A method, device, equipment and storage medium for tracing the source of public opinion release

Through public opinion release traceability methods, relevance clustering and keyword database similarity analysis are used to accurately identify original data in financial public opinion data, solving the problem of public opinion traceability in the financial field and improving the accuracy of analysis and decision-making.

CN118656495BActive Publication Date: 2024-11-22HUNAN DATA IND GRP CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411144625.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-20
Publication Date
2024-11-22
Estimated Expiration
2044-08-20

AI Technical Summary

Technical Problem

It is difficult for existing technology to conduct accurate public opinion traceability analysis in the financial field, resulting in the inability of regulatory authorities to make effective decisions.

Method used

A method for tracing the public opinion release is proposed. By obtaining the public opinion data of the target information platform, it performs correlation clustering and sorting, generates the latest public opinion keyword database, and calculates its similarity with the historical public opinion keyword database, combining the maximum similarity and publishing Time, filter out original public opinion data, and use forward iteration to check the historical public opinion keyword database until the original public opinion data is screened out or the iteration termination conditions are met.

Benefits of technology

It can accurately identify illegal publication and data released through formal channels in financial public opinion data, helping financial data analysts reduce the impact of illegal data on decision-making results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118656495B_ABST
    Figure CN118656495B_ABST
Patent Text Reader

Abstract

The embodiment of the present application belongs to the field of data processing technology, and is applied to the scenario of tracing the release of public opinion data, and relates to a method, device, equipment and storage medium for tracing the release of public opinion, including obtaining target public opinion data; clustering and arranging all public opinion data by correlation; extracting keywords from the clustering results to generate the latest public opinion keyword library; calculating the similarity between the latest public opinion keyword library and all historical public opinion keyword libraries in turn; and screening the original public opinion data corresponding to the public opinion data based on the maximum similarity. The public opinion release tracing method is applied to the scenario of tracing the release of financial news media data, and can identify the illegally released financial guidance data and the financial field guidance data released by formal channels in the complex financial public opinion data through the tracing of short-term and medium-term public opinion, so as to help financial data analysts to conduct more accurate financial situation analysis and reduce the degree to which their analysis and decision-making results are affected by illegal data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and is applied to scenarios for tracing the release of public opinion data, and in particular to a method, device, equipment, and storage medium for tracing the release of public opinion data. Background Art

[0002] The Internet public opinion monitoring system, also known as the public opinion monitoring system, refers to the use of relevant professional public opinion software to capture public opinion information of concern from the complex information on the Internet according to certain rules and methods, and process it through analysis and filtering to finally present public opinion information that matches the needs.

[0003] At present, due to the popularity of personal self-media, more and more information is flooding the entire Internet, resulting in more complicated information and more arduous supervision tasks. In addition, due to the complexity of data information, it is difficult to trace the source of public opinion releases, and there is also a large amount of illegal data, which affects the analysis and decision-making of analysts. Especially in the financial field, facing the complicated financial public opinion data, if accurate public opinion tracing analysis cannot be performed, it is impossible to provide effective decision-making analysis for regulatory authorities. Summary of the invention

[0004] The purpose of the embodiments of the present application is to propose a method, device, equipment and storage medium for tracing the release of public opinion, so as to solve the problem that the existing technology is unable to perform accurate public opinion tracing analysis in the face of complex financial public opinion data.

[0005] In order to solve the above technical problems, the embodiment of the present application provides a method for tracing the source of public opinion release, which adopts the following technical solutions:

[0006] A method for tracing the source of public opinion release includes the following steps:

[0007] Step 201, obtaining all public opinion data released by the target information platform within a preset time interval;

[0008] Step 202: performing correlation clustering on all the public opinion data according to a preset correlation identification strategy to obtain a preliminary clustering result;

[0009] Step 203, extract keywords from all preliminary clustering results to generate the latest public opinion keyword library corresponding to each preliminary clustering result;

[0010] Step 204, sequentially calculating the similarity between each of the latest public opinion keyword libraries and all historical public opinion keyword libraries;

[0011] Step 205, determining, by comparison, the M historical public opinion keyword libraries with the greatest similarity corresponding to each of the latest public opinion keyword libraries, and recording the maximum similarity, where M is a positive integer;

[0012] Step 206, determining whether the maximum similarity exceeds a preset similarity threshold;

[0013] Step 207: if the maximum similarity does not exceed the preset similarity threshold, then compare the release time of all public opinion data in the preliminary clustering result corresponding to the current latest public opinion keyword library, and select the public opinion data with the earliest release time as the original public opinion data;

[0014] Step 208, if the maximum similarity exceeds the preset similarity threshold, a forward iteration method is used to perform a reverse search of the historical public opinion keyword library until the original public opinion data is found or the preset iteration termination condition is reached, and the iteration is terminated.

[0015] Furthermore, the step of clustering and arranging all the public opinion data according to the preset correlation identification strategy to obtain a preliminary clustering result specifically includes:

[0016] According to the preset correlation representation formula:

[0017] ;

[0018] All the public opinion data are clustered and sorted according to their correlation, where: represents the correlation representation value, Indicates the word repetition rate between two pairs of public opinion data. , represents the publisher coefficient, Indicates the publisher’s creative weight in the current public opinion field. Indicates the number of public opinion data in the current public opinion field published by the publisher within the specified time period. Indicates the total number of all public opinion data published by the publisher during the specified time period. The value range is in the interval between, The value range is in the interval between, ;

[0019] Determine whether there is a forwarding, reprinting, or citation relationship between the current public opinion data and the public opinion data for comparison;

[0020] If there is a forwarding, reprinting, or citation relationship, then , the current public opinion data and the public opinion data for comparison are clustered into the same cluster;

[0021] If there is no forwarding, reprinting, or citation relationship, then Method, calculating the correlation representation value between the current public opinion data and the public opinion data for comparison;

[0022] If the correlation representation value exceeds the preset correlation threshold, the current public opinion data and the public opinion data to be compared are clustered into the same cluster;

[0023] If the correlation representation value does not exceed the correlation threshold, the current public opinion data and the public opinion data for comparison are clustered into different clusters.

[0024] Furthermore, the step of extracting keywords from all preliminary clustering results to generate a public opinion keyword library corresponding to each preliminary clustering result specifically includes:

[0025] Using a natural language word segmentation model to perform word segmentation processing on the public opinion data in all preliminary clustering results, wherein the natural language word segmentation model includes an N-Gram word segmentation model;

[0026] Use the TF-IDF word weight calculation method to calculate the weight value of each word in the current preliminary clustering result;

[0027] Based on the weight values, filter out keywords in all preliminary clustering results;

[0028] According to the different preliminary clustering results, all the keywords in the preliminary clustering results are added to different preset public opinion keyword libraries.

[0029] Furthermore, after executing the step of extracting keywords from all preliminary clustering results and generating a public opinion keyword library corresponding to each preliminary clustering result, the method further includes:

[0030] Using label coding, calculate the coding distance between the public opinion keyword library corresponding to each preliminary clustering result and the public opinion keyword library corresponding to other preliminary clustering results;

[0031] Merge the public opinion keyword libraries whose coding distance is less than the preset distance threshold to obtain a new public opinion keyword library;

[0032] The public opinion data in the corresponding preliminary clustering results are merged according to the new public opinion keyword library to obtain updated clustering results.

[0033] Furthermore, the step of sequentially calculating the similarity between each of the public opinion keyword libraries and all the historical public opinion keyword libraries specifically includes:

[0034] Using the cosine similarity algorithm, the similarity between each of the public opinion keyword libraries and all the historical public opinion keyword libraries is calculated in turn;

[0035] The step of determining the M historical public opinion keyword libraries with the greatest similarity corresponding to each of the latest public opinion keyword libraries by comparison, and recording the maximum similarity, specifically includes:

[0036] Obtain the similarity between the current public opinion keyword library and all historical public opinion keyword libraries;

[0037] The similarities between the current public opinion keyword library and all historical public opinion keyword libraries are added as set elements to the preset set;

[0038] Sorting all elements in the preset set from large to small using a heap sort method, obtaining the first M elements after sorting, and generating a target set;

[0039] The first element is selected from the sorted target set and recorded.

[0040] Furthermore, before executing the step of determining whether the maximum similarity exceeds a preset similarity threshold, the method further includes:

[0041] Step 501, setting the historical public opinion keyword library with the greatest similarity to the current latest public opinion keyword library as the target word library;

[0042] Step 502, determining the generation time of the target word library by querying;

[0043] Step 503, calculating the difference between the generation time of the target word library and the current system time;

[0044] Step 504, using a preset time correlation influence formula:

[0045] ;

[0046] Calculate the public opinion influence coefficient of the public opinion data at the generation time of the target word library on the public opinion data at the current system time, where: represents the public opinion influence coefficient, Indicates the difference between the generation time of the target vocabulary and the current system time, Indicates the first set duration. Indicates the second setting duration. represents a fixed calculation factor coefficient, ;

[0047] Step 505, judging whether the public opinion influence coefficient exceeds a preset public opinion influence coefficient threshold value by comparison;

[0048] Step 506: If the public opinion impact coefficient exceeds the preset public opinion impact coefficient threshold, execute step 206;

[0049] Step 507: If the public opinion influence coefficient does not exceed the preset public opinion influence coefficient threshold, the next element is selected from the sorted target set and recorded, and the historical public opinion keyword library corresponding to the next element is updated to the target vocabulary, and steps 502 to 505 are executed until the public opinion influence coefficient exceeds the preset public opinion influence coefficient threshold, and step 206 is executed; if the screening is completed in the sorted target set and the public opinion influence coefficients corresponding to all elements do not exceed the preset public opinion influence coefficient, then the release time of all public opinion data in the preliminary clustering results corresponding to the current latest public opinion keyword library is compared, and the public opinion data with the earliest release time is selected as the original public opinion data.

[0050] Furthermore, the forward iteration method is used to perform reverse search on the historical public opinion keyword library until the original public opinion data is found or the preset iteration termination condition is reached. The iteration termination step specifically includes:

[0051] Step 601, determining the generation time of the target word library by querying, and filtering out all historical public opinion keyword libraries before the target word library based on the generation time;

[0052] Step 602, calculating the similarity between the target word library and all historical public opinion keyword libraries respectively;

[0053] Step 603, by comparison, determining the M historical public opinion keyword libraries with the greatest similarity corresponding to the target word libraries, and recording the maximum similarity;

[0054] Step 604, determining whether the maximum similarity exceeds a preset similarity threshold;

[0055] Step 605: if the maximum similarity does not exceed the similarity threshold, then compare the release time of all public opinion data in the historical preliminary clustering results corresponding to the target word library, and select the public opinion data with the earliest release time as the original public opinion data;

[0056] Step 606: If the maximum similarity exceeds the similarity threshold, the historical public opinion keyword library with the maximum similarity corresponding to the target vocabulary is updated to the new target vocabulary, and steps 601 to 604 are repeated until the original public opinion data is screened out or the preset iteration termination condition is reached, and the iteration is terminated, wherein the iteration termination condition includes that the number of iterations reaches the set value or the public opinion influence coefficients corresponding to all historical public opinion keyword libraries after the iteration do not exceed the preset public opinion influence coefficient threshold.

[0057] In order to solve the above technical problems, the embodiment of the present application also provides a public opinion release tracing device, which adopts the following technical solution:

[0058] A public opinion release tracing device, comprising:

[0059] The public opinion data acquisition module is used to obtain all public opinion data released by the target information platform within a preset time interval;

[0060] A correlation clustering module is used to perform correlation clustering on all the public opinion data according to a preset correlation identification strategy to obtain a preliminary clustering result;

[0061] The public opinion keyword library generation module is used to extract keywords from all preliminary clustering results and generate the latest public opinion keyword library corresponding to each preliminary clustering result;

[0062] A keyword library similarity calculation module is used to sequentially calculate the similarity between each of the latest public opinion keyword libraries and all historical public opinion keyword libraries;

[0063] The maximum similarity word library determination module is used to determine the M historical public opinion keyword libraries with the greatest similarity corresponding to each of the latest public opinion keyword libraries by comparison, and record the maximum similarity, where M is a positive integer;

[0064] A threshold determination module, used to determine whether the maximum similarity exceeds a preset similarity threshold;

[0065] The first branch processing module is used to compare the release time of all public opinion data in the preliminary clustering result corresponding to the current latest public opinion keyword library if the maximum similarity does not exceed the preset similarity threshold, and select the public opinion data with the earliest release time as the original public opinion data;

[0066] The second branch processing module is used to use a forward iteration method to perform a reverse search in the historical public opinion keyword library if the maximum similarity exceeds a preset similarity threshold, until original public opinion data is found or a preset iteration termination condition is reached, and the iteration is terminated.

[0067] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0068] A computer device includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor implements the steps of the above-mentioned method for tracing the source of public opinion release when executing the computer-readable instructions.

[0069] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0070] A computer-readable storage medium having computer-readable instructions stored thereon, wherein the computer-readable instructions, when executed by a processor, implement the steps of the public opinion release tracing method as described above.

[0071] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0072] The method for tracing the source of public opinion release described in the embodiment of the present application is to obtain target public opinion data; cluster and organize all public opinion data by correlation; extract keywords from the clustering results to generate the latest public opinion keyword library; calculate the similarity between the latest public opinion keyword library and all historical public opinion keyword libraries in sequence; and filter the original public opinion data corresponding to the public opinion data based on the maximum similarity. The method for tracing the source of public opinion release is applied to the scenario of tracing the source of financial news media data. It can identify the illegally released financial guidance data and the financial field guidance data released by formal channels in the complex financial public opinion data through the tracing of short- and medium-term public opinion, so as to help financial data analysts to analyze the financial situation more accurately and reduce the influence of illegal data on their analysis and decision-making results. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0074] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0075] Figure 2 It is a flowchart of an embodiment of the method for tracing the source of public opinion release according to the present application;

[0076] Figure 3 yes Figure 2 A flowchart of a specific embodiment of step 203 is shown;

[0077] Figure 4 yes Figure 2 A flowchart of a specific embodiment of step 205 is shown;

[0078] Figure 5 It is a flowchart of a specific embodiment of analyzing the influence of the interval time dimension of public opinion release on the public opinion tracing in the public opinion release tracing method described in this application;

[0079] Figure 6 It is a flowchart of a specific embodiment of tracing the source of public opinion data by forward iterative reverse query in the public opinion release tracing method described in this application;

[0080] Figure 7 It is a structural diagram of an embodiment of a public opinion release tracing device according to the present application;

[0081] Figure 8 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0082] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0083] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0084] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0085] like Figure 1As shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0086] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0087] Terminal devices 101, 102, 103 can be various electronic devices with display screens and supporting web browsing, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III), MP4 (Moving Picture Experts Group Audio Layer IV), laptop computers, desktop computers, etc.

[0088] The server 105 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal devices 101 , 102 , and 103 .

[0089] It should be noted that the public opinion release tracing method provided in the embodiment of the present application is generally executed by a server, and accordingly, the public opinion release tracing device is generally set in the server.

[0090] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0091] Continue to refer Figure 2 , shows a flow chart of an embodiment of a method for tracing the source of public opinion release according to the present application. The method for tracing the source of public opinion release includes the following steps:

[0092] Step 201, obtaining all public opinion data released by the target information platform within a preset time interval.

[0093] In this embodiment, the target information platform includes a financial news media publishing platform, and also includes an information browsing platform frequently used by users, such as an information publishing platform built within the company. The preset time interval is, for example, 1 minute, 1 hour, 1 day, etc. The public opinion data can be the public opinion data of the entire field, or only the public opinion data of certain specific fields, for example, only the public opinion data of the financial field is obtained. The target information platform, the preset time interval, and the field limitation of the public opinion data acquisition can all be freely set by the public opinion data acquisition party according to actual business needs. By obtaining all the public opinion data released by the target information platform within the preset time interval, it can help analysis and decision-making personnel to understand the dynamic changes of the industry more comprehensively and timely.

[0094] Step 202: According to a preset correlation identification strategy, all the public opinion data are clustered and sorted according to their correlation to obtain a preliminary clustering result.

[0095] By identifying the correlation between public opinion data, all the public opinion data are clustered and organized according to their correlation, so as to facilitate the subsequent auxiliary analysis and decision-making personnel to conduct a comprehensive analysis of the industry situation.

[0096] In this embodiment, the step of performing correlation clustering on all public opinion data according to a preset correlation identification strategy to obtain a preliminary clustering result specifically includes:

[0097] According to the preset correlation representation formula:

[0098] ;

[0099] All the public opinion data are clustered and sorted according to their correlation, where: represents the correlation representation value, Indicates the word repetition rate between two pairs of public opinion data. , represents the publisher coefficient, Indicates the publisher’s creative weight in the current public opinion field. Indicates the number of public opinion data in the current public opinion field published by the publisher within the specified time period. Indicates the total number of all public opinion data published by the publisher during the specified time period. The value range is in the interval between, The value range is in the interval between, , wherein, assuming that two pieces of public opinion data are being compared, one piece is the current public opinion data, and the other piece is the public opinion data to be compared, the publisher represents the publisher of the public opinion data to be compared;

[0100] Determine whether there is a forwarding, reprinting, or citation relationship between the current public opinion data and the public opinion data for comparison;

[0101] If there is a forwarding, reprinting, or citation relationship, then , the current public opinion data and the public opinion data for comparison are clustered into the same cluster;

[0102] In this embodiment, the judgment method of whether there is a forwarding, reprinting, or citing relationship can be performed by traversing the current public opinion data and the public opinion data to be compared to check whether there are corresponding key fields between the two. For example, if it is reprinted, it will be marked as "reprinted to". Similarly, forwarding and citing can also be identified by specific characters or key fields. For the public opinion data of the above relationship, it is assumed that the two are associated, and by assigning a value of 1, it is convenient to cluster and organize the association of all the public opinion data.

[0103] If there is no forwarding, reprinting, or citation relationship, then Method, calculating the correlation representation value between the current public opinion data and the public opinion data for comparison;

[0104] In this embodiment, for public opinion data that do not have a direct forwarding, reprinting, or citation relationship, the word repetition rate and the publisher coefficient of the compared public opinion data are summed together to obtain a correlation representation value, and then combined with a preset correlation threshold, it is determined whether the current public opinion data and the compared public opinion data are clustered into the same cluster cluster, thereby ensuring the correlation calculation between public opinion numbers that do not have a direct forwarding, reprinting, or citation relationship, and making the clustering more scientific.

[0105] If the correlation representation value exceeds the preset correlation threshold, the current public opinion data and the public opinion data to be compared are clustered into the same cluster;

[0106] If the correlation representation value does not exceed the correlation threshold, the current public opinion data and the public opinion data for comparison are clustered into different clusters.

[0107] Step 203: extract keywords from all preliminary clustering results to generate the latest public opinion keyword library corresponding to each preliminary clustering result.

[0108] Specifically, when tracing the source of public opinion data, since the entire piece or article of public opinion data contains a large number of words, keyword extraction and the construction of a public opinion keyword library are adopted. This allows the subsequent tracing of public opinion to be done only by comparing the public opinion keyword libraries, thus reducing the amount of data compared during tracing. At the same time, it also reduces the consumption of computing resources and improves the efficiency of tracing and comparison.

[0109] Continue to refer Figure 3 , Figure 3yes Figure 2 The flowchart of a specific embodiment of step 203 shown includes:

[0110] Step 301, using a natural language word segmentation model to perform word segmentation processing on the public opinion data in all preliminary clustering results, wherein the natural language word segmentation model includes an N-Gram word segmentation model;

[0111] Step 302, using the TF-IDF word weight calculation method to calculate the weight value of each word in the current preliminary clustering result;

[0112] Step 303: based on the weight value, filter out the keywords in all the preliminary clustering results.

[0113] Specifically, the words whose weight values ​​exceed the preset weight threshold are selected as keywords;

[0114] Step 304, according to the difference of the preliminary clustering results, the keywords in all the preliminary clustering results are added to different preset public opinion keyword libraries.

[0115] Through the natural language word segmentation model and the TF-IDF word weight calculation method, the weight value of each word segmentation in the corresponding preliminary clustering result is calculated, and the keywords are screened out through the weight value and the preset weight threshold, so as to construct the public opinion keyword library in a more scientific way.

[0116] In addition, the preset weight threshold may not be set, and the weight values ​​of all the segmented words may be directly compared to select the first N segmented words with the largest weight values ​​as the keywords corresponding to the preliminary clustering result, where N is a positive integer.

[0117] Specifically, according to the different preliminary clustering results, the keywords in all preliminary clustering results are added to different preset public opinion keyword libraries respectively. For example, 20 clusters are generated during the preliminary clustering, and the clustered public opinion data corresponding to these 20 clusters are 20 preliminary clustering results. Keywords are extracted for each preliminary clustering result respectively, and the keywords are added to the 20 public opinion keyword libraries accordingly according to the different clustering clusters.

[0118] In this embodiment, after executing the step of extracting keywords from all preliminary clustering results and generating a public opinion keyword library corresponding to each preliminary clustering result, the method also includes: using a label encoding method to calculate the encoding distance between the public opinion keyword library corresponding to each preliminary clustering result and the public opinion keyword library corresponding to other preliminary clustering results; merging the public opinion keyword libraries whose encoding distance is less than a preset distance threshold to obtain a new public opinion keyword library; and merging the public opinion data in the corresponding preliminary clustering results according to the new public opinion keyword library to obtain an updated clustering result.

[0119] In essence, after initially constructing different public opinion keyword libraries, label encoding was used to perform a secondary merger on the same or similar public opinion keyword libraries, thus avoiding the situation where the initial clustering results were incomplete during the initial clustering and minimizing the occurrence of excessive construction of public opinion keyword libraries.

[0120] Step 204, calculate the similarity between each of the latest public opinion keyword libraries and all historical public opinion keyword libraries in turn.

[0121] In this embodiment, the step of sequentially calculating the similarity between each of the public opinion keyword libraries and all the historical public opinion keyword libraries specifically includes: using a cosine similarity algorithm to sequentially calculate the similarity between each of the public opinion keyword libraries and all the historical public opinion keyword libraries.

[0122] In this embodiment, a label encoding method may also be used to sequentially calculate the similarity between each public opinion keyword library and all historical public opinion keyword libraries. Specifically, when the label encoding method is used for calculation, the smaller the encoding distance, the greater the similarity between the two public opinion keyword libraries.

[0123] Step 205, by comparison, determine the M historical public opinion keyword libraries with the greatest similarity corresponding to each of the latest public opinion keyword libraries, and record the maximum similarity, where M is a positive integer.

[0124] Continue to refer Figure 4 , Figure 4 yes Figure 2 The flowchart of a specific embodiment of step 205 shown includes:

[0125] Step 401, obtaining the similarity between the current public opinion keyword library and all historical public opinion keyword libraries;

[0126] Step 402, adding the similarities between the current public opinion keyword library and all historical public opinion keyword libraries as set elements into a preset set;

[0127] Step 403, sorting all elements in the preset set from large to small using a heap sort method, obtaining the first M elements after sorting, and generating a target set;

[0128] Step 404: Filter out the first element from the sorted target set and record it.

[0129] In this embodiment, the M historical public opinion keyword libraries with the largest similarity corresponding to each of the latest public opinion keyword libraries are determined by comparison, and the step of recording the maximum similarity can also be: obtaining the similarity between the current public opinion keyword library and all historical public opinion keyword libraries; adding the similarity between the current public opinion keyword library and all historical public opinion keyword libraries as set elements to a preset set; sorting all elements in the target set from large to small using a bubble sort method, obtaining the first M elements after sorting, and generating a target set; filtering out the first element from the sorted target set and recording it.

[0130] In this embodiment, after obtaining the similarity between the current public opinion keyword library and all historical public opinion keyword libraries, a heap sort or bubble sort method is used to screen out the historical public opinion keyword library with the greatest similarity to the current public opinion keyword library. Since when a public opinion event occurs, there is often a situation where the network water army conducts short-term speculation. Therefore, the historical public opinion keyword library with the greatest similarity to the current public opinion keyword library is screened out to identify whether the newly released public opinion data corresponding to the current public opinion keyword library is the continued fermentation of a certain historical public opinion event, which is helpful for public opinion tracing from the perspective of time series relationship. In the case of a large number of sorting elements, it is faster to use heap sort to screen out the first M elements than bubble sort, and in the case of a small number of sorting elements, bubble sort is more applicable than heap sort. Therefore, in the use priority of heap sort and bubble sort, a reference value can be set here, for example: when the number of historical public opinion keyword libraries does not exceed a certain value, bubble sort is preferred, and when the number of historical public opinion keyword libraries exceeds the value, heap sort is preferred.

[0131] Step 206: determine whether the maximum similarity exceeds a preset similarity threshold.

[0132] In this embodiment, before executing the step of determining whether the maximum similarity exceeds a preset similarity threshold, the method also includes: calculating, based on the difference between the generation time of the historical public opinion keyword library and the current system time, the public opinion influence coefficient of the public opinion data corresponding to the historical public opinion keyword library on the current latest public opinion data released; in combination with actual public opinion events, the longer the time interval, the less likely it is that the latest public opinion data is released on the basis of previous public opinion data, that is, the less likely it is that it is the same public opinion event.

[0133] Continue to refer Figure 5 , Figure 5 It is a flowchart of a specific embodiment of analyzing the influence of the interval time of public opinion release on the public opinion tracing in the public opinion release tracing method described in this application, including:

[0134] Step 501, setting the historical public opinion keyword library with the greatest similarity to the current latest public opinion keyword library as the target word library;

[0135] Step 502, determining the generation time of the target word library by querying;

[0136] Step 503, calculating the difference between the generation time of the target word library and the current system time;

[0137] Step 504, using a preset time correlation influence formula:

[0138] ;

[0139] Calculate the public opinion influence coefficient of the public opinion data at the generation time of the target word library on the public opinion data at the current system time, where: represents the public opinion influence coefficient, Indicates the difference between the generation time of the target vocabulary and the current system time, Indicates the first set duration. Indicates the second setting duration. represents a fixed calculation factor coefficient, ;

[0140] Step 505, judging whether the public opinion influence coefficient exceeds a preset public opinion influence coefficient threshold value by comparison;

[0141] Step 506: If the public opinion impact coefficient exceeds the preset public opinion impact coefficient threshold, execute step 206;

[0142] Step 507: If the public opinion influence coefficient does not exceed the preset public opinion influence coefficient threshold, the next element is selected from the sorted target set and recorded, and the historical public opinion keyword library corresponding to the next element is updated to the target vocabulary, and steps 502 to 505 are executed until the public opinion influence coefficient exceeds the preset public opinion influence coefficient threshold, and step 206 is executed; if the screening is completed in the sorted target set and the public opinion influence coefficients corresponding to all elements do not exceed the preset public opinion influence coefficient, then the release time of all public opinion data in the preliminary clustering results corresponding to the current latest public opinion keyword library is compared, and the public opinion data with the earliest release time is selected as the original public opinion data.

[0143] By calculating the public opinion impact coefficient of the public opinion data at the time of generation of the target word library to the public opinion data at the current system time, the continuous public opinion events in the short and medium term can be traced back to the source, which is more helpful to maintain the network security environment, timely screen out the source of the public opinion events, and avoid the continuous fermentation of public opinion events. The recording includes recording the elements in the set, that is, the similarity, and also includes recording the distinguishing identification information of the historical public opinion keyword library corresponding to the similarity. Each time the next element is screened out and recorded, it can facilitate the next screening by recording information, and also preserve the traces of public opinion tracing, which is convenient for public opinion tracing personnel to view and archive.

[0144] Step 207, if the maximum similarity does not exceed the preset similarity threshold, then compare the release time of all public opinion data in the preliminary clustering result corresponding to the current latest public opinion keyword library, and select the public opinion data with the earliest release time as the original public opinion data.

[0145] Specifically, if the maximum similarity does not exceed the preset similarity threshold, it means that there is no previous public opinion event released in the historical public opinion keyword library. Therefore, according to the release time of all public opinion data in the preliminary clustering results corresponding to the current latest public opinion keyword library, a comparison is made, and the public opinion data with the earliest release time is screened out as the original public opinion data. This is convenient for network management personnel to trace the source of public opinion events, and also facilitates them to identify original events, which helps to establish a standardized Internet speech or publishing environment.

[0146] Step 208, if the maximum similarity exceeds the preset similarity threshold, a forward iteration method is used to perform a reverse search of the historical public opinion keyword library until the original public opinion data is found or the preset iteration termination condition is reached, and the iteration is terminated.

[0147] Continue to refer Figure 6 , Figure 6 It is a flowchart of a specific embodiment of tracing the source of public opinion data by forward iterative reverse query in the public opinion release tracing method described in this application, including:

[0148] Step 601, determining the generation time of the target word library by querying, and filtering out all historical public opinion keyword libraries before the target word library based on the generation time;

[0149] Step 602, calculating the similarity between the target word library and all historical public opinion keyword libraries respectively;

[0150] Step 603, by comparison, determining the M historical public opinion keyword libraries with the greatest similarity corresponding to the target word libraries, and recording the maximum similarity;

[0151] Step 604, determining whether the maximum similarity exceeds a preset similarity threshold;

[0152] Step 605: if the maximum similarity does not exceed the similarity threshold, then compare the release time of all public opinion data in the historical preliminary clustering results corresponding to the target word library, and select the public opinion data with the earliest release time as the original public opinion data;

[0153] Step 606: If the maximum similarity exceeds the similarity threshold, the historical public opinion keyword library with the maximum similarity corresponding to the target vocabulary is updated to the new target vocabulary, and steps 601 to 604 are repeated until the original public opinion data is screened out or the preset iteration termination condition is reached, and the iteration is terminated, wherein the iteration termination condition includes that the number of iterations reaches the set value or the public opinion influence coefficients corresponding to all historical public opinion keyword libraries after the iteration do not exceed the preset public opinion influence coefficient threshold.

[0154] In this embodiment, the step of updating the historical public opinion keyword library with the greatest similarity corresponding to the target keyword library to a new target keyword library includes: updating the target keyword library to the current public opinion keyword library; executing step 204 at the generation time of the current public opinion keyword library to obtain the similarity between the current public opinion keyword library and all historical public opinion keyword libraries respectively; and then executing step 205 to filter out the historical public opinion keyword library with the greatest similarity corresponding to the current public opinion keyword library and update it as a new target keyword library. Specifically, each time the historical public opinion keyword library with the greatest similarity corresponding to the target keyword library is updated to the new target keyword library, it is an operation of forward iterative reverse query of the historical public opinion keyword library, and the forward iterative reverse query is essentially to continuously reverse the historical public opinion keyword library to find the earliest historical public opinion keyword library related to the public opinion data.

[0155] In this embodiment, the preset iteration termination condition is reached, including the number of iterations reaching the set value and the public opinion influence coefficients corresponding to all historical public opinion keyword libraries after the iteration do not exceed the preset public opinion influence coefficient threshold. Correspondingly, the more iterations there are, the earlier the screened historical public opinion keyword library is in the time dimension. Therefore, by setting the termination condition of the number of iterations, the screened original public opinion data is moderate in the time of public opinion tracing, neither too early nor too late. If the public opinion influence coefficients corresponding to all historical public opinion keyword libraries after the iteration do not exceed the preset public opinion influence coefficient threshold, it means that all historical public opinion keyword libraries after the iteration are of little significance to the public opinion tracing, or it can be understood that all historical public opinion keyword libraries after the iteration have little impact on the current public opinion tracing. Therefore, the iteration is terminated and there is no need to obtain earlier public opinion data.

[0156] Correspondingly, when the number of iterations reaches the set value, the iteration is terminated, and the target vocabulary at the time of iteration termination is determined. According to the release time of all public opinion data in the preliminary clustering results corresponding to the target vocabulary, a comparison is made to screen out the public opinion data with the earliest release time as the original public opinion data.

[0157] Correspondingly, when the public opinion influence coefficient corresponding to all historical public opinion keyword libraries after the iteration does not exceed the preset public opinion influence coefficient threshold, the iteration is terminated, and the target vocabulary library at the time of iteration termination is determined. According to the release time of all public opinion data in the preliminary clustering results corresponding to the target vocabulary library, a comparison is performed to screen out the public opinion data with the earliest release time as the original public opinion data.

[0158] Through forward iteration and reverse checking, it is possible to trace the source of public opinion release for continuous public opinion events in the medium and short term, which is more helpful in maintaining the network security environment, timely screening out the source of public opinion events, and locating the relevant responsible persons.

[0159] In this application, the public opinion release tracing method can be applied to the release tracing scenario of financial news media data. By tracing the source of medium- and short-term public opinion, it can identify illegally released financial guidance data and financial field guidance data released through formal channels in complex financial public opinion data, helping financial data analysts to conduct financial situation analysis more accurately and reduce the degree to which their analysis and decision-making results are affected by illegal data.

[0160] This application obtains the target public opinion data; clusters and organizes all public opinion data by correlation; extracts keywords from the clustering results to generate the latest public opinion keyword library; calculates the similarity between the latest public opinion keyword library and all historical public opinion keyword libraries in sequence; and filters the original public opinion data corresponding to the public opinion data based on the maximum similarity. The public opinion release tracing method is applied to the release tracing scenario of financial news media data. It can identify the illegally released financial guidance data and the financial field guidance data released by formal channels in the complex financial public opinion data through the tracing of short-term and medium-term public opinion, helping financial data analysts to conduct more accurate financial situation analysis and reduce the degree to which their analysis and decision-making results are affected by illegal data.

[0161] In the embodiment of the present application, the target public opinion data is obtained; all public opinion data are clustered and sorted according to their relevance; keywords are extracted from the clustering results to generate the latest public opinion keyword library; the similarity between the latest public opinion keyword library and all historical public opinion keyword libraries is calculated in turn; and the original public opinion data corresponding to the public opinion data is screened based on the maximum similarity. The public opinion release tracing method is applied to the release tracing scenario of financial news media data. Through the tracing of short- and medium-term public opinion, it is possible to identify illegally released financial guidance data and financial field guidance data released through formal channels in complex financial public opinion data, helping financial data analysts to conduct more accurate financial situation analysis and reduce the degree to which their analysis and decision-making results are affected by illegal data.

[0162] Further references Figure 7 , as a response to the above Figure 2 The present application provides an embodiment of a device for tracing the source of public opinion release. Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0163] like Figure 7 As shown, the public opinion release tracing device 700 described in this embodiment includes: a public opinion data acquisition module 701, a correlation clustering module 702, a public opinion keyword library generation module 703, a keyword library similarity calculation module 704, a maximum similarity word library determination module 705, a threshold judgment module 706, a first branch processing module 707 and a second branch processing module 708. Among them:

[0164] The public opinion data acquisition module 701 is used to acquire all public opinion data released by the target information platform within a preset time interval;

[0165] The relevance clustering module 702 is used to perform relevance clustering on all the public opinion data according to a preset relevance identification strategy to obtain a preliminary clustering result;

[0166] The public opinion keyword library generation module 703 is used to extract keywords from all preliminary clustering results and generate the latest public opinion keyword library corresponding to each preliminary clustering result;

[0167] The keyword library similarity calculation module 704 is used to sequentially calculate the similarity between each of the latest public opinion keyword libraries and all historical public opinion keyword libraries;

[0168] The maximum similarity word library determination module 705 is used to determine the M historical public opinion keyword libraries with the greatest similarity corresponding to each of the latest public opinion keyword libraries by comparison, and record the maximum similarity, where M is a positive integer;

[0169] A threshold determination module 706 is used to determine whether the maximum similarity exceeds a preset similarity threshold;

[0170] The first branch processing module 707 is used to compare the release time of all public opinion data in the preliminary clustering result corresponding to the current latest public opinion keyword library if the maximum similarity does not exceed the preset similarity threshold, and select the public opinion data with the earliest release time as the original public opinion data;

[0171] The second branch processing module 708 is used to use a forward iteration method to perform a reverse search in the historical public opinion keyword library if the maximum similarity exceeds a preset similarity threshold, until the original public opinion data is found or the preset iteration termination condition is reached, and the iteration is terminated.

[0172] This application obtains the target public opinion data; clusters and organizes all public opinion data by correlation; extracts keywords from the clustering results to generate the latest public opinion keyword library; calculates the similarity between the latest public opinion keyword library and all historical public opinion keyword libraries in sequence; and filters the original public opinion data corresponding to the public opinion data based on the maximum similarity. The public opinion release tracing method is applied to the release tracing scenario of financial news media data. It can identify the illegally released financial guidance data and the financial field guidance data released by formal channels in the complex financial public opinion data through the tracing of short-term and medium-term public opinion, helping financial data analysts to conduct more accurate financial situation analysis and reduce the degree to which their analysis and decision-making results are affected by illegal data.

[0173] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing related hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0174] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0175] To solve the above technical problems, the present application also provides a computer device. Figure 8 , Figure 8 This is a basic structural block diagram of the computer device in this embodiment.

[0176] The computer device 8 includes a memory 8a, a processor 8b, and a network interface 8c which are interconnected through a system bus. It should be noted that: Figure 8 Only a computer device 8 having a memory 8a, a processor 8b, and a network interface 8c is shown, but it should be understood that it is not required to implement all the components shown, and more or fewer components may be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), digital signal processors (DSP), embedded devices, etc.

[0177] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0178] The memory 8a includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 8a can be an internal storage unit of the computer device 8, such as a hard disk or memory of the computer device 8. In other embodiments, the memory 8a can also be an external storage device of the computer device 8, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), etc. equipped on the computer device 8. Of course, the memory 8a can also include both the internal storage unit of the computer device 8 and its external storage device. In this embodiment, the memory 8a is generally used to store the operating system and various application software installed on the computer device 8, such as a computer-readable instruction of a method for tracing the source of public opinion release, etc. In addition, the memory 8a can also be used to temporarily store various types of data that have been output or are to be output.

[0179] In some embodiments, the processor 8b may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other text clustering chips. The processor 8b is generally used to control the overall operation of the computer device 8. In this embodiment, the processor 8b is used to run computer-readable instructions stored in the memory 8a or process data, such as computer-readable instructions for running the public opinion release tracing method.

[0180] The network interface 8c may include a wireless network interface or a wired network interface. The network interface 8c is generally used to establish a communication connection between the computer device 8 and other electronic devices.

[0181] The computer device proposed in this embodiment belongs to the field of data processing technology and is applied to the scenario of tracing the release of public opinion data. This application obtains the target public opinion data; clusters and organizes all public opinion data by correlation; extracts keywords from the clustering results to generate the latest public opinion keyword library; sequentially calculates the similarity between the latest public opinion keyword library and all historical public opinion keyword libraries; and filters the original public opinion data corresponding to the public opinion data based on the maximum similarity. The public opinion release tracing method is applied to the release tracing scenario of financial news media data. By tracing the source of short- and medium-term public opinion, it is possible to identify illegally released financial guidance data and financial field guidance data released through formal channels in complex financial public opinion data, helping financial data analysts to conduct more accurate financial situation analysis and reduce the degree to which their analysis and decision-making results are affected by illegal data.

[0182] The present application also provides another implementation mode, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by a processor so that the processor performs the steps of the public opinion release tracing method as described above.

[0183] The computer-readable storage medium proposed in this embodiment belongs to the field of data processing technology and is applied to the scenario of tracing the release of public opinion data. This application obtains the target public opinion data; clusters and organizes all public opinion data by correlation; extracts keywords from the clustering results to generate the latest public opinion keyword library; sequentially calculates the similarity between the latest public opinion keyword library and all historical public opinion keyword libraries; and filters the original public opinion data corresponding to the public opinion data based on the maximum similarity. The public opinion release tracing method is applied to the release tracing scenario of financial news media data. It can identify the illegally released financial guidance data and the financial field guidance data released by formal channels in the complex financial public opinion data through the tracing of short-term and medium-term public opinion, and help financial data analysts to conduct more accurate financial situation analysis and reduce the degree to which their analysis and decision-making results are affected by illegal data.

[0184] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0185] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A method for tracing the source of public opinion release, characterized in that: The steps include: Step 201, obtaining all public opinion data released by the target information platform within a preset time interval; Step 202, according to a preset correlation identification strategy, all the public opinion data are clustered and sorted by correlation to obtain a preliminary clustering result, wherein the step of clustering and sorting all the public opinion data by correlation to obtain a preliminary clustering result according to a preset correlation identification strategy specifically includes: According to the preset correlation representation formula: According to the preset correlation representation formula: , All the public opinion data are clustered and sorted according to their correlation, where: represents the correlation representation value, Indicates the word repetition rate between two pairs of public opinion data. , represents the publisher coefficient, Indicates the publisher’s creative weight in the current public opinion field. Indicates the number of public opinion data in the current public opinion field published by the publisher within the specified time period. Indicates the total number of all public opinion data published by the publisher within the specified time period. The value range is in the interval between, The value range is in the interval between, ; Determine whether there is a forwarding, reprinting, or citation relationship between the current public opinion data and the public opinion data for comparison; If there is a forwarding, reprinting, or citation relationship, then , the current public opinion data and the public opinion data for comparison are clustered into the same cluster; If there is no forwarding, reprinting, or citation relationship, then Method, calculating the correlation representation value between the current public opinion data and the public opinion data for comparison; If the correlation representation value exceeds the preset correlation threshold, the current public opinion data and the public opinion data to be compared are clustered into the same cluster; If the correlation representation value does not exceed the correlation threshold, the current public opinion data and the public opinion data to be compared are clustered into different clusters; Step 203, extract keywords from all preliminary clustering results to generate the latest public opinion keyword library corresponding to each preliminary clustering result; Step 204, sequentially calculating the similarity between each of the latest public opinion keyword libraries and all historical public opinion keyword libraries; Step 205, determining, by comparison, the M historical public opinion keyword libraries with the greatest similarity corresponding to each of the latest public opinion keyword libraries, and recording the maximum similarity, where M is a positive integer; Step 206, determining whether the maximum similarity exceeds a preset similarity threshold, wherein before executing the step of determining whether the maximum similarity exceeds a preset similarity threshold, the method further includes: Step 501, setting the historical public opinion keyword library with the greatest similarity to the current latest public opinion keyword library as the target word library; Step 502, determining the generation time of the target word library by querying; Step 503, calculating the difference between the generation time of the target word library and the current system time; Step 504, using a preset time correlation influence formula: , Calculate the public opinion influence coefficient of the public opinion data at the generation time of the target word library on the public opinion data at the current system time, where: represents the public opinion influence coefficient, Indicates the difference between the generation time of the target vocabulary and the current system time, Indicates the first set duration. Indicates the second setting duration. represents a fixed calculation factor coefficient, ; Step 505, judging whether the public opinion influence coefficient exceeds a preset public opinion influence coefficient threshold value by comparison; Step 506: If the public opinion impact coefficient exceeds the preset public opinion impact coefficient threshold, execute step 206; Step 507: if the public opinion influence coefficient does not exceed the preset public opinion influence coefficient threshold, the next element is selected from the sorted target set and recorded, and the historical public opinion keyword library corresponding to the next element is updated as the target word library, and steps 502 to 505 are executed until the public opinion influence coefficient exceeds the preset public opinion influence coefficient threshold, and then step 206 is executed; if the screening is completed in the sorted target set and the public opinion influence coefficients corresponding to all elements do not exceed the preset public opinion influence coefficient, then the release time of all public opinion data in the preliminary clustering result corresponding to the current latest public opinion keyword library is compared, and the public opinion data with the earliest release time is selected as the original public opinion data; Step 207: if the maximum similarity does not exceed the preset similarity threshold, then compare the release time of all public opinion data in the preliminary clustering result corresponding to the current latest public opinion keyword library, and select the public opinion data with the earliest release time as the original public opinion data; Step 208, if the maximum similarity exceeds the preset similarity threshold, a forward iteration method is used to perform a reverse search of the historical public opinion keyword library until the original public opinion data is found or the preset iteration termination condition is reached, and the iteration is terminated.

2. The method for tracing the source of public opinion release according to claim 1 is characterized in that: The step of extracting keywords from all preliminary clustering results and generating a public opinion keyword library corresponding to each preliminary clustering result specifically includes: Using a natural language word segmentation model to perform word segmentation processing on the public opinion data in all preliminary clustering results, wherein the natural language word segmentation model includes an N-Gram word segmentation model; Use the TF-IDF word weight calculation method to calculate the weight value of each word in the current preliminary clustering result; Based on the weight values, filter out keywords in all preliminary clustering results; According to the different preliminary clustering results, all the keywords in the preliminary clustering results are added to different preset public opinion keyword libraries.

3. The method for tracing the source of public opinion release according to claim 1 or 2, characterized in that: After executing the step of extracting keywords from all preliminary clustering results and generating a public opinion keyword library corresponding to each preliminary clustering result, the method further includes: Using label coding, calculate the coding distance between the public opinion keyword library corresponding to each preliminary clustering result and the public opinion keyword library corresponding to other preliminary clustering results; Merge the public opinion keyword libraries whose coding distance is less than the preset distance threshold to obtain a new public opinion keyword library; The public opinion data in the corresponding preliminary clustering results are merged according to the new public opinion keyword library to obtain updated clustering results.

4. The method for tracing the source of public opinion release according to claim 1 is characterized in that: The step of sequentially calculating the similarity between each of the public opinion keyword libraries and all the historical public opinion keyword libraries specifically includes: Using the cosine similarity algorithm, the similarity between each of the public opinion keyword libraries and all the historical public opinion keyword libraries is calculated in turn; The step of determining the M historical public opinion keyword libraries with the greatest similarity corresponding to each of the latest public opinion keyword libraries by comparison, and recording the maximum similarity, specifically includes: Obtain the similarity between the current public opinion keyword library and all historical public opinion keyword libraries; The similarities between the current public opinion keyword library and all historical public opinion keyword libraries are added as set elements to the preset set; Sorting all elements in the preset set from large to small using a heap sort method, obtaining the first M elements after sorting, and generating a target set; The first element is selected from the sorted target set and recorded.

5. The method for tracing the source of public opinion release according to claim 1 is characterized in that: The forward iteration method is used to reversely search the historical public opinion keyword library until the original public opinion data is reversed or the preset iteration termination condition is reached. The iteration termination step specifically includes: Step 601, determining the generation time of the target word library by querying, and filtering out all historical public opinion keyword libraries before the target word library based on the generation time; Step 602, calculating the similarity between the target word library and all historical public opinion keyword libraries respectively; Step 603, by comparison, determining the M historical public opinion keyword libraries with the greatest similarity corresponding to the target word libraries, and recording the maximum similarity; Step 604, determining whether the maximum similarity exceeds a preset similarity threshold; Step 605: if the maximum similarity does not exceed the similarity threshold, then compare the release time of all public opinion data in the historical preliminary clustering results corresponding to the target word library, and select the public opinion data with the earliest release time as the original public opinion data; Step 606: If the maximum similarity exceeds the similarity threshold, the historical public opinion keyword library with the maximum similarity corresponding to the target vocabulary is updated to the new target vocabulary, and steps 601 to 604 are repeated until the original public opinion data is screened out or the preset iteration termination condition is reached, and the iteration is terminated, wherein the iteration termination condition includes that the number of iterations reaches the set value or the public opinion influence coefficients corresponding to all historical public opinion keyword libraries after the iteration do not exceed the preset public opinion influence coefficient threshold.

6. A device for tracing the source of public opinion release, characterized in that: include: The public opinion data acquisition module is used to obtain all public opinion data released by the target information platform within a preset time interval; The correlation clustering module is used to perform correlation clustering on all the public opinion data according to a preset correlation identification strategy to obtain a preliminary clustering result, wherein the step of performing correlation clustering on all the public opinion data according to a preset correlation identification strategy to obtain a preliminary clustering result specifically includes: According to the preset correlation representation formula: , All the public opinion data are clustered and sorted according to their correlation, where: represents the correlation representation value, Indicates the word repetition rate between two pairs of public opinion data. , represents the publisher coefficient, Indicates the publisher’s creative weight in the current public opinion field. Indicates the number of public opinion data in the current public opinion field published by the publisher within the specified time period. Indicates the total number of all public opinion data published by the publisher within the specified time period. The value range is in the interval between, The value range is in the interval between, ; Determine whether there is a forwarding, reprinting, or citation relationship between the current public opinion data and the public opinion data for comparison; If there is a forwarding, reprinting, or citation relationship, then , the current public opinion data and the public opinion data for comparison are clustered into the same cluster; If there is no forwarding, reprinting, or citation relationship, then Method, calculating the correlation representation value between the current public opinion data and the public opinion data for comparison; If the correlation representation value exceeds the preset correlation threshold, the current public opinion data and the public opinion data to be compared are clustered into the same cluster; If the correlation representation value does not exceed the correlation threshold, the current public opinion data and the public opinion data to be compared are clustered into different clusters; The public opinion keyword library generation module is used to extract keywords from all preliminary clustering results and generate the latest public opinion keyword library corresponding to each preliminary clustering result; A keyword library similarity calculation module is used to sequentially calculate the similarity between each of the latest public opinion keyword libraries and all historical public opinion keyword libraries; The maximum similarity word library determination module is used to determine the M historical public opinion keyword libraries with the greatest similarity corresponding to each of the latest public opinion keyword libraries by comparison, and record the maximum similarity, where M is a positive integer; A threshold judgment module is used to judge whether the maximum similarity exceeds a preset similarity threshold, wherein before executing the step of judging whether the maximum similarity exceeds the preset similarity threshold, it also includes: Step 501, setting the historical public opinion keyword library with the greatest similarity to the current latest public opinion keyword library as the target word library; Step 502, determining the generation time of the target word library by querying; Step 503, calculating the difference between the generation time of the target word library and the current system time; Step 504, using a preset time correlation influence formula: , Calculate the public opinion influence coefficient of the public opinion data at the generation time of the target word library on the public opinion data at the current system time, where: represents the public opinion influence coefficient, Indicates the difference between the generation time of the target vocabulary and the current system time, Indicates the first set duration. Indicates the second setting duration. represents a fixed calculation factor coefficient, ; Step 505, judging whether the public opinion influence coefficient exceeds a preset public opinion influence coefficient threshold value by comparison; Step 506: If the public opinion impact coefficient exceeds the preset public opinion impact coefficient threshold, execute step 206; Step 507: if the public opinion influence coefficient does not exceed the preset public opinion influence coefficient threshold, the next element is selected from the sorted target set and recorded, and the historical public opinion keyword library corresponding to the next element is updated as the target word library, and steps 502 to 505 are executed until the public opinion influence coefficient exceeds the preset public opinion influence coefficient threshold, and then step 206 is executed; if the screening is completed in the sorted target set and the public opinion influence coefficients corresponding to all elements do not exceed the preset public opinion influence coefficient, then the release time of all public opinion data in the preliminary clustering result corresponding to the current latest public opinion keyword library is compared, and the public opinion data with the earliest release time is selected as the original public opinion data; The first branch processing module is used to compare the release time of all public opinion data in the preliminary clustering result corresponding to the current latest public opinion keyword library if the maximum similarity does not exceed the preset similarity threshold, and select the public opinion data with the earliest release time as the original public opinion data; The second branch processing module is used to use a forward iteration method to perform a reverse search in the historical public opinion keyword library if the maximum similarity exceeds a preset similarity threshold, until original public opinion data is found or a preset iteration termination condition is reached, and the iteration is terminated.

7. A computer device, characterized in that: It includes a memory and a processor, the memory stores computer-readable instructions, and the processor implements the steps of the public opinion release tracing method as described in any one of claims 1 to 5 when executing the computer-readable instructions.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the method for tracing the source of public opinion release as described in any one of claims 1 to 5.