A network public opinion monitoring method and system

Abstract

The application provides a network public opinion monitoring method and system, and relates to the technical field of data processing.The network public opinion monitoring method comprises the following steps: obtaining network public opinion data according to a public opinion event; performing public opinion orientation extraction on the network public opinion data to obtain public opinion triggering data; obtaining public opinion triggering source data according to the public opinion triggering data; performing regulation and control on event update data of the public opinion event according to the public opinion triggering source data, the public opinion triggering data and an expected target, and generating event public relation data for publication on the whole network.The network public opinion monitoring method and system can effectively supervise and regulate the development trend of public opinion.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for monitoring online public opinion. Background Technology

[0002] Online public opinion monitoring refers to the act of collecting, analyzing, and visualizing information across the entire internet through technical means to generate hot topic statistics and dynamic monitoring. Its core objective is to identify online public opinion trends and generate statistical reports using technologies such as natural language processing and data mining.

[0003] In the current process of online public opinion supervision, once a public opinion incident occurs, the monitoring of public opinion on various online platforms is insufficient, which easily leads to the leakage of public opinion data. Moreover, once a public opinion incident occurs, the public opinion public relations related to the incident cannot fully implement public opinion prevention and control measures, which leads to the public opinion incident continuing to develop in a vicious cycle after the public opinion public relations are done, which is not conducive to public opinion management. Summary of the Invention

[0004] In view of the shortcomings of the prior art described above, the purpose of this invention is to provide a method and system for monitoring online public opinion, which solves the problem that once public opinion occurs, the monitoring of public opinion on various online platforms is insufficient, and public opinion data is easily leaked; moreover, once public opinion occurs, the public opinion public relations of related public opinion events cannot fully implement public opinion prevention and control measures, thus causing public opinion to continue to develop in a vicious cycle after the public opinion public relations efforts.

[0005] To achieve the above and other related objectives, the present invention provides a method for monitoring online public opinion, comprising: acquiring online public opinion data based on public opinion events; performing public opinion guidance extraction on the online public opinion data to acquire public opinion triggering data; acquiring public opinion triggering source data based on the public opinion triggering data; and regulating the event update data of the public opinion event based on the public opinion triggering source data, the public opinion triggering data, and the expected goals to generate event public relations data for publication across the entire network.

[0006] In one embodiment of the present invention, obtaining online public opinion data based on public opinion events includes: extracting keywords from the public opinion events to obtain public opinion keywords. Search and public opinion keywords Semantically related keywords as public opinion-related terms ; Keywords related to public opinion Keywords related to public opinion Perform platform conversion to generate platform expressions corresponding to the platform language of the respective network platform. ; through public opinion keywords Keywords related to public opinion platform expressions Conduct public opinion searches on relevant online platforms to obtain online public opinion data.

[0007] In one embodiment of the present invention, public opinion keywords are... Keywords related to public opinion Perform platform conversion to generate platform expressions corresponding to the platform language of the respective network platform. This includes: keywords related to public opinion. Keywords related to public opinion Perform word segmentation to obtain a word data set. The word dataset includes at least one of single-character data and word data; based on the word dataset... Word data Find the first set of terms used by the network platform. ,in, This indicates a subset of terms used on the first platform. , Represents data for each word Corresponding terms used on the primary platform; select the terminology set used on the primary platform. A subset of terms used on several first-tier platforms And select multiple subsets of terms used on the first platform The corresponding first platform terminology Perform a combination, then combine it with other word data. A second combination is performed to generate the first platform's expressive words, wherein each combination includes at least one subset of the first platform's expressive words. The corresponding first platform terminology ;Keywords for public opinion Keywords related to public opinion Semantic transformation is performed to obtain semantic information; based on the semantic information, the second platform expression term corresponding to the network platform is found; based on the first and second platform expression terms, the platform expression term corresponding to the platform language of the corresponding network platform is generated. .

[0008] In one embodiment of the present invention, public opinion guidance extraction is performed on online public opinion data to obtain public opinion triggering data, including: extracting dialogue data from the online public opinion data, the dialogue data including question data and answer data; sorting the dialogues according to the correspondence between the answer data and the question data to obtain related dialogue distribution data; extracting dialogue strings from the related dialogue distribution data to obtain multiple related dialogue strings, the related dialogue strings including question data and answer data that are hierarchically related; and performing related sentiment analysis on the question data and answer data in the related dialogue strings to obtain related sentiment classification and related sentiment feedback trends, which are used as public opinion triggering data.

[0009] In one embodiment of the present invention, dialogue sorting is performed based on the correspondence between answer data and question data to obtain associated dialogue distribution data, including: calculating the correspondence between each answer data and each question data to obtain a correspondence set. ; Set the corresponding degree Each correspondence in Correspondence threshold Comparison: When Correspondence Greater than the corresponding threshold When the value exceeds the corresponding degree threshold, it will be greater than the threshold value. Correspondence The corresponding answer data serves as the answer to the question data, generating a single dialogue; the single dialogue is then evaluated and categorized to determine the corresponding topic focus. and the direction of a single conversation correlation between ; relevance Greater than the set correlation threshold The same topic points to By comparing the content association levels between corresponding individual conversations, we can obtain the topic references for each conversation. The relationships between individual dialogues under the corresponding dimension are analyzed, including hierarchical relationships and parallel hierarchical relationships. The individual dialogues corresponding to the relationships are sorted according to the corresponding hierarchical arrangement order to obtain the distribution data of the related dialogues.

[0010] In one embodiment of the present invention, the correlation degree is... Greater than the set correlation threshold The same topic points to By comparing the content association levels between corresponding individual conversations, we can obtain the topic references for each conversation. The relationships between individual conversations within the corresponding dimension include: linking each individual conversation to the relevant topic. Semantic transformation is performed on the dimensions to obtain the topic content; causal relationships are compared between the topic content corresponding to each individual dialogue to obtain the causal correlation degree between the individual dialogues. Based on the relevance of a single dialogue Relevance Corresponding association weight and causal correlation The correlation level index was calculated. ,in, When related hierarchical indicators Greater than the indicator threshold When this happens, a hierarchical relationship is established between the corresponding individual dialogues; when related hierarchical indicators... Less than the indicator threshold When the time is reached, a parallel hierarchical relationship is established between the corresponding single dialogues.

[0011] In an embodiment of the present application, the problem data and the answer data in the associated dialogue string are subjected to associated sentiment analysis to obtain associated sentiment classification and associated sentiment feedback trend as public opinion triggering data, including: performing sentiment detection on all associated dialogue strings under the same topic direction dimension to obtain sentiment elements and element evaluation; sorting all sentiment elements to obtain associated sentiment classification and a sentiment element set corresponding to each associated sentiment classification; calculating an associated sentiment feedback value of the sentiment element according to the degree value of the element evaluation corresponding to the sentiment element and the associated factor of the sentiment element in the dialogue string; and obtaining the associated sentiment feedback trend according to the associated sentiment feedback value.

[0012] In an embodiment of the present application, the public opinion triggering data includes associated sentiment classification, associated sentiment feedback trend and associated dialogue string; the public opinion triggering source data is obtained according to the public opinion triggering data, including: performing dialogue searching on the associated dialogue string according to the associated sentiment classification to obtain problem data and answer data arranged in the order of the associated dialogue string; and performing upstream data searching on the associated dialogue string corresponding to the associated sentiment classification according to the associated sentiment feedback trend to obtain initial data corresponding to the associated sentiment feedback trend as the public opinion triggering source data, wherein the initial data includes one of initial problem data and initial answer data.

[0013] In an embodiment of the present application, the event update data of the public opinion event is regulated according to the public opinion triggering source data, the public opinion triggering data and the expected target to generate event public relations data for publication on the whole network, including: performing data conversion on the associated sentiment feedback trend, the associated sentiment classification of the public opinion triggering source data and the public opinion triggering source data to obtain public opinion guidance of the public opinion triggering source data; comparing the public opinion guidance with the expected guidance corresponding to the expected target to find the public opinion guidance closest to the expected guidance and calculate the deviation degree of the public opinion guidance relative to the expected guidance; adjusting the associated sentiment feedback trend and the associated sentiment classification according to the deviation degree to obtain expected sentiment trend and expected sentiment classification; adjusting the public opinion triggering source data according to the deviation degree, the expected sentiment trend and the expected sentiment classification to obtain expected triggering source data; generating expected event data based on the expected triggering source data; and regulating the event update data of the public opinion event according to the expected event data to generate event public relations data for publication on the whole network.

[0014] ​​​To achieve the above object and other related objects, the present application further provides a network public opinion monitoring system, comprising: an acquisition unit configured to acquire network public opinion data according to a public opinion event; an extraction unit configured to perform public opinion-oriented extraction on the network public opinion data to acquire public opinion triggering data; a query unit configured to acquire public opinion triggering source data according to the public opinion triggering data; and a publishing unit configured to regulate and control event update data of the public opinion event according to the public opinion triggering source data, the public opinion triggering data and an expected target, to generate event public relations data for publication on the whole network.

[0015] As described above, the network public opinion monitoring method and system of the present application have the following beneficial effects: by using the public opinion event to set diversified search keywords adapted to different network platforms to regulate and control the acquisition of network public opinion data of different network platforms, the sufficiency of retrieval of each network platform can be effectively improved, and the retrieval coverage of network public opinion data on each network platform is ensured; by performing public opinion-oriented extraction on the acquired network public opinion data, various public opinion orientations and public opinion triggering source data can be quickly locked; and in the further regulation and control of the public opinion event, the public opinion triggering source data can be converted based on the expected target to obtain expected event data corresponding to the expected target, so that the event update data after the evolution of the public opinion event can be adjusted according to the expected event data, and the adjusted data can be published as event public relations data, thereby realizing the regulation and control of the evolution trend of the public opinion towards the expected target, such as strengthening or weakening the influence of the public opinion, so as to realize the monitoring and regulation and control of the development trend of the public opinion. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 The flowchart of the network public opinion monitoring method provided by the embodiment of the present application is shown.

[0017] Figure 2 The tree structure diagram of the associated conversation distribution data provided by the embodiment of the present application is shown.

[0018] Figure 3 The structural block diagram of the network public opinion monitoring system provided by the embodiment of the present application is shown.

[0019] Figure 4 The structural diagram of the electronic device of the embodiment of the present application is shown.

[0020] Element number explanation: electronic device 1; network public opinion monitoring system 11; memory 12; processor 13; acquisition unit 111; extraction unit 112; query unit 113; publishing unit 114. DETAILED DESCRIPTION

[0021] Following, the advantages and effects of the present application can be easily understood by those skilled in the art from the description. The present application can also be implemented or applied by different specific embodiments, and the details in the description can be modified or changed based on different views and applications without departing from the spirit of the present application. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0022] It should be noted that the diagrams provided in the following embodiments only illustrate the basic concepts of the present application in a schematic manner, and only show the components related to the present application in the diagrams, not the number of components, shapes and sizes when actually implemented. The shapes, numbers and proportions of the components can be arbitrarily changed, and the layout of the components can be more complex.

[0023] In the following description, a large number of details are discussed to provide a more thorough explanation of the embodiments of the present application, however, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details, and in other embodiments, the known structures and devices are shown in the form of block diagrams rather than in the form of details, to avoid making the embodiments of the present application difficult to understand.

[0024] The present application provides a network public opinion monitoring method, by regulating the acquisition of network public opinion data of different network platforms according to public opinion events, the sufficiency of retrieval of each network platform can be effectively improved, and by extracting the network public opinion data obtained in the direction of public opinion, the rapid locking of various public opinion oriented and public opinion triggering source data can be realized, then, in the process of public opinion regulation, the event update data after the evolution of public opinion events can be adjusted based on the expected target to be published as event public relations data, so that the evolution trend of public opinion can be regulated to develop towards the expected target by publishing event public relations, thereby realizing effective regulation of the development of public opinion.

[0025] Figure 1 The flowchart of the network public opinion monitoring method in an exemplary embodiment of the present application is shown, which is applied in a network public opinion monitoring system, including steps S10-S40. The technical solutions of the present application will be described in detail below in combination with Figure 1 The technical solutions of the present application will be described in detail below in combination with

[0026] First, step S10 is performed, network public opinion data is acquired according to public opinion events.

[0027] When acquiring online public opinion data in an online public opinion monitoring system, since the spread of public opinion events requires a certain number of events, the data can be obtained from various online platforms based on public opinion events that have occurred some time ago.

[0028] In step S10, obtaining online public opinion data based on public opinion events may further include:

[0029] Extracting keywords from public opinion events to obtain public opinion keywords. ;

[0030] Search and public opinion keywords Semantically related keywords as public opinion-related terms ;

[0031] Key words for public opinion Keywords related to public opinion Perform platform conversion to generate platform expressions corresponding to the platform language of the respective network platform. ;

[0032] Through public opinion keywords Keywords related to public opinion platform expressions Conduct public opinion searches on relevant online platforms to obtain online public opinion data.

[0033] When acquiring online public opinion data based on public opinion events, the online public opinion monitoring system first processes the public opinion events to extract keywords from them, thereby obtaining public opinion keywords that reflect the information of the events. Specifically, this can be achieved by breaking down keywords into various types, then filtering them, and performing semantic comparisons with public opinion events to obtain public opinion keywords that match the events. Then, based on the derived public opinion keywords... By performing similar word searches, keywords related to public opinion can be obtained. Semantic related terms for public opinion In order to achieve the goal of utilizing public opinion keywords Keywords related to public opinion To improve the accuracy of online public opinion data searches, it's crucial to consider the potential differences in terminology across various online platforms. Furthermore, to ensure the accuracy of public opinion keywords... Keywords related to public opinion It can be applied to corresponding online platforms, and through the online public opinion monitoring system, it can further realize the analysis of public opinion keywords. Keywords related to public opinion Perform keyword conversion on relevant online platforms to transform public opinion keywords. Keywords related to public opinion Convert to platform-specific terms applicable to the corresponding online platforms This allows for the use of relevant online keywords to search for public opinion. Keywords related to public opinion platform expressions Simultaneously conducting public opinion searches on relevant online platforms allows for a more comprehensive and thorough search of public opinion across these platforms, improving the completeness of online public opinion data when acquired from these platforms.

[0034] Among them, public opinion keywords Keywords related to public opinion Perform platform conversion to generate platform expressions corresponding to the platform language of the respective network platform. It may further include:

[0035] Key words in public opinion Keywords related to public opinion Perform word segmentation to obtain a word data set. The word data set includes at least one of single-character data and word data;

[0036] Based on word and phrase data set Word data in Find the first set of terms used by the network platform. ,in, This indicates a subset of terms used on the first platform. , Represents data for each word The corresponding terminology used on the first platform;

[0037] Select the first platform's terminology set A subset of terms used on several first-tier platforms And select multiple subsets of terms used on the first platform The corresponding first platform terminology Perform a combination, then combine it with other word data. A second combination is performed to generate the first platform's expressive words, wherein each combination includes at least one subset of the first platform's expressive words. The corresponding first platform terminology ;

[0038] Key words in public opinion Keywords related to public opinion Perform semantic transformation to obtain semantic information;

[0039] Find the second platform expression word corresponding to the network platform according to the semantic information;

[0040] Generate the platform expression word corresponding to the platform language of the corresponding network platform according to the first platform expression word and the second platform expression word .

[0041] When converting the public opinion keyword and the public opinion related word into the platform expression word corresponding to the platform language of the corresponding network platform , first split each keyword or related word in the public opinion keyword and the public opinion related word to obtain single - character data and word data. Since there are not only single - character platform languages but also word - based platform languages in the platform language of the network platform, and there may be more than one expression form for each platform language. For example, "WeChat" in the chat tool is used as "Mou Xin", "Green Bubble", "VX", etc. on a certain network platform. Therefore, after splitting out the set of word data , multiple sets of first - platform words corresponding to the word data (which can be a single - character data or word data) can be obtained ' . And by constructing a subset of first - platform words , each subset of first - platform words may contain multiple sets of first - platform words .

[0042] When combining word data based on several subsets of first - platform words corresponding to the network platform in the public opinion event ' , it can be to first select the first - platform words corresponding to several subsets of first - platform words for a first combination, and then, after the first combination, combine with other word data in the word data set for a second combination to generate the first - platform expression word. Of course, it can also be to directly combine with other word data in the word data set after selecting the first - platform words corresponding to several subsets of first - platform words to obtain the first - platform expression word at once. Among them, when selecting the first - platform words corresponding to several subsets of first - platform words ' , any first - platform word in the subset of first - platform words can be selected . When selecting the first - platform words corresponding to several subsets of first - platform words , any first - platform word in the subset of first - platform words The combination of any other first platform word in the other first platform word subset of the other first platform can be combined to achieve the combination of the derived first platform expression word, ensuring the diversity of the search keywords for the network public opinion data retrieval.

[0043] In addition, in order to further improve the diversity of the search keywords, the network public opinion monitoring system will further respectively convert the public opinion keywords and the public opinion related words Accordingly, after obtaining the corresponding semantic information, further network platform language query is carried out based on the derived semantics, so as to obtain the second platform expression word corresponding to the network platform. It is worth noting that the public opinion keywords and the public opinion related words The real semantics of the words are different.

[0044] By using the public opinion keywords and the public opinion related words The first platform expression word is obtained by splitting the words and analyzing the single word and single word. The second platform expression word is obtained by using the public opinion keywords and the public opinion related words The platform expression word based on the network platform language Can improve the comprehensiveness of platform language conversion and improve the coverage of public opinion search using platform expression word .

[0045] Then, the public opinion keywords , the public opinion related words and the platform expression words Carry out public opinion search on the corresponding network platform to obtain network public opinion data, which can further include:

[0046] Search data on the corresponding network platform by using the public opinion keywords , the public opinion related words and the platform expression words , and obtain the search result data;

[0047] The comprehensive semantic conversion is carried out on each search result data to obtain the search semantics.

[0048] The search semantics and the corresponding semantics of the public opinion event are compared in terms of semantic association.

[0049] When the semantic association degree of the search semantics and the corresponding semantics of the public opinion event reaches the semantic association degree threshold, the search result data is taken as the network public opinion data.

[0050] The network public opinion monitoring system carries out the public opinion search on the corresponding network platform by using the public opinion keywords , public opinion related words and platform expression words In the process of searching public opinion of the corresponding network platform, the data of the corresponding network platform can be searched by using the public opinion keywords , public opinion related words and platform expression words to obtain search result data under corresponding data retrieval. In the process of using the search result data, in order to solve the problem of retrieval illusion caused by the high similarity of the keywords in the network platform database, but actually unrelated to the public opinion event, after obtaining the search result data, the corresponding semantics of the search result data is summarized and then compared with the corresponding semantics of the public opinion event. The semantic correlation is compared to achieve the semantic correlation between the search semantics and the corresponding semantics of the public opinion event when the semantic correlation reaches the semantic correlation threshold, and then the search result data is used as the network public opinion data, so that the correspondence between the network public opinion data and the public opinion event can be effectively ensured when the network public opinion data is obtained through the public opinion event. Thus, it is convenient for subsequent accurate analysis and determination of public opinion triggering data, public opinion triggering source data, etc. based on network public opinion data.

[0051] Then, step S20 is performed to extract public opinion oriented from the network public opinion data to obtain public opinion triggering data.

[0052] After the network public opinion monitoring system obtains the network public opinion data, it further extracts the public opinion oriented from the network public opinion data to analyze and obtain the public opinion triggering data, so as to determine the corresponding public opinion triggering source data based on the public opinion triggering data, and to realize the control of the event update data of the public opinion event by using the public opinion triggering source data and the public opinion triggering data, so as to accurately cope with the public opinion crisis.

[0053] In step S20, the public opinion oriented extraction is performed on the network public opinion data to obtain the public opinion triggering data, which can further include:

[0054] Extracting conversation data in the network public opinion data, the conversation data including question data and answer data;

[0055] According to the correspondence between the answer data and the question data, the conversation is sorted to obtain associated conversation distribution data;

[0056] The associated conversation distribution data is extracted to obtain a plurality of associated conversation strings, the associated conversation string including question data and answer data associated layer by layer;

[0057] The question data and answer data in the associated conversation string are analyzed to obtain associated emotional classification and associated emotional feedback trend as public opinion triggering data.

[0058] When extracting public opinion guidance from online public opinion data, the online public opinion monitoring system can first extract dialogue data from the data. This extraction can be based on dialogues between different users, yielding corresponding question and answer data for each user. Alternatively, dialogue data can be obtained using language recognition models and / or text dialogue models. After obtaining the dialogue data, further dialogue analysis is performed to determine the correspondence between the question and answer data, resulting in related dialogue distribution data. This addresses situations where answers are given multiple times in real-world dialogues, or where other users have already answered the same question in other dialogues. Dialogue analysis ensures accuracy and correspondence, resulting in a more reasonable distribution of related dialogue data. This related dialogue distribution data can be understood as a tree-like structure establishing the correspondence between question and answer data and their hierarchical order based on different users and topics (including topic direction). Figure 2 The distribution data of associated dialogues in this tree structure is shown.

[0059] Furthermore, after obtaining the associated dialogue distribution data, the dialogue strings within this data are extracted to identify interconnected dialogue strings containing multiple questions and answers. Then, sentiment analysis is performed on each associated topic string to classify sentiment and determine sentiment feedback trends, thus providing data for public opinion events. This sentiment analysis may include analyzing the focus of public opinion on each associated topic string and the attitudes expressed towards that focus, thereby determining the focus category of the corresponding associated topic string and the attitude trends towards that focus during discussion. Of course, the sentiment classification and sentiment feedback trends can also be other categories and trends.

[0060] This includes analyzing the dialogues based on the correspondence between answer data and question data to obtain related dialogue distribution data, which may further include:

[0061] Calculate the correspondence between each answer and each question to obtain the correspondence set. ;

[0062] Correspondence set Each correspondence in Correspondence threshold Comparison:

[0063] When Correspondence greater than a corresponding degree threshold value If the corresponding degree is greater than the corresponding degree threshold value corresponding answer data as the answer to the question data to generate a single dialogue;

[0064] topic evaluation classification is performed on the single dialogue to obtain a corresponding topic direction and the association degree between the single dialogue and the topic direction

[0065] content association level comparison is performed between the single dialogues corresponding to the same topic direction with an association degree greater than a set association degree threshold value to obtain the association relationship between the single dialogues corresponding to each topic direction in the corresponding dimension, wherein the association relationship includes a superior-inferior hierarchical relationship and a parallel hierarchical relationship;

[0066] The single dialogues corresponding to the association relationship are associated and sorted according to the corresponding hierarchical arrangement order to obtain association dialogue distribution data.

[0067] In the dialogue carding process of the network public opinion monitoring system, the answer data and the answer data are first used to calculate the corresponding degree between each answer data and each question data to construct a corresponding degree set . When calculating the corresponding degree, a question and answer matching model can be used, for example, a similarity measurement model, a cosine similarity algorithm is used to calculate the semantic correlation between the question and the answer as the corresponding degree, of course, other corresponding degree calculation models can also be used to calculate the corresponding degree between each answer data and each question data. Then, the association relationship between each single dialogue is determined based on the corresponding degree between the answer data and the answer data, and then the corresponding association relationship is used to construct the association dialogue distribution data. Specifically, by comparing each corresponding degree with the corresponding degree threshold value , it is determined whether there is a certain corresponding relationship between the question data and the answer data. If the corresponding degree is greater than the corresponding degree threshold value , it means that there is a corresponding relationship, and a single dialogue can be generated. By analogy, multiple single dialogues composed of question data and answer data can be obtained. Then, the discussion topic evaluation classification is performed on each single dialogue to determine the corresponding topic direction and the association degree between the single dialogue and the topic direction ​​​​​​When classifying topics for evaluation, one can use a topic classification model to classify each dialogue, or other classification models can be used. Furthermore, the resulting classification can be a known classification; alternatively, to increase flexibility, a summary of the topic focus of the current dialogue can be provided.

[0068] Specifically, semantic recognition is performed on each individual dialogue, and multiple topic orientations are summarized from different analytical perspectives. The correlation between each topic orientation and the individual dialogue is then calculated. And select the correlation degree Greater than the set correlation threshold The same topic points to The corresponding single dialogue simultaneously summarizes the same topic. Single conversations are extracted, and then the extracted conversations are compared in terms of content relevance to identify those that share the same topic. Is the relationship between individual dialogues hierarchical or parallel? Based on the association relationship within individual dialogues, can we achieve targeting of the same topic? All corresponding single dialogues are sorted together to obtain a tree-structured distribution of associated dialogues.

[0069] In the process of sorting the single dialogues corresponding to the association relationship according to the corresponding hierarchical arrangement to obtain the association dialogue distribution data, the single dialogues with upper and lower hierarchical association relationships can be sorted level by level as the first association dialogue distribution data to represent the distribution of the hierarchical extension relationship of the topic; the single dialogues with parallel association relationships can be sorted in parallel as the second association dialogue distribution data to represent the distribution of topics that are all at the same level; then, the association dialogue distribution data is obtained based on the first association dialogue distribution data and the second association dialogue distribution data.

[0070] Furthermore, the correlation Greater than the set correlation threshold The same topic points to By comparing the content association levels between corresponding individual conversations, we can obtain the topic references for each conversation. The relationships between individual conversations under the corresponding dimension can include:

[0071] Each individual dialogue will be assigned to a corresponding topic. The semantics of the dimensions are transformed to obtain the topic content;

[0072] By comparing the causal relationships of the topics corresponding to each individual dialogue, the degree of causal correlation between the individual dialogues is obtained. ;

[0073] Based on the relevance of a single dialogue Relevance Corresponding association weight and causal correlation The correlation level index was calculated. ,in, ;

[0074] When related hierarchical indicators Greater than the indicator threshold In this case, a hierarchical relationship is established between the corresponding individual dialogues;

[0075] When related hierarchical indicators Less than the indicator threshold In this case, a parallel hierarchical relationship is established between the corresponding individual dialogues.

[0076] Establish the direction of each topic through the online public opinion monitoring system. In the process of establishing connections between individual dialogues within the corresponding dimension, based on the same topic... By semantically transforming each individual dialogue into its corresponding dimension, we can obtain the corresponding topic. The topic content, that is, transforming a single conversation into one centered around a specific topic. Then, the topic content is used for reasonable narration. Next, the causal relationship between the topic content of each individual dialogue is compared to determine whether each question and answer data points to the same topic. Causal correlation in topic content under different dimensions In determining the degree of causal relationship In this case, a causal correlation prediction model can be obtained by first training the model based on the topic content and the labeled causal correlation between the topics, and then identifying and predicting the corresponding causal correlation based on different topic content. Of course, other methods can also be used to measure the degree of causal relationship. The calculation of causal correlation. Then, the relevance of a single conversation is analyzed. The degree of correlation obtained by looking up the table Corresponding association weight To achieve the management of related hierarchical indicators The calculation can then be performed by associating hierarchical indicators. With indicator threshold By comparing the relevant hierarchical indicators, the appropriate level of indicators can be determined. Greater than the indicator threshold When this happens, a hierarchical relationship is established between the corresponding individual dialogues, and the associated hierarchical indicators are... Less than the indicator threshold If so, a parallel hierarchical relationship is established between the corresponding single dialogues. Specifically, the correlation level indicator The calculation formula can be expressed as , represents the correlation degree of another single dialogue, represents the correlation weight corresponding to the correlation degree of another single dialogue.

[0077] In addition, the question data and answer data in the associated dialogue string are subjected to correlation sentiment analysis to obtain correlation sentiment classification and correlation sentiment feedback trend as public opinion triggering data, including:

[0078] The sentiment of all associated dialogue strings under the same topic direction dimension is detected to obtain sentiment elements and element evaluation; wherein the element evaluation can be "extremely", "very", "very", "good", "general", etc.

[0079] All sentiment elements are sorted and classified to obtain correlation sentiment classification and the sentiment element set corresponding to each correlation sentiment classification;

[0080] According to the degree value of the element evaluation corresponding to the sentiment element and the correlation factor of the sentiment element in the dialogue string , the correlation sentiment feedback value of the sentiment element is calculated.

[0081] According to the correlation sentiment feedback value, the correlation sentiment feedback trend is obtained.

[0082] Through the network public opinion monitoring system in the process of correlation sentiment analysis, the sentiment elements and element evaluation in all associated dialogue strings under the same topic direction dimension are first obtained. Then, the sentiment elements are classified to obtain the correlation sentiment classification corresponding to each sentiment element, and thus the sentiment element set corresponding to each correlation sentiment classification is obtained. For each sentiment element, the corresponding degree value is obtained according to the element evaluation, and the correlation factor corresponding to the level is obtained according to the level of the sentiment element in the dialogue string. Then, based on the degree value and the correlation factor , the correlation sentiment feedback value of the sentiment element is calculated. Finally, the correlation sentiment feedback trend is drawn using the correlation sentiment feedback value corresponding to each sentiment element.

[0083] Preferably, the public opinion triggering data can include correlation sentiment classification, correlation sentiment feedback trend and associated dialogue string.

[0084] Then, step S30 is performed to obtain the public opinion trigger source data according to the public opinion trigger data.

[0085] After obtaining the public opinion trigger data through the network public opinion monitoring system, the associated dialogue string in the public opinion trigger source data is analyzed and searched to determine the public opinion trigger source data corresponding to the public opinion trigger data, that is, to form the source data corresponding to the public opinion trigger data.

[0086] In step S30, the public opinion trigger source data is obtained according to the public opinion trigger data, which can further include:

[0087] According to the associated emotional classification, the associated dialogue string is searched to obtain question data and answer data arranged in order of the associated dialogue string;

[0088] According to the associated emotional feedback trend, the associated dialogue string corresponding to the associated emotional classification is searched to obtain initial data corresponding to the associated emotional feedback trend as the public opinion trigger source data, wherein the initial data includes one of initial question data and initial answer data.

[0089] In the process of searching for the public opinion trigger source data through the network public opinion monitoring system, first, the associated emotional classification in the public opinion trigger data is used to realize searching for all question data and answer data corresponding to the associated emotional classification in the associated dialogue string, and the question data and the answer data are arranged in order of the hierarchical order of the associated dialogue string. Then, according to the associated emotional feedback trend in the public opinion trigger data, upstream data of all question data and answer data under the associated emotional classification is searched to find initial data corresponding to the associated emotional feedback trend, such as initial question data or initial answer data, as the public opinion trigger source data. That is, the corresponding dialogue under the associated emotional feedback trend and the associated emotional classification is caused by the public opinion trigger source data, which is the source data.

[0090] It is worth noting that the source data refers to a series of public opinion data information with the associated emotional feedback trend caused by a certain public opinion topic orientation in the public opinion data, and the associated emotional feedback trend can directly reflect the further extension degree of the source data leading to discussion or public opinion fermentation.

[0091] Then, step S40 is performed to regulate the event update data of the public opinion event according to the public opinion trigger source data, the public opinion trigger data and the expected target, to generate event public relations data for publication on the whole network.

[0092] In step S40, the event update data of the public opinion event is regulated according to the public opinion trigger source data, the public opinion trigger data and the expected target to generate event public relations data for publication on the whole network, which can further include:

[0093] corresponding to the expected target, find the public opinion orientation closest to the expected orientation, and calculate the deviation of the public opinion orientation relative to the expected orientation;

[0094] corresponding to the expected target, find the public opinion orientation closest to the expected orientation, and calculate the deviation of the public opinion orientation relative to the expected orientation;

[0095] corresponding to the expected target, find the public opinion orientation closest to the expected orientation, and calculate the deviation of the public opinion orientation relative to the expected orientation;

[0096] corresponding to the expected target, find the public opinion orientation closest to the expected orientation, and calculate the deviation of the public opinion orientation relative to the expected orientation;

[0097] corresponding to the expected target, find the public opinion orientation closest to the expected orientation, and calculate the deviation of the public opinion orientation relative to the expected orientation;

[0098] corresponding to the expected target, find the public opinion orientation closest to the expected orientation, and calculate the deviation of the public opinion orientation relative to the expected orientation;

[0099] corresponding to the expected target, find the public opinion orientation closest to the expected orientation, and calculate the deviation of the public opinion orientation relative to the expected orientation;

[0100] According to the deviation degree, the expected emotional trend and the expected emotional classification, the public opinion trigger source data is adjusted to obtain the expected trigger source data. The model training can be performed on the public opinion trigger source data according to the deviation degree, the expected emotional trend and the expected emotional classification, so as to establish the trigger source prediction model. Thus, the corresponding expected trigger source data can be predicted according to the corresponding deviation degree, the expected emotional trend and the expected emotional classification.

[0101] Of course, according to the deviation degree, the expected emotional trend and the expected emotional classification, the public opinion trigger source data is adjusted to obtain the expected trigger source data. The model training can be performed on the public opinion trigger source data according to the deviation degree, the expected emotional trend and the expected emotional classification, so as to establish the trigger source prediction model. Thus, the corresponding expected trigger source data can be predicted according to the corresponding deviation degree, the expected emotional trend and the expected emotional classification.

[0102] In addition, according to the expected event data, the event update data of the public opinion event is regulated to generate event public relations data for publication on the whole network. The further regulation can include: according to the expected event data, the event update data of the public opinion event is deleted or retained to generate event public relations data for publication on the whole network. Due to different expected targets, the expected event data corresponding to the expected guidance can be deleted in the event update data to eliminate the further influence of public opinion. Of course, the expected event data can also be retained in the event update data to expand the influence range of public opinion.

[0103] Please refer to 3, the present application also provides a network public opinion monitoring system 11, comprising: an acquisition unit 111, configured to acquire network public opinion data according to a public opinion event; an extraction unit 112, configured to extract public opinion guidance from the network public opinion data to obtain public opinion trigger data; a query unit 113, configured to acquire public opinion trigger source data according to the public opinion trigger data; and a publishing unit 114, configured to regulate the event update data of the public opinion event according to the public opinion trigger source data, the public opinion trigger data and an expected target, to generate event public relations data for publication on the whole network.

[0104] It should be noted that the network public opinion monitoring system 11 provided by the above embodiments and the network public opinion monitoring method provided by the above embodiments belong to the same concept, wherein the specific operation of each module and unit has been described in detail in the method embodiments, which will not be repeated here. In actual application, the above functions can be distributed by different functional modules according to the need, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above, and this is not limited here.

[0105] Referring to Figure 4 The electronic device 1 can include a memory 12, a processor 13 and a bus, and can further include a computer program, such as a network public opinion monitoring program, stored in the memory 12 and executable on the processor 13.

[0106] The memory 12 includes at least one type of readable storage medium, such as a flash memory, a mobile hard disk, a multimedia card, a card-type memory (e.g., an SD or DX memory, etc.), a magnetic memory, a disk, an optical disk, etc. In some embodiments, the memory 12 can be an internal storage unit of the electronic device 1, such as a mobile hard disk of the electronic device 1. In other embodiments, the memory 12 can also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 1. Further, the memory 12 can include both an internal storage unit and an external storage device of the electronic device 1. The memory 12 can be used not only to store application software and various data installed on the electronic device 1, such as the code of the network public opinion monitoring, but also to temporarily store data that has been output or will be output.

[0107] The processor 13 can be composed of an integrated circuit in some embodiments, such as a single packaged integrated circuit or a plurality of packaged integrated circuits with the same or different functions, including a combination of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors and various control chips, etc. The processor 13 is the control unit of the electronic device 1, which connects various components of the entire electronic device 1 through various interfaces and lines, and executes various functions and processes data of the electronic device 1 by running or executing programs or modules (such as the network public opinion monitoring program) stored in the memory 12 and calling data stored in the memory 12.

[0108] The processor 13 executes the operating system and various application programs installed on the electronic device 1. The processor 13 executes the application programs to implement the steps in the above network public opinion monitoring method.

[0109] The computer program can be divided into one or more modules, which are stored in the memory 12 and executed by the processor 13 to complete the present application. The one or more modules can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the electronic device 1. For example, the computer program can be divided into units in the network public opinion monitoring system.

[0110] The integrated units in the form of software function modules described above can be stored in a computer readable storage medium, which can be non-volatile or volatile. The software function modules described above are stored in a storage medium, including a plurality of instructions for enabling a computer device (which can be a personal computer, a computer device, or a network device, etc.) or a processor to execute part of the network public opinion monitoring method described in various embodiments of the present application.

[0111] In summary, the network public opinion monitoring method and system disclosed by the present application can effectively improve the sufficiency of retrieval of each network platform, ensure the retrieval coverage of network public opinion data in each network platform, and quickly lock various public opinion oriented and public opinion triggering source data by extracting the public opinion oriented network public opinion data obtained. In the further regulation process of the public opinion event, the public opinion triggering source data can be converted based on the expected target to obtain expected event data corresponding to the expected target, so as to realize the corresponding adjustment of the event update data after the evolution of the public opinion event according to the expected event data, and publish the adjusted event public relations data, so as to realize the regulation of the evolution trend of the public opinion towards the expected target, such as strengthening or weakening the influence of the public opinion, so as to realize the monitoring and regulation of the development trend of the public opinion. Therefore, the present application effectively overcomes the shortcomings of the prior art and has high industrial utilization value.

[0112] The above embodiments only exemplarily illustrate the principles and effects of the present application, and are not used to limit the present application. Any person skilled in the art can modify or change the above embodiments without departing from the spirit and scope of the present application. Therefore, all equivalent modifications or changes made by those skilled in the art without departing from the spirit and technical thought disclosed by the present application should be covered by the claims of the present application.

Claims

1. A method for monitoring online public opinion, characterized in that, include: Obtain online public opinion data based on public opinion events; The online public opinion data is subjected to public opinion guidance extraction to obtain data on public opinion triggers; Based on the aforementioned public opinion triggering data, obtain the data on the source of the public opinion triggering event; Based on the source data of the public opinion incident, the data on the triggering of the public opinion incident, and the expected goals, the event update data of the public opinion event is adjusted to generate event public relations data for publication across the entire network; The online public opinion data is subjected to public opinion guidance extraction to obtain public opinion triggering data, including: Extract dialogue data from the online public opinion data, the dialogue data including question data and answer data; Based on the correspondence between the answer data and the question data, the dialogue is analyzed to obtain the associated dialogue distribution data; The associated dialogue distribution data is subjected to dialogue string extraction to obtain multiple associated dialogue strings, which include the question data and the answer data that are associated in layers. Perform sentiment analysis on the question data and answer data in the associated dialogue string to obtain sentiment classification and sentiment feedback trend, which can be used as the public opinion triggering data; Based on the aforementioned public opinion source data, the public opinion triggering data, and the expected goals, the event update data of the public opinion event is adjusted to generate event public relations data for publication across the entire network, including: The associated sentiment feedback trend, associated sentiment classification, and the data transformation of the public opinion source data are used to obtain the public opinion guidance of the public opinion source data. The public opinion guidance is compared with the expected guidance corresponding to the expected goal, the public opinion guidance that is closest to the expected guidance is found, and the deviation of the public opinion guidance from the expected guidance is calculated. Based on the deviation, the associated sentiment feedback trend and the associated sentiment classification are adjusted accordingly to obtain the expected sentiment trend and the expected sentiment classification. Based on the deviation, the expected sentiment trend, and the expected sentiment classification, the data of the public opinion triggering source is adjusted to obtain the expected triggering source data; Based on the expected source data, expected event data is generated; Based on the expected event data, the event update data of the public opinion event is adjusted to generate event public relations data for publication across the entire network.

2. The online public opinion monitoring method according to claim 1, characterized in that, Obtain online public opinion data based on public opinion events, including: Keyword extraction was performed on the aforementioned public opinion event to obtain public opinion keywords. ; Search for keywords related to public opinion. Semantically related keywords as public opinion-related terms ; For the aforementioned public opinion keywords Related keywords to the aforementioned public opinion Perform platform conversion to generate platform expressions corresponding to the platform language of the respective network platform. ; Through the aforementioned public opinion keywords The aforementioned public opinion-related keywords and the platform expression words Conduct public opinion searches on relevant online platforms to obtain the aforementioned online public opinion data.

3. The online public opinion monitoring method according to claim 2, characterized in that, For the aforementioned public opinion keywords Related keywords to the aforementioned public opinion Perform platform conversion to generate platform expressions corresponding to the platform language of the respective network platform. ,include: The aforementioned public opinion keywords Related keywords to the aforementioned public opinion Perform word segmentation to obtain a word data set. The word data set includes at least one of single-character data and word data; According to the word data set Word data The first platform terminology set corresponding to the network platform is found. ,in, This indicates a subset of terms used on the first platform. , Represents data for each word The corresponding terminology used on the first platform; Select the terminology set of the first platform Several subsets of terms used on the first platform And select multiple subsets of terms used on the first platform The corresponding first platform terminology Perform a combination, then combine it with other word data. A secondary combination is performed to generate the first platform expression terms, wherein the first combination includes at least one subset of the first platform terms. The corresponding first platform terminology ; The aforementioned public opinion keywords Related keywords to the aforementioned public opinion Perform semantic transformation to obtain semantic information; Based on the semantic information, the second platform expression term corresponding to the network platform is obtained; Based on the first platform terminology and the second platform terminology, generate platform terminology corresponding to the platform language of the respective network platform. .

4. The online public opinion monitoring method according to claim 1, characterized in that, Based on the correspondence between the answer data and the question data, the dialogue is analyzed to obtain associated dialogue distribution data, including: Calculate the correspondence between each answer data and each question data to obtain a correspondence set. ; The correspondence set Each correspondence in Correspondence threshold Comparison: When the correspondence Greater than the corresponding degree threshold When the value is greater than the corresponding degree threshold, it will be greater than the threshold value. The corresponding degree The corresponding answer data serves as the answer to the question data, thereby generating a single dialogue; The single dialogue is evaluated and categorized to determine the corresponding topic orientation. and the single dialogue and the topic orientation. correlation between ; correlation Greater than the set correlation threshold The same topic refers to The content association hierarchy is compared between the corresponding single dialogues to obtain the topic pointers for each dialogue. The relationships between individual dialogues under the corresponding dimension, wherein the relationships include hierarchical relationships and parallel hierarchical relationships; The individual dialogues corresponding to the aforementioned relationships are sorted according to their respective hierarchical arrangement to obtain the associated dialogue distribution data.

5. The online public opinion monitoring method according to claim 4, characterized in that, correlation Greater than the set correlation threshold The same topic refers to The content association hierarchy is compared between the corresponding single dialogues to obtain the topic pointers for each dialogue. The relationships between individual dialogues under the corresponding dimension include: Each of the aforementioned single dialogues will be assigned to the corresponding topic. The semantics of the dimensions are transformed to obtain the topic content; By comparing the causal relationships of the topics corresponding to each individual dialogue, the causal correlation degree between each individual dialogue is obtained. ; Based on the relevance of the single dialogue The degree of correlation Corresponding association weight and causal correlation The correlation level index was calculated. ,in, ; When the associated hierarchical indicators Greater than the indicator threshold When this happens, the hierarchical relationship is established between the corresponding single dialogues; When the associated hierarchical indicators Less than the indicator threshold When this happens, the parallel hierarchical relationship is established between the corresponding single dialogues.

6. The online public opinion monitoring method according to claim 1, characterized in that, Sentiment analysis is performed on the question and answer data in the associated dialogue strings to obtain sentiment classification and sentiment feedback trends, which are used as the public opinion triggering data, including: Sentiment detection is performed on all related dialogue strings under the same topic-pointing dimension to obtain sentiment elements and element evaluations; All the emotional elements are sorted and classified to obtain the associated emotional classification class and the emotional element set corresponding to each associated emotional classification class; The degree value of the evaluation of the corresponding emotional element. and the correlation factor of the emotional element at the level of the dialogue string. The associated emotional feedback value of the emotional element is calculated. ; The associated sentiment feedback trend is obtained based on the associated sentiment feedback value.

7. The online public opinion monitoring method according to claim 1, characterized in that, The data triggered by public opinion includes related sentiment classification, related sentiment feedback trends, and related dialogue strings; Based on the aforementioned public opinion triggering data, obtain the public opinion triggering source data, including: Based on the sentiment classification, a dialogue search is performed on the associated dialogue strings to obtain question data and answer data arranged in the order of the associated dialogue strings; Based on the associated sentiment feedback trend, upstream data search is performed on the associated dialogue strings corresponding to the associated sentiment classification to obtain the initial data corresponding to the associated sentiment feedback trend, which is used as the source data for triggering public opinion. The initial data includes one of initial question data and initial answer data.

8. A monitoring system for use in the online public opinion monitoring method according to any one of claims 1-7, characterized in that, include: The acquisition unit is used to acquire online public opinion data based on public opinion events; The extraction unit is used to perform public opinion guidance extraction on the online public opinion data to obtain public opinion triggering data. The query unit is used to obtain the source data of the public opinion incident based on the public opinion incident data; The publishing unit is used to regulate the event update data of the public opinion event based on the source data of the public opinion event, the data of the public opinion event, and the expected goals, and generate event public relations data for publication across the entire network.

Images