Public opinion monitoring and analyzing system and method for smart travel big data

Abstract

The invention discloses an intelligent travel big data public opinion monitoring and analysis system and method, relates to the technical field of public opinion monitoring, and solves the problems that auditing of the authenticity of acquired public opinion information is not considered in the prior art, so that follow-up selection and decision-making of tourists on scenic spots are affected, and the experience of tourists is affected. And negative effects on economic income and reputation of the scenic spot are caused. The method comprises the following steps: collecting public opinion information from multiple propagation media to construct a public opinion information base; setting keywords and screening out public opinion information from a public opinion information base according to the keywords; constructing and training a public opinion type judgment model; determining the type of the public opinion information in the query time range according to a public opinion type judgment model; constructing a public opinion correction factor to correct the type of the public opinion information; according to the method, the authenticity of the public opinion information is verified, the correct cognition of tourists on the scenic spot is improved, and meanwhile, the economic development of the scenic spot is promoted to a great extent.

Description

Technical Field

[0001] The present invention belongs to the field of Internet technology, relates to public opinion monitoring technology, and specifically is a big data public opinion monitoring and analysis system and method for intelligent culture and tourism. Background Art

[0002] The big data public opinion monitoring and analysis system for intelligent culture and tourism not only has a monitoring function, but also can provide targeted public opinion guidance and response suggestions according to the public opinion analysis results. This helps the culture and tourism department or scenic area to take measures in a timely manner, guide the trend of public opinion, reduce the impact of negative public opinion, and enhance the public's trust and satisfaction with the culture and tourism industry; the construction and application of the big data public opinion monitoring and analysis system for intelligent culture and tourism is an important manifestation of the intelligent development of the culture and tourism industry; by introducing big data technology and intelligent algorithms, the system can realize the automated and intelligent processing and analysis of public opinion information, improving the efficiency and quality of public opinion monitoring and response. This will help to promote the culture and tourism industry to develop in a more intelligent and efficient direction.

[0003] The prior art (CN116955621A) discloses a monitoring method applicable to tourism public opinion, including the following steps: determining the targets to be monitored and the data sources to be collected; using web crawler technology to capture the text data related to the monitoring targets in the data sources, and filtering and preprocessing the captured data to remove irrelevant information; performing data mining and public opinion analysis on the preprocessed text data, extracting the key information therein, classifying and summarizing the key information, and marking the sentiment of the text; visualizing the analysis results and generating corresponding reports and analysis results; the prior art performs data mining and public opinion analysis on the text data, without considering the verification of the accuracy and authenticity of the obtained public opinion information, and at the same time, does not conduct positive public opinion guidance and correction of wrong public opinion, thus affecting the subsequent choices and decisions of tourists for scenic areas and having a negative impact on the economic income and reputation of scenic areas.

[0004] The present invention provides a big data public opinion monitoring and analysis system and method for intelligent culture and tourism to solve the above technical problems. Summary of the Invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; for this purpose, the present invention proposes a big data public opinion monitoring and analysis system for intelligent culture and tourism, which is used to solve the technical problems that the prior art does not consider the verification of the accuracy and authenticity of the obtained public opinion information, and at the same time, does not conduct positive public opinion guidance and correction of wrong public opinion, thus affecting the subsequent choices and decisions of tourists for scenic areas and having a negative impact on the economic income and reputation of scenic areas.

[0006] To achieve the above object, a first aspect of the present invention provides a smart cultural and tourism big data public opinion monitoring and analysis system, including: a public opinion collection module and a connected public opinion analysis module;

[0007] Public opinion collection module: Collect public opinion information from multiple communication media to construct a public opinion information database; set keywords and screen a number of public opinion information from the public opinion information database according to the keywords; sort the number of public opinion information in chronological order; wherein, the keywords include: scenic spot name, custom query term; public opinion information includes: occurrence time of public opinion, end time, scenic spot name and custom query term; and the occurrence time and end time are determined according to the popularity, click-through rate or search rate of the platform where the public opinion is located;

[0008] Public opinion analysis module: Construct and train a public opinion type judgment model; determine the type of public opinion information within the query time range according to the public opinion type judgment model; and,

[0009] Construct a public opinion correction factor to correct the type of public opinion information; wherein, the types of public opinion information include: positive, negative and neutral.

[0010] Preferably, the system includes: a public opinion feedback module connected to the public opinion analysis module: used to push the type of public opinion information to the user, and used to update the type of public opinion information.

[0011] Preferably, the setting of keywords and screening a number of public opinion information from the public opinion information database according to the keywords includes:

[0012] S110: Mark the scenic spot name as the first keyword A, and mark the custom query term as the second keyword Bj; wherein, Bj represents the jth second keyword; j = 1, 2, 3,..., n; n is a positive integer;

[0013] S120: Set the query interval duration of public opinion information;

[0014] S130: Set the format of the jth marker as Search all historical public opinion information in the public opinion information database in turn within the query time range, obtain the historical public opinion information corresponding to different markers, and count the number of historical public opinion information corresponding to different markers and mark it as YQG j 。

[0015] It should be noted that the public opinion information database includes historical public opinion information and public opinion information within the set query interval duration.

[0016] The present invention screens according to the relevance of search results to ensure that the obtained public opinion information is closely related to the keywords; screening by keywords is conducive to screening according to the keywords of public opinion information, and giving priority to paying attention to public opinion information with greater influence and high attention; screening by timeliness is conducive to ensuring that the screened public opinion information is the latest to reflect the current public opinion dynamics and trends.

[0017] Preferably, setting the query interval duration of the public opinion information includes:

[0018] Retrieve the public opinion information of the scenic area from the public opinion information database, and count the number k of negative public opinion occurrences in the corresponding scenic area within two years from the current time. The average duration of each historical public opinion is marked as CXS k ; Obtain the income of the scenic area when there is no public opinion in each historical month; where k = 0, 1, 2, …, n; n ∈ [0, +∞];

[0019] Judge whether the occurrence of the most recent public opinion makes the income of the scenic area in that month less than the preset low-income threshold; if so, mark that the occurrence of the public opinion has an impact on the local scenic area's income as SR exists and set SR = 1; if not, mark that the occurrence of the public opinion has no impact on the local scenic area's income as SR and set SR = 0;

[0020] Count the duration of the end of several public opinion information as T k ;

[0021] Through the formula | Calculate the range influence factor;

[0022] Through the formula CQS = CXS k × FYZ to calculate the query interval duration.

[0023] It should be noted that the specific value of the low-income threshold is set by the local cultural and tourism department staff according to the income of the local scenic area in a specific month.

[0024] The present invention determines the query time range of public opinion information by considering the number of historical negative public opinion information in the corresponding scenic area and the influence degree of historical public opinion on the current public opinion, which is conducive to capturing and evaluating the type of current real-time public opinion faster.

[0025] Preferably, constructing and training the public opinion type judgment model includes:

[0026] S210: Retrieve the historical public opinion information corresponding to different markers and obtain the type of historical public opinion information;

[0027] S220: Retrieve the public opinion type judgment model, and mark the number of historical public opinion information corresponding to the marker j = 1 as YQG 1 ; Will The corresponding historical public opinion information is input into the public opinion type judgment model in sequence; it is judged whether the type of the historical public opinion information output by the public opinion type judgment model is consistent with the type of the marked historical public opinion information; if so, the output result of the public opinion type judgment model is marked as accurate; if not, the output result of the public opinion type judgment model is marked as inaccurate; count the number of public opinion information with accurate output results and record it as YQZ 1 ; Among them, the public opinion type judgment model is constructed based on an artificial intelligence model;

[0028] Through the formula YQZ 1 / 2YQG 1 Calculate the accuracy rate of the corresponding historical public opinion information when j = 1;

[0029] S230: Input the corresponding historical public opinion information into the public opinion type judgment model in sequence, repeat step S220, and obtain the accuracy rates of the historical public opinion information corresponding to different markers;

[0030] S240: Use the number of keywords as the abscissa and the accuracy rate as the ordinate to construct a rectangular coordinate system, count the total number of the curve slopes between adjacent two points in the rectangular coordinate system, and the number of the curve slopes between adjacent two points in the rectangular coordinate system that are greater than or equal to 0; judge whether the proportion of the number of the curve slopes greater than or equal to 0 in the total number is greater than the proportion threshold; if so, complete the training of the public opinion type judgment model; if not, with the first term as Increase the number of the historical public opinion information input into the public opinion type judgment model in sequence with a common difference of 2 until the proportion of the number of the curve slopes greater than or equal to 0 in the total number is greater than the proportion threshold, and then end the process of continuing the training.

[0031] It should be noted that when the number of the historical public opinion information corresponding to different markers is odd, the number of the historical public opinion information input into the public opinion type judgment model is determined according to the method of rounding.

[0032] In the present invention, by inputting the historical public opinion information into the public opinion type judgment model and comparing the consistency between the type output by the model and the marked type, the accuracy of the model can be intuitively evaluated. This evaluation method is both direct and effective, which helps to timely discover the problems and deficiencies of the model; during the model training process, by adjusting the number of the input public opinion information, the difficulty and complexity of the training can be flexibly controlled, which helps to gradually improve the performance of the model until the accuracy rate requirement is met; by constructing a rectangular coordinate system and counting the curve slopes between adjacent two points, the change trend of the model performance can be intuitively reflected.

[0033] Preferably, the public opinion type judgment model is constructed based on an artificial intelligence model, including:

[0034] Obtain standard training data; integrate all historical public opinion information through word cloud generation technology to form a displayed word cloud, and sort the displayed words in the displayed word cloud in descending order of font size to form a judgment factor sequence; wherein, the standard training data includes standard input data consistent with the content attributes of the judgment factor sequence, and standard output data consistent with the content attributes of the judgment factor sequence; the standard output data is the type of public opinion information; use the standard training data to train the artificial intelligence model;

[0035] Mark the trained artificial intelligence model as a public opinion type judgment model; wherein, the artificial intelligence model includes a convolutional neural network model or a long short-term memory neural network model.

[0036] Preferably, determining the type of public opinion information within the query time range according to the public opinion type judgment model includes:

[0037] Retrieve the public opinion information within the query time range, and obtain the displayed word cloud of the public opinion information within the query time range, and analyze the sentiment types of several displayed words in the displayed word cloud based on the sentiment dictionary; wherein, the sentiment types include: favorability degree a, favorability degree b, favorability degree 0, favorability degree -b, favorability degree -a; count the number of all displayed words in the displayed word cloud and the number of displayed words with a favorability degree greater than 0 in the sentiment type; divide the number of displayed words with a favorability degree greater than 0 in the sentiment type by the number of all displayed words to obtain the favorability ratio; wherein, a > b > 0;

[0038] Judge whether the favorability ratio is greater than 80%; if so, mark the corresponding public opinion information as positive; if not, judge whether the favorability ratio is less than 20%; if so, mark the corresponding public opinion information as negative;

[0039] If not, mark the corresponding public opinion information as neutral.

[0040] The present invention analyzes the sentiment types of displayed words based on the sentiment dictionary, and can accurately judge the sentiment tendency of public opinion information; by counting the number of all displayed words in the displayed word cloud and the number of displayed words with a favorability degree greater than 0 in the sentiment type, the overall sentiment tendency of public opinion information can be quantitatively evaluated, and this quantitative method helps to more objectively judge the nature of public opinion information; judging the type of public opinion information according to the favorability ratio realizes the intelligent classification of public opinion information, and this classification method is both fast and accurate, which helps to quickly respond to and process different types of public opinion information.

[0041] Preferably, constructing a public opinion correction factor to correct the type of public opinion information includes:

[0042] Retrieve public opinion information marked as neutral, and count the number of likes DZL, forwarding ZFL, and comments corresponding to the neutral public opinion information with a favorability ratio in the range of [20%, 40%] ∪ [60%, 80%]; remove invalid comments from the comments; mark the number of comments after removing invalid comments as PLZ, where invalid comments are: comments with less than three words, repetitive comments, and comments with non-text content; mark the number of invalid comments as WXL;

[0043] By formula The public opinion correction factor is calculated; where α and β are correction coefficients;

[0044] Determine whether the public opinion correction factor is greater than 0; if so, replace the public opinion information marked as neutral with the positive one; if not, determine whether the public opinion correction factor is less than 0; if so, replace the public opinion information marked as neutral with the negative one; if not, do not replace the public opinion information marked as neutral.

[0045] The present invention reduces the misjudgment rate of public opinion information and improves the accuracy and practicality of judging the type of public opinion information by further subdividing and correcting neutral public opinion information.

[0046] Preferably, the correction coefficient is obtained by:

[0047] Get the number of digits e of the like value DZL and the number of digits w of the forwarding value ZFL,

[0048] The calculation method of the correction coefficient α is: α = 10 -e ;

[0049] The calculation method of the correction coefficient β is: β = 10 -w ; Among them, the value range of e and w is [0, +∞].

[0050] To achieve the above-mentioned purpose, the second aspect of the present invention provides a method for monitoring and analyzing public opinion of smart cultural tourism big data, comprising:

[0051] Collect public opinion information from multiple communication media to build a public opinion information database; set keywords and filter out a number of public opinion information from the public opinion information database based on the keywords; sort the several public opinion information in chronological order; the keywords include: scenic spot name, custom query words; public opinion information includes: occurrence time, end time, scenic spot name and custom query words of the public opinion; and the occurrence time and end time are determined according to the popularity and click rate or search rate of the platform where the public opinion is located;

[0052] Construct and train a public opinion type judgment model; determine the type of public opinion information within the query time range according to the public opinion type judgment model; and,

[0053] Construct an opinion sentiment correction factor to correct the type of opinion sentiment information; among them, the types of opinion sentiment information include: positive, negative, and neutral.

[0054] Compared with the prior art, the beneficial effects of the present invention are:

[0055] 1. By collecting opinion sentiment information from multiple communication media, the present invention ensures the comprehensiveness and diversity of information, and avoids biases caused by the singularity of information sources; by setting keywords to screen information, it is beneficial to accurately locate the opinion sentiment related to the theme, improving the accuracy of information; by constructing and training an opinion sentiment type judgment model, the rapid judgment of the type of opinion sentiment information is realized, greatly improving the processing efficiency; the automated process reduces manual intervention, lowers labor costs, and at the same time improves the processing speed and consistency; the opinion sentiment type judgment model can be adjusted and optimized according to actual needs to adapt to different opinion sentiment environments and requirements.

[0056] 2. For the opinion sentiment information judged to be neutral, the present invention further limits the range and constructs an opinion sentiment correction factor to correct the type of opinion sentiment information, improving the accuracy and practicality of the judgment of the type of opinion sentiment information; by classifying and correcting the type of opinion sentiment information, clear and accurate opinion sentiment information can be provided for decision-makers, helping them make more informed decisions; by obtaining three types of opinion sentiment information, namely positive, negative, and neutral, it helps tourists comprehensively and objectively understand the attitudes and views of other publics towards different topics. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0058] Figure 1 It is a schematic diagram of the module relationship included in the present invention;

[0059] Figure 2 It is a schematic diagram of the specific steps for obtaining opinion sentiment information of the present invention;

[0060] Figure 3 It is a schematic diagram of the specific steps for training the opinion sentiment type judgment model of the present invention;

[0061] Figure 4 It is a schematic diagram of the specific steps for revising the type of opinion sentiment information of the present invention;

[0062] Figure 5 It is a schematic diagram of the process for detecting and analyzing opinion sentiment of the present invention. Specific Embodiments

[0063] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative work shall fall within the protection scope of the present invention.

[0064] Please refer to Figure 1 , an embodiment of the first aspect of the present invention provides a smart cultural and tourism big data public opinion monitoring and analysis system, including: a public opinion collection module, a public opinion analysis module, and a public opinion feedback module connected thereto;

[0065] Public opinion collection module: Collect public opinion information from multiple communication media to construct a public opinion information database; Set keywords and screen a number of public opinion information from the public opinion information database according to the keywords; The number of public opinion information is sorted in the order of occurrence time; Among them, the keywords include: scenic spot name, custom query word; Public opinion information includes: the occurrence time, end time, scenic spot name and custom query word of the public opinion; The occurrence time and end time are determined according to the popularity, click-through rate or search rate of the platform where the public opinion is located;

[0066] Public opinion analysis module: Construct and train a public opinion type judgment model; Determine the type of public opinion information within the query time range according to the public opinion type judgment model; And,

[0067] Construct a public opinion correction factor to correct the type of public opinion information; Among them, the types of public opinion information include: positive, negative and neutral;

[0068] Public opinion feedback module: Used to push the type of public opinion information to the user, and used to update the type of public opinion information.

[0069] Please refer to Figure 2 , the specific steps of obtaining public opinion information,

[0070] S110: Mark the scenic spot name as the first keyword A, and mark the custom query word as the second keyword Bj; Among them, Bj represents the jth second keyword; j = 1, 2, 3,..., n; n is a positive integer;

[0071] S120: Retrieve the public opinion information of the scenic spot from the public opinion information database, count the number of negative public opinions k that occurred in the corresponding scenic spot within two years from the current time, and mark the average duration of each historical public opinion as CXS k ; Obtain the income of the scenic spot when there is no public opinion in each historical month; Among them, k = 0, 1, 2,..., n; n ∈ [0, +∞);

[0072] Determine whether the occurrence of the most recent public opinion sentiment has made the scenic area revenue in the current month less than the preset low-income threshold; if so, mark that the occurrence of the public opinion sentiment has an impact on the local scenic area revenue as SR exists, and set SR = 1; if not, mark that the occurrence of the public opinion sentiment has no impact on the local scenic area revenue as SR, and set SR = 0;

[0073] Statistically mark the duration of the end of the public opinion sentiment of several pieces of public opinion information as T k ;

[0074] Through the formula |Calculate the range influence factor;

[0075] Through the formula CQS = CXS k × FYZ to calculate the query interval duration;

[0076] S130: Set the format of the j-th marker as In the query time range, sequentially search the sentiment information database for all historical public opinion sentiment information corresponding to the markers from j = n to j = 1, obtain the historical public opinion sentiment information corresponding to different markers, and statistically mark the number of historical public opinion sentiment information corresponding to different markers as YQG j 。

[0077] For example, there is a scenic area named A. Use the name of the scenic area as the first keyword; the custom query terms are: service level, attitude of scenic area service staff, and environmental sanitation status of the scenic area; use the custom query terms as the second keyword; now collect the public opinion sentiment information of Scenic Area A from multiple communication media such as news, social media, mainstream portal websites, forums, Weibo, blogs, WeChat official accounts, and OTA platforms to construct a public opinion sentiment information database; based on the keywords, screen out 20 pieces of public opinion sentiment information from the public opinion sentiment information database;

[0078] Retrieve the public opinion sentiment information of the scenic area from the public opinion sentiment information database. Statistically mark that the number of negative public opinion sentiment occurrences in the corresponding scenic area in the past two years is 2 times, which occurred on May 13, 2023 and lasted for 9 days to end, and on June 20, 2023 and lasted for 5 days to end; the average duration of the two public opinion sentiments is 7 days;

[0079] Obtain the average revenue of Scenic Area A in June for 20, 21, and 22 years as 310,000 yuan and use 310,000 yuan as the set low-income threshold for June;

[0080] Obtain the revenue of Scenic Area A during the public opinion sentiment occurrence in June 2023, which is the most recent time, as 280,000 yuan, lower than the set low-income threshold of 310,000 yuan in June; then the public opinion sentiment in June has an impact on the revenue of Scenic Area A, and assign this revenue impact as 1;

[0081] Through the formula |Calculate that the range influence factor is 1.7;

[0082] Through the formula CQS = CXS k × FYZ = 7 × 1.7 = The query interval duration of 11.9 days is calculated

[0083] Set the format of the first marker as A, and 20 corresponding public opinion information are searched; the formats of the second markers are: A + service level, A + attitude of scenic area service staff, and A + environmental hygiene status of scenic area, and 10 corresponding public opinion information are searched; the formats of the third markers are: A + service level + attitude of scenic area service staff and A + service level + environmental hygiene status of scenic area, and 5 corresponding public opinion information are searched; the format of the fourth marker is: A + service level + attitude of scenic area service staff + environmental hygiene status of scenic area, and 2 corresponding public opinion information are searched.

[0084] The present invention calculates the query interval duration by considering the impact of the most recent public opinion event on the current scenic area; it is beneficial to monitor public opinion more comprehensively and reasonably, enable tourists to have a more comprehensive understanding of the scenic area, and objectively provide tourists with an evaluation of the public opinion of the scenic area.

[0085] Please refer to Figure 3 , the specific steps of training the public opinion type judgment model, including:[[]]

[0086] S210: Retrieve the historical public opinion information corresponding to different markers, and obtain the types of historical public opinion information;

[0087] S220: Retrieve the public opinion type judgment model, and mark the number of historical public opinion information corresponding to the marker j = 1 as YQG 1 ; and The corresponding historical public opinion information is sequentially input into the public opinion type judgment model; judge whether the type of the historical public opinion information output by the public opinion type judgment model is consistent with the type of the marked historical public opinion information; if so, mark the output result of the public opinion type judgment model as accurate; if not, mark the output result of the public opinion type judgment model as inaccurate; count the number of public opinion information with accurate output results and record it as YQZ 1 ; among them, the public opinion type judgment model is constructed based on an artificial intelligence model;

[0088] Through the formula YQZ 1 / 2YQG 1 Calculate the accuracy rate of the historical public opinion information corresponding to j = 1;

[0089] S230: The corresponding historical public opinion information is sequentially input into the public opinion type judgment model, and step S220 is repeated to obtain the accuracy rates of the historical public opinion information corresponding to different markers;

[0090] S240: Use the number of keywords as the abscissa and the accuracy rate as the ordinate to construct a rectangular coordinate system. Count the total number of the curve slopes between adjacent two points in the rectangular coordinate system, and the number of the curve slopes between adjacent two points in the rectangular coordinate system that are greater than or equal to 0; Determine whether the proportion of the number of the curve slopes greater than or equal to 0 in the total number is greater than the proportion threshold; If yes, complete the training of the public opinion type judgment model; If not, use the first term as the common difference is 2, and increase the number of historical public opinion information input to the public opinion type judgment model in turn until the proportion of the number of the curve slopes greater than or equal to 0 in the total number is greater than the proportion threshold, and then end the process of continuing the training.

[0091] For example, retrieve the public opinion type judgment model and input that is, 10 corresponding historical public opinion information into the public opinion type judgment model in turn. The number of public opinion information with accurate output results through the public opinion type judgment model is 5;

[0092] Input that is, 5, 3, and 1 corresponding historical public opinion information into the public opinion type judgment model in turn. The number of public opinion information with accurate output results through the public opinion type judgment model is 3, 2, and 1 respectively;

[0093] Through the formula YQZ 1 / 2YQG 1 Calculate that the accuracy rate of the corresponding public opinion information when j = 1 is 50%, the accuracy rate of the corresponding public opinion information when j = 2 is 60%, the accuracy rate of the corresponding public opinion information when j = 3 is 66.6%, and the accuracy rate of the corresponding public opinion information when j = 5 is 100%;

[0094] Use the number of keywords as the abscissa and the accuracy rate as the ordinate to construct a rectangular coordinate system. The total number of the curve slopes between adjacent two points in the rectangular coordinate system is 4; The number of the curve slopes between adjacent two points in the rectangular coordinate system that are greater than or equal to 0 is 4, and the proportion of the number of the curve slopes greater than or equal to 0 in the total number is 100%. According to the accuracy requirements of the model training by the staff in this field, set the proportion threshold to 95%; Since 100% is greater than 95%, the training of the public opinion type judgment model is completed.

[0095] Please refer to Figure 4, The specific steps for revising the public opinion information type are as follows: retrieve the public opinion information within the query time range, and obtain the displayed word cloud diagram of the public opinion information within the query time range. Analyze the sentiment types of several displayed words in the displayed word cloud diagram based on the sentiment dictionary. Among them, the sentiment types include: a favorable degree of 1, a favorable degree of 0.5, a favorable degree of 0, a favorable degree of -0.5, and a favorable degree of -1. Count the number of all displayed words in the displayed word cloud diagram and the number of displayed words with a favorable degree greater than 0 among the sentiment types. Divide the number of displayed words with a favorable degree greater than 0 among the sentiment types by the number of all displayed words to obtain the favorable degree ratio.

[0096] Judge whether the favorable degree ratio is greater than 80%; if so, mark the corresponding public opinion information as positive; if not, judge whether the favorable degree ratio is lower than 20%; if so, mark the corresponding public opinion information as negative.

[0097] If not, mark the corresponding public opinion information as neutral. Count the number of likes DZL, the number of forwards ZFL, and the number of comments corresponding to the neutral public opinion information whose favorable degree ratio is in the interval [20%, 40%] ∪ [60%, 80%]. Eliminate the invalid comments in the number of comments. Mark the number of comments after eliminating the invalid comments as PLZ, where the invalid comments are: comments with less than three characters, repetitive comments, and comments with non-text content. Mark the number of invalid comments as WXL. ∪ represents the union symbol.

[0098] Through the formula Calculate the public opinion correction factor; where both α and β are correction coefficients.

[0099] Judge whether the public opinion correction factor is greater than 0; if so, replace the public opinion information marked as neutral with the mark of positive; if not, judge whether the public opinion correction factor is less than 0; if so, replace the public opinion information marked as neutral with the mark of negative; if not, do not replace the public opinion information marked as neutral.

[0100] For example, retrieve the public opinion information within the query time range, and obtain the displayed word cloud diagram of the public opinion information within the query time range, which contains a total of 25 displayed words. Based on the analysis of the displayed word cloud diagram using the sentiment dictionary, the sentiment types of the displayed words are obtained as follows: 7 with a favorable degree of 1, 8 with a favorable degree of 0.5, 3 with a favorable degree of 0, 4 with a favorable degree of -0.5, and 3 with a favorable degree of -1.

[0101] Count the number of all displayed words in the displayed word cloud diagram and the number of displayed words with a favorable degree greater than 0 among the sentiment types is 15. Then the favorable degree ratio is: 15 / 25 = 60%. Then mark the corresponding public opinion information as neutral.

[0102] Retrieve and count the neutral public opinion information whose favorability ratio is in the range of [20%, 40%] ∪ [60%, 80%], and count the like value corresponding to the neutral public opinion information as 3658, the forwarding value as 1279, and the number of comments; remove invalid comments from the number of comments; the number of comments after removing invalid comments is 2541; mark the number of invalid comments as 129;

[0103] By formula Calculate the public opinion correction factor Since 2.7 is greater than 0, the public opinion information type marked as neutral is replaced with positive.

[0104] See also Figure 5 The second aspect of the present invention provides a method for monitoring and analyzing public opinion of smart cultural and tourism big data, comprising:

[0105] Collect public opinion information from multiple communication media to build a public opinion information database; set keywords and filter out a number of public opinion information from the public opinion information database based on the keywords; sort the several public opinion information in chronological order; the keywords include: scenic spot name, custom query words; public opinion information includes: occurrence time, end time, scenic spot name and custom query words of the public opinion; and the occurrence time and end time are determined according to the popularity and click rate or search rate of the platform where the public opinion is located;

[0106] Construct and train a public opinion type judgment model; determine the type of public opinion information within the query time range according to the public opinion type judgment model; and,

[0107] Construct a public opinion correction factor to correct the type of public opinion information; the types of public opinion information include: positive, negative and neutral.

[0108] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0109] The working principle of the present invention is as follows: public opinion information is collected from multiple communication media to construct a public opinion information database; keywords are set and a number of public opinion information are screened out from the public opinion information database according to the keywords; the several public opinion information are sorted in chronological order; a public opinion type judgment model is constructed and trained; the type of public opinion information within the query time range is determined according to the public opinion type judgment model; and a public opinion correction factor is constructed to correct the type of public opinion information.

[0110] The above embodiments are only used to illustrate the technical method of the present invention and not to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A smart cultural tourism big data public opinion monitoring and analysis system, characterized in that: include: Public opinion collection module and the public opinion analysis module connected thereto; Public opinion collection module: collects public opinion information from multiple communication media to build a public opinion information database; Setting keywords and filtering out a number of pieces of public opinion information from the public opinion information database according to the keywords; sorting the pieces of public opinion information according to the order of occurrence; wherein the keywords include: scenic spot name, custom query words; Public opinion analysis module: build and train a public opinion type judgment model; determine the type of public opinion information within the query time range based on the public opinion type judgment model; and, Construct a public opinion correction factor to correct the type of public opinion information; the types of public opinion information include: positive, negative and neutral.

2. According to claim 1, a smart cultural tourism big data public opinion monitoring and analysis system is characterized by: Also includes: The public opinion feedback module connected to the public opinion analysis module is used to push the type of public opinion information to users and to update the type of public opinion information.

3. According to claim 1, a smart cultural tourism big data public opinion monitoring and analysis system is characterized by: The setting of keywords and screening out a number of pieces of public opinion information from the public opinion information database according to the keywords include: S110: Mark the scenic spot name as the first keyword A, and mark the custom query word as the second keyword Bj; wherein Bj represents the jth second keyword; j=1, 2, 3, ..., n; n is a positive integer; S120: Setting the query interval duration of public opinion information; S130: Set the jth marker format to Search all historical public opinion information from the public opinion information database in turn within the query time range, obtain the historical public opinion information corresponding to different markers, and count the number of historical public opinion information corresponding to different markers YQG j .

4. According to claim 3, a smart cultural tourism big data public opinion monitoring and analysis system is characterized in that: The setting of the query interval duration of public opinion information includes: Retrieve the public opinion information of the scenic spot from the public opinion information database, and count the number of negative public opinions k that occurred in the corresponding scenic spot within two years from the current time. The average duration of each historical public opinion is marked as CXS k ; Get the historical monthly income of the scenic spot when there is no public opinion; where k = 0, 1, 2, ..., n; n∈[0, +∞]; Determine whether the income of the scenic spot in the month when the public opinion information occurs is less than the preset low income threshold; if yes, the occurrence of the public opinion on the income of the local scenic spot is marked as having an impact SR, and SR = 1 is set; if no, the occurrence of the public opinion on the income of the local scenic spot is marked as having no impact SR, and SR = 0 is set; The time when the public opinion of several pieces of public opinion information ends is marked as T k ; By formula Calculate the range impact factor; Through the formula CQS = CXS k ×FYZ calculates the query interval duration.

5. According to claim 1, a smart cultural tourism big data public opinion monitoring and analysis system is characterized by: The construction and training of the public opinion type judgment model includes: S210: Retrieving historical public opinion information corresponding to different markers, and obtaining the type of historical public opinion information; S220: Retrieve the public opinion type judgment model, and mark the number of historical public opinion information corresponding to the marker j=1 as YQG1; The corresponding historical public opinion information is sequentially input into the public opinion type judgment model; it is judged whether the type of historical public opinion information output by the public opinion type judgment model is consistent with the type of marked historical public opinion information; if yes, the output result of the public opinion type judgment model is marked as accurate; if not, the output result of the public opinion type judgment model is marked as inaccurate; the number of public opinion information with accurate output results is counted and recorded as YQZ1; wherein, the public opinion type judgment model is constructed based on the artificial intelligence model; The accuracy rate of historical public opinion information corresponding to j=1 is calculated by the formula YQZ1 / 2YQG1; S230: The corresponding historical public opinion information is sequentially input into the public opinion type judgment model, and step S220 is repeated to obtain the accuracy rate of the historical public opinion information corresponding to different markers; S240: Use the number of keywords as the horizontal coordinate and the accuracy rate as the vertical coordinate to construct a rectangular coordinate system, count the total number of curve slopes between two adjacent points in the rectangular coordinate system, and the number of curve slopes between two adjacent points in the rectangular coordinate system that are greater than or equal to 0; determine whether the proportion of the number of curve slopes greater than or equal to 0 to the total number is greater than a proportion threshold; if yes, complete the training of the public opinion type judgment model; if no, take the first item as The tolerance is 2, and the number of historical public opinion information input into the public opinion type judgment model is increased successively until the proportion of the number of curve slopes greater than or equal to 0 to the total number is greater than the proportion threshold, and the training process is terminated.

6. According to claim 5, a smart cultural tourism big data public opinion monitoring and analysis system is characterized in that: The public opinion type judgment model is constructed based on an artificial intelligence model and includes: Obtain standard training data; integrate all historical public opinion information through word cloud generation technology to form a display word cloud, and integrate the display words in the display word cloud into a judgment factor sequence by arranging them in descending order according to font size; wherein the standard training data includes standard input data consistent with the content attributes of the judgment factor sequence, and standard output data consistent with the content attributes of the judgment factor sequence; the standard output data is the type of public opinion information; use the standard training data to train the artificial intelligence model; The trained artificial intelligence model is marked as a public opinion type judgment model; wherein the artificial intelligence model includes a convolutional neural network model or a long short-term memory neural network model.

7. According to claim 1, a smart cultural tourism big data public opinion monitoring and analysis system is characterized by: Determining the type of public opinion information within the query time range according to the public opinion type judgment model includes: Retrieve public opinion information within the query time range, and obtain a display word cloud diagram of the public opinion information within the query time range, and analyze the sentiment types of several display words in the display word cloud diagram based on the sentiment dictionary; wherein the sentiment types include: favorability a, favorability b, favorability 0, favorability -b, favorability -a; count the number of all display words in the display word cloud diagram and the number of display words with a favorability greater than 0 in the sentiment type; divide the number of display words with a favorability greater than 0 in the sentiment type by the number of all display words to obtain a favorability ratio; wherein, a>b>0; The types of public opinion information are divided according to comparison thresholds.

8. According to claim 1, a smart cultural tourism big data public opinion monitoring and analysis system is characterized by: The constructing of the public opinion correction factor to correct the type of public opinion information includes: Retrieve public opinion information marked as neutral, select the number of likes DZL, the number of reposts ZFL and the number of comments corresponding to the public opinion information type of neutral; remove invalid comments from the number of comments; mark the number of comments after removing invalid comments as PLZ; mark the number of invalid comments as WXL; By formula The public opinion correction factor is calculated; where α and β are correction coefficients; Determine whether the public opinion correction factor is greater than 0; if so, replace the public opinion information marked as neutral with the positive one; if not, determine whether the public opinion correction factor is less than 0; if so, replace the public opinion information marked as neutral with the negative one; if not, do not replace the public opinion information marked as neutral.

9. The intelligent cultural tourism big data public opinion monitoring and analysis system according to claim 8 is characterized in that: The method for obtaining the correction coefficient includes: Get the number of digits e of the like value DZL and the number of digits w of the forwarding value ZFL, The correction coefficient α is calculated as follows: α = 10 -e ; The correction coefficient β is calculated as follows: β = 10 -w ; Among them, the value range of e and w is [0, +∞].

10. A method for monitoring and analyzing public opinion of smart cultural and tourism big data, adapted to a system for monitoring and analyzing public opinion of smart cultural and tourism big data according to claims 1-9, characterized in that: include: Collect public opinion information from multiple communication media to build a public opinion information database; Set keywords and filter out several pieces of public opinion information from the public opinion information database based on the keywords; The plurality of public opinion information are sorted in chronological order; wherein the keywords include: scenic spot name, custom query words; Construct and train a public opinion type judgment model; determine the type of public opinion information within the query time range according to the public opinion type judgment model; and, Construct a public opinion correction factor to correct the type of public opinion information; the types of public opinion information include: positive, negative and neutral.

Images