and storage medium for tracking group speech and opinions to address the above technical issues.

[0006] A method for tracking group speech opinions, the method comprising:

[0007] Obtain the comment data of each user in the group about the target object and the character attribute information of each user at the sampling time point from the social platform;

[0008] Perform part-of-speech tagging on the comment data to obtain multiple segments. Build a dependency syntax tree based on the dependency relationships between the segments. Parse the dependency syntax tree to obtain the opinion features in the comment data.

[0009] Obtain a corresponding polarity weight value based on the frequency of occurrence of the opinion feature, perform a weighted summation of the opinion polarity score corresponding to each opinion feature in the feature score table and the polarity weight value to obtain a personal speech opinion polarity score, and determine the personal speech opinion based on the opinion feature corresponding to the personal speech opinion polarity score in the feature score table;

[0010] Calculate the corresponding user weight value based on the character attribute information of each user, calculate the opinion weight score based on the polarity score of each user's personal opinion and the user weight value, and determine the group opinion based on the opinion feature corresponding to the opinion weight score in the feature score table;

[0011] The group opinions and views corresponding to different sampling time points are connected in chronological order to obtain a change context of the group opinions and views; the change context of the group opinions and views is used for tracking the group opinions and views.

[0012] In one embodiment, the method further includes: obtaining the group speech opinions at the i-th sampling time point, determining whether the group speech opinions at the i-th sampling time point are the same as the group speech opinions at the i-1-th sampling time point; if they are the same, deleting the group speech opinions at the i-th sampling time point; if they are not the same, retaining the group speech opinions at the i-th sampling time point, wherein i is the serial number of the sampling time point, i is less than or equal to N, and N is the number of sampling time points; iteratively updating the value of i, obtaining the group speech opinions at the next sampling time point, until i is greater than N, stopping the iteration, and connecting the current group speech opinions in chronological order to obtain a change context of the group speech opinions; the change context of the group speech opinions is used for tracking the group speech opinions.

[0013] In one embodiment, it also includes: judging whether the group speech opinions at the i-th sampling time point are the same as those at the i-1-th sampling time point, including: calculating the similarity of the group speech opinions at the i-th time point and the i-1-th time point; if the similarity is greater than a first threshold, the group speech opinions at the i-th time point and the i-1-th time point are the same; otherwise, they are not the same.

[0014] In one embodiment, the method further includes: determining whether the group's opinion on the target object at the current sampling time point is the same as the group's opinion on the target object at the previous sampling time point at the current sampling time point; if not, the group's opinion on the target object at the current sampling time point is the group's abrupt opinion.

[0015] In one embodiment, it further includes: determining whether the user's personal opinion on the target object and the group opinion on the target object are the same; if not, the target user's personal opinion and the group opinion do not match.

[0016] In one embodiment, the method further includes: calculating a corresponding polarity weight value based on a ratio of the frequency of occurrence of the opinion feature in the comment data to the total number of occurrences of each opinion feature.

[0017] A method for analyzing group speech opinions, characterized in that the method comprises:

[0018] Obtain the comment data of each user in the group about the target object and the character attribute information of each user at the sampling time point from the social platform;

[0019] Construct a comment information-word matrix based on the frequency of occurrence of each word in each comment data, where each row represents a comment data, each column represents a word, and the elements in the comment-word matrix represent the frequency of occurrence of the word in the current comment data;

[0020] Perform non-negative matrix factorization on the comment-word matrix to obtain the first matrix and the second matrix, where the columns of the first matrix represent words and the rows represent the weights of personal opinions on the words; the rows of the second matrix represent the weights of comment information on personal opinions, and the columns represent personal opinions;

[0021] The user weight value of each user is calculated based on the character attribute information of each user and the second matrix, and the row vectors in the first matrix are weighted and summed according to the user weight value of each user to obtain the opinion weight score;

[0022] Obtain a pre-constructed feature score table, and determine the group's opinions based on the opinion features corresponding to the opinion weight scores in the feature score table;

[0023] The group opinions and views corresponding to different sampling time points are connected in chronological order to obtain a change context of the group opinions and views; the change context of the group opinions and views is used for tracking the group opinions and views.

[0024] A group speech opinion tracking device, comprising:

[0025] The data acquisition module is used to obtain the comment data of each user in the group about the target object and the character attribute information of each user at the sampling time point from the social platform;

[0026] The opinion feature extraction module is used to perform part-of-speech tagging on the comment data to obtain multiple segments, construct a dependency syntax tree based on the dependency relationships between the segments, and parse the dependency syntax tree to obtain the opinion features in the comment data;

[0027] A module for determining personal speech opinions, configured to obtain a corresponding polarity weight value based on the frequency of occurrence of opinion features, obtain a personal speech opinion polarity score based on a weighted sum of the opinion polarity scores corresponding to each opinion feature in the feature score table and the polarity weight value, and determine the personal speech opinion based on the opinion feature corresponding to the personal speech opinion polarity score in the feature score table;

[0028] The group speech opinion determination module is used to calculate the corresponding user weight value based on the character attribute information of each user, calculate the opinion weight score based on the polarity score of each user's personal speech opinion and the user weight value, and determine the group speech opinion based on the opinion feature corresponding to the opinion weight score in the feature score table;

[0029] The result output module is used to connect the group opinions corresponding to different sampling time points in chronological order to obtain the change context of the group opinions; the change context of the group opinions is used to track the group opinions.

[0030] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0031] Obtain the comment data of each user in the group about the target object and the character attribute information of each user at the sampling time point from the social platform;

[0032] Perform part-of-speech tagging on the comment data to obtain multiple segments. Build a dependency syntax tree based on the dependency relationships between the segments. Parse the dependency syntax tree to obtain the opinion features in the comment data.

[0033] Obtain a corresponding polarity weight value based on the frequency of occurrence of the opinion feature, perform a weighted summation of the opinion polarity score corresponding to each opinion feature in the feature score table and the polarity weight value to obtain a personal speech opinion polarity score, and determine the personal speech opinion based on the opinion feature corresponding to the personal speech opinion polarity score in the feature score table;

[0034] Calculate the corresponding user weight value based on the character attribute information of each user, calculate the opinion weight score based on the polarity score of each user's personal opinion and the user weight value, and determine the group opinion based on the opinion feature corresponding to the opinion weight score in the feature score table;

[0035] The group opinions and views corresponding to different sampling time points are connected in chronological order to obtain a change context of the group opinions and views; the change context of the group opinions and views is used for tracking the group opinions and views.

[0036] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

[0037] Obtain the comment data of each user in the group about the target object and the character attribute information of each user at the sampling time point from the social platform;

[0038] Perform part-of-speech tagging on the comment data to obtain multiple segments. Build a dependency syntax tree based on the dependency relationships between the segments. Parse the dependency syntax tree to obtain the opinion features in the comment data.

[0039] Obtain a corresponding polarity weight value based on the frequency of occurrence of the opinion feature, perform a weighted summation of the opinion polarity score corresponding to each opinion feature in the feature score table and the polarity weight value to obtain a personal speech opinion polarity score, and determine the personal speech opinion based on the opinion feature corresponding to the personal speech opinion polarity score in the feature score table;

[0040] Calculate the corresponding user weight value based on the character attribute information of each user, calculate the opinion weight score based on the polarity score of each user's personal opinion and the user weight value, and determine the group opinion based on the opinion feature corresponding to the opinion weight score in the feature score table;

[0041] The group opinions and views corresponding to different sampling time points are connected in chronological order to obtain a change context of the group opinions and views; the change context of the group opinions and views is used for tracking the group opinions and views.

[0042] The above-mentioned group speech and opinion tracking method, device, computer equipment and storage medium can comprehensively cover the opinions expressed by the target group at each time point by obtaining user comment data and user information at different time points from the social platform, perform part-of-speech tagging and dependency syntax analysis on the comment data, accurately extract the opinion features in the comments, and calculate the polarity weight value based on its frequency of occurrence, thereby determining the individual speech and opinion. The user weight value calculated in combination with user information can measure the representativeness of the user in the group, and the group opinion is calculated based on the polarity score of the individual speech and opinion, thereby ensuring the accuracy of the group speech and opinion. By connecting the group speech and opinion at different time points in chronological order, it is possible to construct the change context of the group speech and opinion, making the evolution process of the group opinion clearer. The embodiment of the present invention can realize the dynamic tracking of group speech and opinion, and provide efficient and accurate support for public opinion analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 1 is a flow chart of a method for tracking group speech opinions in one embodiment;

[0044] Figure 2 A schematic diagram of a process for obtaining the changing trends of group opinions and views in one embodiment;

[0045] Figure 3 A structural block diagram of a group speech opinion tracking device in one embodiment;

[0046] Figure 4 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0048] In one embodiment, Figure 1 As shown, a method for tracking group speech opinions is provided, including the following steps:

[0049] Step 102 : obtaining comment data on the target object by each user in the group at the sampling time point and the character attribute information of each user from the social platform.

[0050] A group can be all users of the same product. In this application scenario, the target object is the product. It can also be opinion camps on the same event, such as opinion camp A and opinion camp B on a certain event. In this application scenario, the target object is the event and the group is all people in the opinion camp.

[0051] A user's profile information, which can include their identity, importance within the group, number of followers, frequency of posting, number of followers, and number of likes, is used to calculate their weight within the group. The user's importance within the group is determined by constructing a social network graph using a graph database or graph processing tool, determining how to represent nodes (users) and edges (relationships), calculating centrality metrics using a graph algorithm library, and directly outputting the graph centrality score as the importance metric.

[0052] Step 104 , perform part-of-speech tagging on the comment data to obtain multiple segments, construct a dependency syntax tree based on the dependency relationships between the segments, and parse the dependency syntax tree to obtain opinion features in the comment data.

[0053] Part-of-speech tagging is performed on the comment information of each user at each time point to obtain the tagging results of the comment information. Based on the tagging results of the comment information, the dependency relationship between each segment in the comment information is constructed, where each segment is a character, word or phrase, and may also include symbols. According to the dependency relationship between each segment in the comment information, the segment as the opinion feature in the comment information is extracted, and the number of occurrences of the opinion feature in the comment information is counted; based on the opinion feature in the comment information and its number of occurrences, the user's personal opinion on the target object at that time point is determined.

[0054] Part-of-speech tagging involves categorizing individual segments within a review, such as nouns, verbs, and adjectives (it can also include characters, names, and actions). This tagging helps determine the role and meaning of each segment within the review. Dependency building involves identifying interdependencies between segments within a review, such as whether an adjective modifies a noun or what action a character performs.

[0055] Based on the dependencies between the fragments in the comment information, we extract the fragments that serve as opinion features. Opinion features include sentiment words, degree words, negation words, domain-specific words, core topic words, grammatical structures, emoticons, and special characters.

[0056] Sentiment Words: Positive sentiment words such as "like," "satisfied," "excellent," and "great" typically express positive sentiment; negative sentiment words such as "hate," "disappointed," "terrible," and "poor" typically express negative sentiment. Degree Words: Intensifying words such as "very," "extremely," and "super" are used to strengthen sentiment words; weakening words such as "slightly," "a little," and "not quite" are used to weaken sentiment words. Negative words such as "no," "no," "none," and "non" can change the sentiment of the entire sentence. Domain-Specific Vocabulary: Specific vocabulary related to the review object, which can indicate the user's views on specific aspects. Core Topic Vocabulary: Topical and focal words in a review, such as "service," "price," and "quality," are often closely related to the user's specific views. Grammatical Structure: Subjective Sentences: These typically include structures such as "I feel" and "I think" to express personal opinions; comparative structures such as "better than..." and "not as good as..." are used to compare different objects. Emoticons and special characters.

[0057] Opinion features also include evaluation words: adjectives and adverbs used for evaluation, such as "fast", "cheap", "delicious", etc. Verbs: verbs that express opinions or actions, such as "recommend", "like", "avoid", "complain", etc.

[0058] Opinion features also include contextual and pragmatic features: the context of a sentence in a comment, including the content of the preceding and following sentences, may be crucial to understanding the opinion; pragmatic features, such as expressions such as sarcasm and exaggeration, may affect the interpretation of the opinion.

[0059] Step 106: Obtain the corresponding polarity weight value based on the frequency of occurrence of the opinion feature, perform weighted summation of the opinion polarity score and polarity weight value corresponding to each opinion feature in the feature score table to obtain the personal speech opinion polarity score, and determine the personal speech opinion based on the opinion feature corresponding to the personal speech opinion polarity score in the feature score table.

[0060] The feature score table is obtained by establishing a correspondence between opinion features and opinion polarity scores. The correspondence between different opinion features and opinion polarity scores can be established in advance, and then the opinion polarity score of each opinion feature in the review information can be determined based on this correspondence. Specifically, taking the group of all users of the same product as an example, analyze user opinions on a product. All opinion features in the review information can be divided into three categories: satisfied, dissatisfied, and neutral. The opinion polarity score of a satisfactory opinion feature is 100 points, the opinion polarity score of an dissatisfied opinion feature is 0 points, and the opinion polarity score of a neutral opinion feature is 50 points.

[0061] Taking the group as an example, all users who use the same product, the weight value Q of an opinion feature is calculated according to the following formula:

[0062] Q = the number of occurrences of this opinion feature / the total number of occurrences of all opinion features

[0063] According to the weight value of each opinion feature, the feature score of each opinion feature is weighted and summed to obtain the user's opinion score. Assuming the opinion score is 80, because 80 is closer to 100 points, the user's personal opinion at that time point is determined to be satisfied; assuming the opinion score is 70, because 70 is closer to 50 points, the user's personal opinion at that time point is determined to be neutral; assuming the opinion score is 20, because 20 is closer to 0 points, the user's personal opinion at that time point is determined to be dissatisfied.

[0064] Step 108, calculate the corresponding user weight value based on the character attribute information of each user, calculate the opinion weight score based on the polarity score of each user's personal opinion and the user weight value, and determine the group opinion based on the opinion feature corresponding to the opinion weight score in the feature score table.

[0065] The user's opinion score C at that time point is calculated according to the following formula:

[0066]

[0067] Where j is the index of the opinion feature, m is the number of opinion features in the comment information, Qj is the weight value of the j-th opinion feature, and Ej is the j-th opinion feature.

[0068] Taking the example of a group consisting of all users of the same product, each user's user weight is determined based on their identity, number of followers, speaking frequency, number of followers, number of likes, and other information. The higher the user's identity, the more followers they have, the more frequently they speak, the more followers they are, and the more likes they receive, the greater their user weight. Based on each user's user weight, the weighted sum of their corresponding opinion weight scores is calculated to yield the opinion weight score. By calculating user weights, it is possible to reasonably measure the influence of each user on group opinion when group representation is unclear or when there are multiple group representatives, thereby improving the accuracy of tracking group speech and opinion.

[0069] Assuming that the opinion weight score at this point in time is 80, since 80 is closer to 100 points, the group opinion at this point in time is determined to be satisfied; assuming that the opinion weight score at this point in time is 70, since 70 is closer to 50 points, the group opinion at this point in time is determined to be neutral; assuming that the opinion weight score at this point in time is 20, since 20 is closer to 0 points, the group opinion at this point in time is determined to be dissatisfied.

[0070] The user's opinion score W at that time point is calculated according to the following formula:

[0071]

[0072] Where j is the user's serial number, L is the number of users, Rj is the user weight value of the j-th user, and Ej is the opinion weight score of the j-th user.

[0073] Step 110: The group opinions and views corresponding to different sampling time points are connected in chronological order to obtain the changing context of the group opinions and views.

[0074] The evolution of group opinion and speech is used to track group opinion and speech. This can clearly demonstrate the evolution of group opinion over time. By connecting group opinions and speech at different points in time, it can identify the direction of change (e.g., from support to opposition, or from negativity to positivity). It can also identify key moments of opinion change, such as whether group opinion changes significantly after an event. Speech tracking can be used to promptly identify potential public opinion risks, allowing for appropriate guidance and intervention before negative sentiment escalates, thereby preventing the spread of destabilizing factors.

[0075] In the above-mentioned group speech and opinion tracking method, by obtaining user comment data and user information at different time points from the social platform, it is possible to comprehensively cover the opinions expressed by the target group at each time point, perform part-of-speech tagging and dependency syntax analysis on the comment data, accurately extract the opinion features in the comments, and calculate the polarity weight value based on its frequency of occurrence, thereby determining the individual speech and opinion. By combining the user information to calculate the user weight value, it is possible to measure the representativeness of the user in the group, and calculate the group opinion based on the polarity score of the individual speech and opinion, thereby ensuring the accuracy of the group speech and opinion. By connecting the group speech and opinion at different time points in chronological order, it is possible to construct the context of the change of group speech and opinion, making the evolution process of group opinion clearer. The embodiment of the present invention can realize the dynamic tracking of group speech and opinion, and provide efficient and accurate support for public opinion analysis.

[0076] In one embodiment, the group speech opinions corresponding to different sampling time points are connected in chronological order to obtain a change context of the group speech opinions, including: obtaining the group speech opinions at the i-th sampling time point, judging whether the group speech opinions at the i-th sampling time point are the same as the group speech opinions at the i-1-th sampling time point; if they are the same, deleting the group speech opinions at the i-th sampling time point; if they are not the same, retaining the group speech opinions at the i-th sampling time point, where i is the sampling time point sequence number, i is less than or equal to N, and N is the number of sampling time points; iteratively updating the value of i to obtain the group speech opinions at the next sampling time point until i is greater than N, stopping the iteration, connecting the current group speech opinions in chronological order to obtain a change context of the group speech opinions; and the change context of the group speech opinions is used for tracking the group speech opinions.

[0077] In this embodiment, if Figure 2 As shown, a flowchart for obtaining the changing context of group speech opinions is provided, including the following steps:

[0078] S201, determining whether i is greater than N, where N is the number of time points and i is the sequence number of the time point;

[0079] S202, when i is greater than N, the loop ends and the group opinions at all retained time points are connected in chronological order to obtain the context of opinion changes;

[0080] S203, when i is not greater than N, determining whether the group opinion at the i-th time point is the same as the group opinion at the i-1-th time point;

[0081] S204, if they are the same, delete the group speech opinions at the i-th time point;

[0082] S205, if they are not the same, then the group opinion at the i-th time point is retained, and i is updated with the value i plus 1, wherein the group opinion at the first time point does not need to be judged and is retained by default.

[0083] The group opinion at the first time point does not need to be evaluated; it is retained by default. The group opinion at the second time point is determined to be the same as the first. If so, the group opinion at the second time point is deleted; otherwise, it is retained. Set i to 3 and continue the next round until i equals N+1. The loop ends and all retained group opinion points are linked together to derive the evolution of opinion.

[0084] In one embodiment, determining whether the group speech opinions at the i-th sampling time point are the same as those at the i-1-th sampling time point includes: calculating the similarity between the group speech opinions at the i-th time point and the i-1-th time point; if the similarity is greater than a first threshold, the group speech opinions at the i-th time point and the i-1-th time point are the same; otherwise, they are not the same.

[0085] In one embodiment, the method further includes: determining whether the group's speech opinions on the target object at the current sampling time point are the same as the group's speech opinions on the target object at the previous sampling time point at the current sampling time point; if not, the group's speech opinions at the current sampling time point are the group's abrupt speech opinions.

[0086] In this embodiment, the current time point can also be any time point. Using the above-mentioned technical means, it is possible to determine whether the target group's group speech and opinions at a certain time point are consistent with the previous group speech and opinions. If the target group's group speech and opinions regarding the target object at this time point are the same as the target group's group speech and opinions regarding the target object at the previous time point, then the group speech and opinions at this time point are consistent with the previous group speech and opinions. If they are different, then the group speech and opinions at the current time point are determined to be the target group's abrupt speech and opinions.

[0087] In one embodiment, the method further includes: determining whether the user's personal opinion on the target object and the group opinion on the target object are the same; if not, the target user's personal opinion does not match the group opinion.

[0088] In this embodiment, the above-mentioned technical means can be used to determine whether the target user's personal speech and opinions match the group speech and opinions of the target group to which they belong. If the target user's personal speech and opinions on the target object are the same as the group speech and opinions of the target group on the target object, then they match. If they are different, then it is determined that the target user's personal speech and opinions do not match the group speech and opinions.

[0089] To determine whether the target user's personal speech and opinion on the target object and the target group's group speech and opinion on the target object are the same, the similarity between the target user's personal speech and opinion on the target object and the target group's group speech and opinion on the target object can be calculated. If the similarity is greater than a second threshold, it is considered that the target user's personal speech and opinion on the target object and the target group's group speech and opinion on the target object are the same; otherwise, they are not the same.

[0090] In one embodiment, obtaining the corresponding polarity weight value according to the frequency of occurrence of the opinion feature includes: calculating the corresponding polarity weight value according to the ratio of the frequency of occurrence of the opinion feature in the comment data to the total number of occurrences of each opinion feature.

[0091] In another embodiment, a method for analyzing opinions in group speech is provided, the method comprising:

[0092] Obtain the comment data of each user in the group about the target object and the character attribute information of each user at the sampling time point from the social platform;

[0093] Construct a comment information-word matrix based on the frequency of occurrence of each word in each comment data, where each row represents a comment data, each column represents a word, and the elements in the comment-word matrix represent the frequency of occurrence of the word in the current comment data;

[0094] Perform non-negative matrix factorization on the comment-word matrix to obtain the first matrix and the second matrix, where the columns of the first matrix represent words and the rows represent the weights of personal opinions on the words; the rows of the second matrix represent the weights of comment information on personal opinions, and the columns represent personal opinions;

[0095] The user weight value of each user is calculated based on the character attribute information of each user and the second matrix, and the row vectors in the first matrix are weighted and summed according to the user weight value of each user to obtain the opinion weight score;

[0096] Obtain a pre-constructed feature score table, and determine the group's opinions based on the opinion features corresponding to the opinion weight scores in the feature score table;

[0097] The group opinions and views corresponding to different sampling time points are connected in chronological order to obtain the changing context of group opinions and views; the changing context of group opinions and views is used to track group opinions and views.

[0098] It should be noted that a user's review can contain multiple personal opinions. For example, a user may identify a product as having three advantages and two disadvantages, which could be five personal opinions in total. Furthermore, different reviews can contain the same personal opinions, meaning different users can identify a product as having the same advantages and disadvantages.

[0099] Count the frequency of each word in each comment information to generate a comment information-word matrix V. Decompose V into two non-negative matrices W and H, so that V≈W*H. H is the first matrix and W is the second matrix. W is an n×k matrix, where n is the number of comment information and k is the number of personal opinions. Each column in W represents a personal opinion, and each row represents the weight of a comment information on these personal opinions. H is a k×m matrix, where m is the number of unique words. Each column represents a word, and each row represents the weight of a personal opinion on these words. That is, each row of the H matrix describes a personal opinion. Each column of the W matrix represents the distribution of a user's comment information on various personal opinions.

[0100] Furthermore, the group speech opinions are determined based on the character attribute information of each user, the first matrix and the second matrix, including: determining the attribute weight value of each user based on the character attribute information of each user and the second matrix; weighted summing up each row vector in the first matrix according to the attribute weight value of each user to obtain an opinion weight score; and determining the group speech opinions based on the opinion weight score.

[0101] Each user's attribute weight can be determined based on their personal attribute information. Each column in the W matrix can then be adjusted based on the attribute weights. Matrix calculations are then performed using the adjusted W matrix and the H matrix to obtain opinion weight scores. The opinion weight scores are then used to determine the group's opinion regarding the target object.

[0102] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. In addition, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0103] In one embodiment, Figure 3 As shown, a group speech opinion tracking device is provided, comprising:

[0104] The data acquisition module 302 is used to obtain the comment data of each user in the group on the target object and the character attribute information of each user at the sampling time point from the social platform;

[0105] The opinion feature extraction module 304 is used to perform part-of-speech tagging on the comment data to obtain multiple segments, construct a dependency syntax tree based on the dependency relationships between the segments, and parse the dependency syntax tree to obtain opinion features in the comment data;

[0106] The personal speech opinion determination module 306 is configured to obtain a corresponding polarity weight value based on the frequency of occurrence of the opinion feature, perform a weighted sum of the opinion polarity score and the polarity weight value corresponding to each opinion feature in the feature score table to obtain a personal speech opinion polarity score, and determine the personal speech opinion based on the opinion feature corresponding to the personal speech opinion polarity score in the feature score table;

[0107] The group speech opinion determination module 308 is used to calculate the corresponding user weight value based on the character attribute information of each user, calculate the opinion weight score based on the individual speech opinion polarity score and the user weight value of each user, and determine the group speech opinion based on the opinion feature corresponding to the opinion weight score in the feature score table;

[0108] The result output module 310 is used to connect the group opinions corresponding to different sampling time points in chronological order to obtain the change context of the group opinions; the change context of the group opinions is used for tracking the group opinions.

[0109] In one embodiment, it is also used to obtain the group speech opinions at the i-th sampling time point, determine whether the group speech opinions at the i-th sampling time point are the same as the group speech opinions at the i-1-th sampling time point, and if they are the same, delete the group speech opinions at the i-th sampling time point; if they are not the same, retain the group speech opinions at the i-th sampling time point, where i is the sampling time point sequence number, i is less than or equal to N, and N is the number of sampling time points; iteratively update the value of i to obtain the group speech opinions at the next sampling time point, until i is greater than N, stop iteration, connect the current group speech opinions in chronological order, and obtain the change context of group speech opinions; the change context of group speech opinions is used to track group speech opinions.

[0110] In one embodiment, it is also used to determine whether the group speech opinions at the i-th sampling time point are the same as those at the i-1-th sampling time point, including: calculating the similarity of the group speech opinions at the i-th time point and the i-1-th time point; if the similarity is greater than a first threshold, the group speech opinions at the i-th time point and the i-1-th time point are the same; otherwise, they are not the same.

[0111] In one embodiment, it is also used to determine whether the group's speech opinions on the target object at the current sampling time point are the same as the group's speech opinions on the target object at the previous sampling time point at the current sampling time point. If they are not the same, the group's speech opinions at the current sampling time point are the group's abrupt speech opinions.

[0112] In one embodiment, it is also used to determine whether the user's personal opinion on the target object and the group opinion on the target object are the same. If they are not the same, the target user's personal opinion and the group opinion do not match.

[0113] In one embodiment, it is also used to calculate the corresponding polarity weight value according to the ratio of the frequency of occurrence of the opinion feature in the comment data to the total number of occurrences of each opinion feature.

[0114] For the specific limitations of the group speech and opinion tracking device, please refer to the limitations of the group speech and opinion tracking method above, and will not be repeated here. The various modules in the above-mentioned group speech and opinion tracking device can be implemented in whole or in part through software, hardware, or a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of the above modules.

[0115] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, memory, network interface, display screen and input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a method for tracking group speech opinions is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad provided on the computer device housing, or an external keyboard, touchpad or mouse.

[0116] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0117] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps of the method in the above embodiment when executing the computer program.

[0118] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method in the above embodiment are implemented.

[0119] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. 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 can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0120] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0121] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and such modifications and improvements are intended to fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A method for tracking group speech opinions, characterized in that: The method comprises: Obtain the comment data of each user in the group about the target object and the character attribute information of each user at the sampling time point from the social platform; Perform part-of-speech tagging on the comment data to obtain multiple segments. Build a dependency syntax tree based on the dependency relationships between the segments. Parse the dependency syntax tree to obtain the opinion features in the comment data. Obtain a corresponding polarity weight value based on the frequency of occurrence of the opinion feature, perform a weighted summation of the opinion polarity score corresponding to each opinion feature in the feature score table and the polarity weight value to obtain a personal speech opinion polarity score, and determine the personal speech opinion based on the opinion feature corresponding to the personal speech opinion polarity score in the feature score table; Calculate the corresponding user weight value based on the character attribute information of each user, calculate the opinion weight score based on the polarity score of each user's personal opinion and the user weight value, and determine the group opinion based on the opinion feature corresponding to the opinion weight score in the feature score table; The group opinions and views corresponding to different sampling time points are connected in chronological order to obtain a change context of the group opinions and views; the change context of the group opinions and views is used for tracking the group opinions and views.

2. The method according to claim 1, characterized in that By connecting the opinions of the groups corresponding to different sampling time points in chronological order, we can obtain the following changes in the opinions of the groups: Obtain the group opinion at the i-th sampling time point, and determine whether the group opinion at the i-th sampling time point is the same as the group opinion at the i-1-th sampling time point. If they are the same, delete the group opinion at the i-th sampling time point; if they are not the same, retain the group opinion at the i-th sampling time point, where i is the sampling time point number, i is less than or equal to N, and N is the number of sampling time points. Iteratively update the value of i to obtain the group opinion at the next sampling time point. When i is greater than N, the iteration is stopped and the current group opinion is connected in chronological order to obtain the change context of the group opinion. The change context of the group opinion is used for tracking the group opinion.

3. The method according to claim 2, characterized in that Determining whether the group opinions at the i-th sampling time point and the i-1-th sampling time point are the same includes: Calculate the similarity between the group opinions at the i-th time point and the i-1-th time point. If the similarity is greater than a first threshold, the group opinions at the i-th time point and the i-1-th time point are the same; otherwise, they are not the same.

4. The method according to claim 1, wherein The method further comprises: Determine whether the group's opinion on the target object at the current sampling time point is the same as the group's opinion on the target object at the previous sampling time point. If not, the group's opinion at the current sampling time point is the group's abrupt opinion.

5. The method according to claim 1, wherein The method further comprises: Determine whether the user's personal opinion on the target object is the same as the group opinion on the target object. If not, the target user's personal opinion does not match the group opinion.

6. The method according to claim 1, characterized in that The corresponding polarity weight values ​​obtained according to the frequency of occurrence of opinion features include: The corresponding polarity weight value is calculated based on the ratio of the frequency of occurrence of opinion features in the comment data to the total number of occurrences of each opinion feature.

7. A method for analyzing group speech opinions, characterized in that: The method comprises: Obtain the comment data of each user in the group about the target object and the character attribute information of each user at the sampling time point from the social platform; Construct a comment information-word matrix based on the frequency of occurrence of each word in each comment data, where each row represents a comment data, each column represents a word, and the elements in the comment-word matrix represent the frequency of occurrence of the word in the current comment data; Perform non-negative matrix factorization on the comment-word matrix to obtain the first matrix and the second matrix, where the columns of the first matrix represent words and the rows represent the weights of personal opinions on the words; the rows of the second matrix represent the weights of comment information on personal opinions, and the columns represent personal opinions; The user weight value of each user is calculated based on the character attribute information of each user and the second matrix, and the row vectors in the first matrix are weighted and summed according to the user weight value of each user to obtain the opinion weight score; Obtain a pre-constructed feature score table, and determine the group's opinions based on the opinion features corresponding to the opinion weight scores in the feature score table; The group opinions and views corresponding to different sampling time points are connected in chronological order to obtain a change context of the group opinions and views; the change context of the group opinions and views is used for tracking the group opinions and views.

8. A group speech and opinion tracking device, characterized in that: The device comprises: The data acquisition module is used to obtain the comment data of each user in the group about the target object and the character attribute information of each user at the sampling time point from the social platform; The opinion feature extraction module is used to perform part-of-speech tagging on the comment data to obtain multiple segments, construct a dependency syntax tree based on the dependency relationships between the segments, and parse the dependency syntax tree to obtain the opinion features in the comment data; A module for determining personal speech opinions, configured to obtain a corresponding polarity weight value based on the frequency of occurrence of opinion features, obtain a personal speech opinion polarity score based on a weighted sum of the opinion polarity scores corresponding to each opinion feature in the feature score table and the polarity weight value, and determine the personal speech opinion based on the opinion feature corresponding to the personal speech opinion polarity score in the feature score table; The group speech opinion determination module is used to calculate the corresponding user weight value based on the character attribute information of each user, calculate the opinion weight score based on the polarity score of each user's personal speech opinion and the user weight value, and determine the group speech opinion based on the opinion feature corresponding to the opinion weight score in the feature score table; The result output module is used to connect the group opinions corresponding to different sampling time points in chronological order to obtain the change context of the group opinions; the change context of the group opinions is used to track the group opinions.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images