Public opinion monitoring analysis method based on RPA

Abstract

The invention relates to the field of public opinion monitoring, in particular to an RPA-based public opinion monitoring analysis method, which comprises the following steps that: an RPA module acquires corresponding public opinion information; de-duplicating the collected public opinion information; carrying out structured processing; the title feature words are compared; comparing the content feature words; solving a public opinion evaluation value; when it is judged that the single piece of public opinion information does not meet the standard of the negative public opinion information according to the public opinion evaluation value, whether the next piece of public opinion information meets the standard of the negative public opinion information or not is judged, or whether the public opinion information meets the standard of the negative public opinion information or not is judged secondarily; judging the level of the public opinion event; judging the increase and decrease of the public opinion event level; and the public opinion event level is corrected, so that the accuracy and response speed of public opinion monitoring analysis are improved.

Description

Technical Field

[0001] The present invention relates to the field of public opinion monitoring, and in particular to a public opinion monitoring and analysis method based on RPA. Background Art

[0002] Public opinion monitoring refers to the use of modern information technology to track online media, social hot spots and emergencies in real time, analyze and study the content of information disseminated on the Internet and its changing trends, evaluate the impact of information on society and make corresponding early warnings.

[0003] China Patent Publication No.: CN114707045B discloses a public opinion monitoring method based on big data, the method comprising: S1, collecting public opinion information from big data to obtain original public opinion information; S2, performing data processing on the collected original public opinion information to obtain hot public opinion information; S3, transmitting the hot public opinion information to the terminal in real time. The present invention first classifies and groups the public opinion information with similar semantics through semantic analysis, and then collects the useful information in the classified and grouped public opinion information, which can greatly filter out useless information, thereby improving the accuracy of public opinion judgment; then, collect characteristic phrases that appear multiple times from the useful information, and filter out characteristic phrases that appear less frequently, which can further filter out useless information and further improve the accuracy of public opinion judgment; then, filter out characteristic phrases with shorter lengths. It can be seen that for the public opinion information collected in the same time period, it is impossible to automatically identify whether it is positive or negative public opinion information, nor can it determine the impact of the public opinion event. Summary of the invention

[0004] To this end, the present invention provides a public opinion monitoring and analysis method based on RPA to overcome the problem in the prior art that it is impossible to automatically and accurately monitor the trend of public opinion information within the same period of time.

[0005] To achieve the above object, the present invention provides a public opinion monitoring and analysis method based on RPA, comprising:

[0006] The RPA module collects corresponding public opinion information based on the set keywords on the corresponding network platform in response to the periodic time node;

[0007] Deduplication of the collected public opinion information;

[0008] Performing structural processing on the public opinion information after deduplication based on the title and content, wherein the structural processing includes public opinion text extraction and public opinion text word segmentation;

[0009] Compare the title text with the title feature words stored in the database;

[0010] Comparing the content text with the content feature words stored in the database;

[0011] Obtain the public opinion evaluation value;

[0012] When it is determined according to the public opinion evaluation value that a single piece of public opinion information does not meet the standard of negative public opinion information, determining whether the next single piece of public opinion information meets the standard of negative public opinion information, or, determining again whether the public opinion information meets the standard of negative public opinion information;

[0013] The level of a public opinion event is determined based on the ratio of the amount of negative public opinion information collected in a single cycle to the total amount of public opinion information after deduplication;

[0014] Determine the rise and fall of the level of public opinion events based on the expansion rate of negative public opinion information;

[0015] The public opinion event level is modified based on the amount of negative public opinion information after deduplication.

[0016] Furthermore, the public opinion evaluation value is set to E, where: Among them, α is the title influence coefficient, set α = 0.35, β is the content influence coefficient, set β = 0.85, n is the number of title feature words in the title text, N is the total number of words in the title text after word segmentation, m is the number of content feature words in the content text, and M is the total number of words in the content text after word segmentation.

[0017] Further, when the RPA module determines that a single piece of public opinion information does not meet the standard of negative public opinion information based on the public opinion evaluation value, it determines whether the next single piece of public opinion information meets the standard of negative public opinion information.

[0018] Or, reduce the length of the word segmentation of the public opinion text and re-determine whether the public opinion information meets the standard of negative public opinion information,

[0019] Or, based on the comparison between the structure of the sentence containing each content feature word and the structure of several sentences containing the content feature words stored in the database, it is determined whether the single piece of public opinion information meets the standard of negative public opinion information.

[0020] Furthermore, the RPA module is provided with several length adjustment methods for the length of the public opinion text segmentation, and each length adjustment method has a different adjustment range for the length of the public opinion text segmentation.

[0021] Furthermore, the process of the RPA module secondarily determining whether a single piece of public opinion information meets the standard of negative public opinion information based on the similarity ratio between the structure of the sentence containing each content feature word and the structure of several sentences containing the content feature word stored in the database includes:

[0022] Comparing the similarity ratio with a preset similarity ratio;

[0023] If the similarity ratio is greater than or equal to the preset similarity ratio, the public opinion information is determined to be negative public opinion information.

[0024] Furthermore, the RPA module determines the level of a public opinion event based on the proportion of the amount of negative public opinion information collected in a single cycle to the total amount of public opinion information after deduplication; the level of the public opinion event includes level one to level five.

[0025] Furthermore, the process by which the RPA module determines whether it is necessary to adjust the level of the public opinion event based on the expansion rate of negative public opinion information includes increasing the level of the public opinion event to a corresponding value according to the difference between the expansion rate and a preset expansion rate threshold, or increasing or decreasing the level of the public opinion event to a corresponding value according to the difference between the preset expansion rate threshold and the expansion rate.

[0026] Furthermore, the RPA module is provided with several adjustment methods for the level of the public opinion event, and each level adjustment method has a different adjustment range for the level of the public opinion event.

[0027] Furthermore, the process in which the RPA module determines whether the level of the public opinion event needs to be corrected based on the amount of the negative public opinion information after deduplication includes:

[0028] Comparing the amount of negative public opinion information after deduplication with a preset correction amount threshold;

[0029] If the amount of the negative public opinion information after deduplication is less than the preset correction amount threshold and the level of the public opinion event is greater than or equal to level 2, the level of the public opinion event will be reduced by one level.

[0030] Furthermore, the RPA module is provided with several correction methods for the frequency of collecting public opinion information, and each correction method has a different correction range for the collection frequency.

[0031] Compared with the prior art, the beneficial effect of the present invention lies in: the present invention removes duplications and performs structured processing on the collected public opinion information, then compares the title feature words and the content feature words respectively, and then obtains the public opinion evaluation value, and when it is determined according to the public opinion evaluation value that a single piece of public opinion information does not meet the standard of negative public opinion information, it is determined whether the next single piece of public opinion information meets the standard of negative public opinion information, or, it is determined again whether the public opinion information meets the standard of negative public opinion information, and the level of the public opinion event and the increase or decrease of the level of the public opinion event and the correction of the level of the public opinion event are determined, thereby realizing automatic monitoring of public opinion monitoring and improving the accuracy of public opinion monitoring analysis.

[0032] Furthermore, the present invention sets a public opinion evaluation value according to the results of the title text comparison and the content text comparison, thereby solving the problem that the nature of the captured public opinion information cannot be automatically evaluated.

[0033] Furthermore, when the present invention determines that a single piece of public opinion information does not meet the standard of negative public opinion information based on the public opinion evaluation value, it determines whether the next single piece of public opinion information meets the standard of negative public opinion information, or reduces the length of the public opinion text segmentation and re-determines whether the public opinion information meets the standard of negative public opinion information, or re-determines whether the single piece of public opinion information meets the standard of negative public opinion information based on the comparison of the structure of the sentence containing each content feature word with the structure of several sentences containing the content feature words stored in the database. Thereby, the accuracy of public opinion information recognition is improved.

[0034] Furthermore, the present invention is provided with several length adjustment methods for the length of the word segments of the public opinion text, so that when the initial identification of public opinion information is uncertain, the word segmentation standard can be adjusted in time, thereby improving the accuracy of public opinion information identification.

[0035] Furthermore, the present invention secondary determines whether a single piece of public opinion information meets the standard of negative public opinion information based on the similarity ratio between the structure of the sentence containing each content feature word and several sentence structures containing the content feature words stored in the database, thereby further improving the accuracy of public opinion information identification.

[0036] Furthermore, the present invention divides the levels of public opinion events based on the amount of negative public opinion information collected in a single cycle and the proportion of the total amount of public opinion information after deduplication, so as to accurately evaluate the impact of public opinion events.

[0037] Furthermore, the present invention determines whether the level of the public opinion event needs to be adjusted based on the expansion rate of the negative public opinion information, thereby dynamically adjusting the impact and direction of the public opinion event.

[0038] Furthermore, the present invention determines whether the level of the public opinion event needs to be revised based on the amount of the negative public opinion information after deduplication, thereby further improving the accuracy of the impact of the public opinion event.

[0039] Furthermore, the present invention is also provided with several correction methods for the frequency of collecting public opinion information, thereby realizing accurate monitoring of public opinion information. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a flow chart of a public opinion monitoring and analysis method based on RPA according to an embodiment of the present invention;

[0041] Figure 2 A flowchart of an embodiment of the present invention for determining whether a single piece of public opinion information meets the standard of negative public opinion information;

[0042] Figure 3 This is a flow chart of a method for adjusting the length of word segments of public opinion text according to an embodiment of the present invention;

[0043] Figure 4 The present invention is a flowchart of an embodiment of the present invention for determining whether the level of the public opinion event needs to be adjusted. DETAILED DESCRIPTION

[0044] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0045] It should be pointed out that the data in this embodiment are obtained by comprehensive analysis and evaluation of the historical analysis data and corresponding historical analysis results of the RPA module of the present invention in the three months before this analysis. Before this analysis, the RPA module of the present invention determines the values ​​of the preset parameter standards for this analysis based on the results of 142,583 online public opinion collection and analysis conducted cumulatively in the previous three months. It can be understood by those skilled in the art that the method of the present invention can determine the single parameter mentioned above by selecting the value with the highest proportion as the preset standard parameter according to the data distribution, using weighted summation to use the obtained value as the preset standard parameter, substituting each historical data into a specific formula and using the value obtained by the formula as the preset standard parameter or other selection methods, as long as the system of the present invention can clearly define the different specific situations in the single judgment process through the obtained values.

[0046] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0047] See also Figure 1 , Figure 2 , Figure 3 as well as Figure 4 As shown, they are respectively a flow chart of a public opinion monitoring and analysis method based on RPA according to an embodiment of the present invention; a flow chart of determining whether a single piece of public opinion information meets the standard of negative public opinion information according to an embodiment of the present invention; a flow chart of an adjustment method for the length of word segments of a public opinion text according to an embodiment of the present invention; and a flow chart of determining whether the level of a public opinion event needs to be adjusted according to an embodiment of the present invention.

[0048] An embodiment of the present invention provides a public opinion monitoring and analysis method based on RPA, including:

[0049] Step S1, the RPA module collects corresponding public opinion information according to the set keywords on the corresponding network platform in response to the periodic time node;

[0050] Step S2, removing duplicates from the collected public opinion information;

[0051] Step S3, performing structural processing on the public opinion information after deduplication based on the title and content, wherein the structural processing includes public opinion text extraction and public opinion text segmentation;

[0052] Step S4, comparing the title text with the title feature words stored in the database;

[0053] Step S5, comparing the content text with the content feature words stored in the database;

[0054] Step S6, obtaining a public opinion evaluation value based on the comparison result;

[0055] Step S7, when it is determined that a single piece of public opinion information does not meet the standard of negative public opinion information according to the obtained public opinion evaluation value, whether the next piece of public opinion information meets the standard of negative public opinion information is determined, or the length of the public opinion text segmentation is reduced and then the public opinion information is re-determined whether it meets the standard of negative public opinion information, or a second determination is made whether a single piece of public opinion information meets the standard of negative public opinion information based on a comparison between the structure of the sentence containing each content feature word and the structure of several sentences containing the content feature words stored in the database;

[0056] Step S8, preliminarily determining the level of the public opinion event based on the proportion of the amount of negative public opinion information collected in a single cycle to the total amount of public opinion information after deduplication;

[0057] Step S9, determining the rise or fall of the level of the public opinion event based on the expansion rate of the negative public opinion information;

[0058] Step S10, correcting the public opinion event level based on the amount of negative public opinion information after deduplication.

[0059] Specifically, the RPA is Robotic Process Automation, that is, robotic process automation. RPA is an automation module used to determine control, judgment and adjustment.

[0060] Specifically, online platforms such as news websites, forums, and self-media platforms, without specific restrictions, are used to publish public opinion information.

[0061] Specifically, the public opinion evaluation value is E, where: Among them, α is the title influence coefficient, set α = 0.35, β is the content influence coefficient, set β = 0.85, n is the number of title feature words in the title text, N is the total number of words in the title text after word segmentation, m is the number of content feature words in the content text, and M is the total number of words in the content text after word segmentation. The public opinion evaluation value is based on the results of comparing the title text with the title feature words stored in the database and the content text with the content feature words stored in the database, and different influence coefficients are set accordingly, thereby accurately solving the problem of not being able to automatically evaluate the nature of the captured public opinion information.

[0062] Specifically, the RPA module determines whether a single piece of public opinion information meets the standard of negative public opinion information based on the public opinion evaluation value, wherein:

[0063] The first judgment is that the RPA module determines that the single piece of public opinion information does not meet the standard of negative public opinion information, and determines whether the next single piece of public opinion information meets the standard of negative public opinion information; the first judgment satisfies that the public opinion evaluation value is less than the first preset public opinion evaluation value of 0.50%;

[0064] The second judgment is that the RPA module judges that the single piece of public opinion information does not meet the standard of negative public opinion information, and the RPA module reduces the length of the public opinion text segmentation based on the difference between the public opinion evaluation value and the first preset public opinion evaluation value, and makes a second judgment on whether the public opinion information meets the standard of negative public opinion information; the second judgment satisfies that the public opinion evaluation value is greater than or equal to the first preset public opinion evaluation value and is less than 2.86% of the second preset public opinion evaluation value;

[0065] The third judgment is that the RPA module determines that the single piece of public opinion information does not meet the standard of negative public opinion information, and secondarily judges whether the single piece of public opinion information meets the standard of negative public opinion information based on the comparison of the structure of the sentence containing each content feature word with the structure of several sentences containing the content feature words stored in the database; the third judgment satisfies that the public opinion evaluation value is greater than or equal to the second preset public opinion evaluation value and less than 6.95% of the third preset public opinion evaluation value;

[0066] The fourth judgment is that the RPA module determines that the single piece of public opinion information meets the standards of negative public opinion information, and determines that the public opinion information is negative public opinion information; the fourth judgment satisfies that the public opinion evaluation value is greater than or equal to the third preset public opinion evaluation value.

[0067] Specifically, the RPA module calculates the difference between the public opinion evaluation value and the first preset public opinion evaluation value under the second judgment, and records the difference as the public opinion difference. The RPA module determines the adjustment method for the word segmentation length of the public opinion text based on the public opinion difference, wherein:

[0068] The first length adjustment method is that the RPA module uses a first preset length adjustment coefficient of 0.95 to reduce the length of the public opinion text segmentation to a corresponding value; the first length adjustment method satisfies that the public opinion difference is less than the first preset public opinion difference of 0.86%;

[0069] The second length adjustment method is that the RPA module uses the second preset length adjustment coefficient 0.89 to reduce the length of the public opinion text segmentation to the corresponding value; the second length adjustment method satisfies that the public opinion difference is greater than or equal to the first preset public opinion difference and less than the second preset public opinion difference 1.78%;

[0070] The third length adjustment method is that the RPA module uses the third preset length adjustment coefficient 0.80 to reduce the word segmentation length of the public opinion text to a corresponding value; the third length adjustment method satisfies that the public opinion difference is greater than or equal to the second preset public opinion difference.

[0071] Specifically, the RPA module determines whether a single piece of public opinion information meets the standard of negative public opinion information based on the similarity ratio between the structure of the sentence containing each content feature word and the sentence structures containing the content feature word stored in the database under the third judgment, wherein:

[0072] The first type of determination is that the RPA module determines that a single piece of public opinion information does not meet the standard of negative public opinion information, and determines whether the next single piece of public opinion information meets the standard of negative public opinion information; the first type of determination satisfies that the similarity ratio is less than the preset similarity ratio of 75.58%;

[0073] The second type of judgment is that the RPA module judges that the single piece of public opinion information meets the standard of negative public opinion information, and judges that the public opinion information is negative public opinion information; the second type of judgment satisfies that the similarity ratio is greater than or equal to the preset similarity ratio.

[0074] Specifically, the RPA module preliminarily determines the level of the public opinion event based on the proportion of the amount of negative public opinion information collected in a single cycle to the total amount of public opinion information after deduplication, where:

[0075] The first level determination method is that the RPA module determines that the public opinion event is a first-level public opinion event; the first level determination method satisfies that the proportion is less than a first preset proportion of 25%;

[0076] The second level determination method is that the RPA module determines that the public opinion event is a second-level public opinion event; the second level determination method satisfies that the proportion is greater than or equal to the first preset proportion and less than 45% of the second preset proportion;

[0077] The third level determination method is that the RPA module determines that the public opinion event is a third-level public opinion event; the third level determination method satisfies that the proportion is greater than or equal to the second preset proportion and less than 65% of the third preset proportion;

[0078] The fourth level determination method is that the RPA module determines that the public opinion event is a fourth level public opinion event; the fourth level determination method satisfies that the proportion is greater than or equal to the third preset proportion and less than 85% of the fourth preset proportion;

[0079] The fifth level determination method is that the RPA module determines that the public opinion event is a fifth-level public opinion event; the fifth level determination method satisfies that the proportion is greater than or equal to the fourth preset proportion.

[0080] Specifically, the RPA module determines whether the public opinion event level needs to be adjusted based on the expansion rate of negative public opinion information, including:

[0081] The expansion rate is compared with a preset expansion rate threshold, and the preset expansion rate threshold is set to 2885 items / hour;

[0082] If the expansion rate is greater than a preset expansion rate threshold, the public opinion event level is increased to a corresponding value according to the difference between the expansion rate and the preset expansion rate threshold;

[0083] If the expansion rate is equal to the preset expansion rate threshold, the public opinion event level is not adjusted;

[0084] If the expansion rate is less than the preset expansion rate threshold, the public opinion event level is reduced to a corresponding value according to the difference between the preset expansion rate threshold and the expansion rate;

[0085] The expansion rate is the ratio of the number of deduplicated negative public opinion information in a single cycle to the length of a single cycle.

[0086] Specifically, the RPA module calculates the difference between the expansion rate and the preset expansion rate threshold, and records the difference as the expansion difference. The RPA module determines the adjustment method for the public opinion event level based on the expansion difference, wherein:

[0087] The first level adjustment method is that the RPA module increases the level of the public opinion event by one level; the first level adjustment method satisfies that the expansion difference is less than the preset expansion difference of 1024 items / hour, and the level of the public opinion event is less than or equal to level 4;

[0088] The second level adjustment method is that the RPA module increases the level of the public opinion event by two levels; the second level adjustment method satisfies that the expansion difference is greater than or equal to the preset expansion difference of 2557 items / hour, and the level of the public opinion event is less than or equal to level three.

[0089] Specifically, the RPA module determines whether the public opinion event level needs to be corrected based on the amount of negative public opinion information after deduplication, including:

[0090] Compare the number of negative public opinion information after deduplication with the preset correction number threshold, and set the preset correction number threshold to 10;

[0091] If the amount of the negative public opinion information after deduplication is less than the preset correction amount threshold and the level of the public opinion event is greater than or equal to level 2, the level of the public opinion event will be reduced by one level.

[0092] Specifically, the RPA module modifies the frequency of collection based on the total amount of the public opinion information collected in a single period, wherein:

[0093] The first frequency correction method is that the RPA module uses the first preset frequency correction coefficient 0.9 to reduce the duration of a single collection cycle to a corresponding value; the first frequency correction method satisfies that the total amount is less than the first preset total amount of 5500 pieces / hour;

[0094] The second frequency correction method is that the RPA module uses the second preset frequency correction coefficient 0.8 to reduce the duration of a single collection cycle to a corresponding value; the second frequency correction method satisfies that the total amount is greater than or equal to the first preset total amount and less than the second preset total amount of 12000 pieces / hour;

[0095] The third frequency correction method is that the RPA module uses the third preset frequency correction coefficient 0.7 to reduce the duration of a single acquisition cycle to a corresponding value; the third frequency correction method satisfies that the total amount is greater than or equal to the second preset total amount.

[0096] Example 1

[0097] Take the public opinion monitoring and analysis of a listed company as an example:

[0098] Based on the company name "XXXXXX Company" and "trade" as keywords, the corresponding information is retrieved and downloaded from news websites, forums, self-media platforms and other corresponding network platforms every two hours;

[0099] From 8:00 to 10:00 on November XX, 2023, a total of 4,500 pieces of public opinion information were collected. After downloading, the collected public opinion information was deduplicated, and 26 pieces of public opinion information were obtained after deduplication.

[0100] After deduplication, only the title and corresponding content of each piece of public opinion information are extracted. The RPA module automatically enters the extracted title and content into the structured template. For a single piece of public opinion information,

[0101] Compare the entered single piece of public opinion information with the title feature words and content feature words stored in the database.

[0102] The results of the comparison are as follows:

[0103] The number of title feature words in the title text 2,

[0104] The total number of words in the title text after word segmentation is 12.

[0105] The number of content feature words in the content text is 25.

[0106] The total number of words in the content text after word segmentation is 1224.

[0107] Based on the comparison results, the public opinion evaluation value is Then E=7.57%, and the obtained public opinion evaluation value is greater than or equal to the third preset public opinion evaluation value of 6.95%, and it is determined that the public opinion information is negative public opinion information.

[0108] After comparing the 26 pieces of public opinion information obtained one by one, a total of 19 negative public opinion information were obtained, and the negative public opinion information accounted for 73%. It was determined that the public opinion events about "XXXXXX Company" and "Trade" in this time period were level 4 public opinion events.

[0109] Example 2

[0110] Take the public opinion monitoring and analysis of a certain subject as an example:

[0111] Based on the name "XXX" as the keyword, the corresponding information is retrieved and downloaded from news websites, forums, self-media platforms and other corresponding network platforms every hour;

[0112] From 12:00 to 13:00 on November XX, 2023, a total of 14,500 pieces of public opinion information were collected. After downloading, the collected public opinion information was deduplicated, and 37 pieces of public opinion information were obtained after deduplication.

[0113] After deduplication, only the title and corresponding content of each piece of public opinion information are extracted. The RPA module automatically enters the extracted title and content into the structured template. For a single piece of public opinion information,

[0114] Compare the entered single piece of public opinion information with the title feature words and content feature words stored in the database.

[0115] The results of the comparison are as follows:

[0116] The number of title feature words in the title text 1,

[0117] The total number of words in the title text after word segmentation is 14.

[0118] The number of content feature words in the content text is 26.

[0119] The total number of words in the content text after word segmentation is 3225.

[0120] Based on the comparison results, the public opinion evaluation value is Then the public opinion evaluation value E=3.19%, and the obtained public opinion evaluation value is greater than or equal to the second preset public opinion evaluation value of 2.86% and less than the third preset public opinion evaluation value of 6.95%. The RPA module determines whether a single piece of public opinion information meets the standard of negative public opinion information based on the comparison of the structure of the sentence containing the 26 content feature words with the sentence structures containing the content feature words stored in the database. After comparison and statistics, the similarity ratio is 52.25%, then the public opinion information is not negative public opinion information.

[0121] After comparing the 37 pieces of public opinion information obtained one by one, a total of 12 pieces of negative public opinion information were obtained, and the negative public opinion information accounted for 32%. It was determined that the public opinion event about the subject in this time period was a secondary public opinion event.

[0122] The RPA module further detects that the expansion rate within the 1-hour period is 3558 items / hour, and the expansion difference is 3558-2885=673 items / hour, so the level of the public opinion event is increased by one level. After adjustment, the public opinion event of the subject is finally obtained as a level three public opinion event.

[0123] And the number of the negative public opinion information after deduplication, 12 pieces, is greater than the preset correction number threshold of 10, so there is no need to correct the adjusted public opinion event.

[0124] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0125] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A public opinion monitoring and analysis method based on RPA, characterized in that: include: The RPA module collects corresponding public opinion information based on the set keywords on the corresponding network platform in response to the periodic time node; Deduplication of the collected public opinion information; Performing structural processing on the public opinion information after deduplication based on the title and content, wherein the structural processing includes public opinion text extraction and public opinion text word segmentation; Compare the title text with the title feature words stored in the database; Comparing the content text with the content feature words stored in the database; Obtain the public opinion evaluation value; When it is determined according to the public opinion evaluation value that a single piece of public opinion information does not meet the standard of negative public opinion information, determining whether the next single piece of public opinion information meets the standard of negative public opinion information, or, determining again whether the public opinion information meets the standard of negative public opinion information; The level of a public opinion event is determined based on the ratio of the amount of negative public opinion information collected in a single cycle to the total amount of public opinion information after deduplication; Determine the rise and fall of the level of public opinion events based on the expansion rate of negative public opinion information; The public opinion event level is modified based on the amount of negative public opinion information after deduplication.

2. The RPA-based public opinion monitoring and analysis method according to claim 1 is characterized in that: The public opinion evaluation value is set to E, where: Among them, α is the title influence coefficient, set α = 0.35, β is the content influence coefficient, set β = 0.85, n is the number of title feature words in the title text, N is the total number of words in the title text after word segmentation, m is the number of content feature words in the content text, and M is the total number of words in the content text after word segmentation.

3. The RPA-based public opinion monitoring and analysis method according to claim 2 is characterized in that: When the RPA module determines that a single piece of public opinion information does not meet the standard of negative public opinion information based on the public opinion evaluation value, it determines whether the next single piece of public opinion information meets the standard of negative public opinion information. Or, reduce the length of the word segmentation of the public opinion text and re-determine whether the public opinion information meets the standard of negative public opinion information, Or, based on the comparison between the structure of the sentence containing each content feature word and the structure of several sentences containing the content feature words stored in the database, it is determined whether the single piece of public opinion information meets the standard of negative public opinion information.

4. The RPA-based public opinion monitoring and analysis method according to claim 3 is characterized in that: The RPA module is provided with several length adjustment methods for the length of the public opinion text segmentation, and each length adjustment method has a different adjustment range for the length of the public opinion text segmentation.

5. The RPA-based public opinion monitoring and analysis method according to claim 4 is characterized in that: The process of the RPA module secondarily determining whether a single piece of public opinion information meets the standard of negative public opinion information based on the similarity ratio between the structure of the sentence containing each content feature word and the structure of several sentences containing the content feature words stored in the database includes: Comparing the similarity ratio with a preset similarity ratio; If the similarity ratio is greater than or equal to the preset similarity ratio, the public opinion information is determined to be negative public opinion information.

6. The RPA-based public opinion monitoring and analysis method according to claim 5 is characterized in that: The RPA module determines the level of a public opinion event based on the amount of negative public opinion information collected in a single cycle and the proportion of the total amount of public opinion information after deduplication; the level of the public opinion event includes level one to level five.

7. The RPA-based public opinion monitoring and analysis method according to claim 6 is characterized in that: The process by which the RPA module determines whether it is necessary to adjust the level of the public opinion event based on the expansion rate of negative public opinion information includes increasing the level of the public opinion event to a corresponding value according to the difference between the expansion rate and a preset expansion rate threshold, or increasing or decreasing the level of the public opinion event to a corresponding value according to the difference between the preset expansion rate threshold and the expansion rate.

8. The RPA-based public opinion monitoring and analysis method according to claim 7 is characterized in that: The RPA module is provided with several adjustment methods for the level of the public opinion event, and each level adjustment method has a different adjustment range for the level of the public opinion event.

9. The RPA-based public opinion monitoring and analysis method according to claim 8 is characterized in that: The process in which the RPA module determines whether the level of the public opinion event needs to be corrected based on the amount of the negative public opinion information after deduplication includes: Comparing the amount of negative public opinion information after deduplication with a preset correction amount threshold; If the amount of the negative public opinion information after deduplication is less than the preset correction amount threshold and the level of the public opinion event is greater than or equal to level 2, the level of the public opinion event will be reduced by one level.

10. The RPA-based public opinion monitoring and analysis method according to claim 9 is characterized in that: The RPA module is provided with several correction methods for the frequency of collecting public opinion information, and each correction method has a different correction range for the collection frequency.

Images