Public opinion monitoring methods, devices, electronic equipment and non-volatile storage media

Abstract

This application discloses a method, apparatus, electronic device, and non-volatile storage medium for public opinion monitoring. It relates to the field of artificial intelligence. The method includes: extracting word segmentation vectors from target text data and clustering the word segmentation vectors to obtain multiple target clusters; determining individuals in a first population when the target parameters of the target clusters do not meet preset parameter conditions; iterating the individuals in the first population multiple times through selection, crossover, and mutation operations, and performing simulated annealing on the individuals in the first population after each iteration until the number of iterations reaches a preset number of genetic iterations to obtain individuals in a second population; and sending a target alarm message when the growth rate of the number of individuals in the second population exceeds a preset growth rate threshold. This application solves the technical problem of poor efficiency in public opinion monitoring caused by the current scattered nature of voice agent dialogue information, requiring manual screening and analysis.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and more specifically, to a public opinion monitoring method, device, electronic device, and non-volatile storage medium. Background Technology

[0002] Telephone banking, as a traditional channel for customer feedback, can effectively collect customer issues. However, because telephone banking is entirely voice-based, the data is not structured and cannot effectively extract trending keywords, making it difficult to effectively identify sudden public opinion events.

[0003] Currently, voice agent dialogue information is scattered, requiring manual post-event screening and analysis, and has technical problems such as large recognition workload and poor efficiency in public opinion monitoring.

[0004] There is currently no effective solution to the above problems. Summary of the Invention

[0005] This application provides a public opinion monitoring method, device, electronic device, and non-volatile storage medium to at least solve the technical problem of poor public opinion monitoring efficiency caused by the current scattered voice agent dialogue information requiring manual screening and analysis.

[0006] According to one aspect of the embodiments of this application, a public opinion monitoring method is provided, comprising: extracting word segmentation vectors from target text data and clustering the word segmentation vectors to obtain multiple target clusters, wherein the target text data is converted from call voice data within a preset time period; determining a first population individual when the target parameters of the target cluster do not meet preset parameter conditions, wherein each first population individual corresponds to one target text data, and the preset parameter conditions are the target parameters corresponding to the target cluster when the public opinion heat value of the call voice data does not exceed a preset heat threshold; iterating the first population individual multiple times through selection, crossover, and mutation operations, and performing simulated annealing operation on the iterated first population individual after each iteration until the number of iterations reaches a preset number of genetic iterations to obtain a second population individual; sending a target alarm message when the growth rate of the number of target population individuals in the second population exceeds a preset growth rate threshold, wherein the number of target population individuals is the number of second population individuals corresponding to the same target cluster in the second population individual.

[0007] Optionally, the process of iterating through selection, crossover, and mutation operations on the first population individuals includes: determining the first fitness of each first population individual; performing a selection operation on the first population individuals based on the first fitness and a preset selection rule; performing a crossover operation on the first population individuals after the selection operation based on the crossover operator; performing a mutation operation on the first population individuals after the crossover operation based on the mutation operator; determining the second fitness of the new generation of first population individuals obtained after the crossover, mutation, selection, and simulated annealing operations; and repeating the crossover, mutation, and selection operations until the number of iterations exceeds the preset number of genetic iterations.

[0008] Optionally, the selection operation for individuals in the first population based on the first fitness and the preset selection rules includes: determining the target probability corresponding to each individual in the first population based on the first fitness, wherein the target probability is used to characterize the probability that the individual in the first population is selected into the next generation population; determining the cumulative probability corresponding to each individual in the first population based on the target probability, wherein the cumulative probability is used to indicate the probability interval corresponding to the individual in the first population; generating a first random number within the preset value interval, and selecting the individual in the first population corresponding to the probability interval where the first random number is located into the next generation population.

[0009] Optionally, the first group of individuals contains multiple feature data corresponding to the target text data; the crossover operation on the selected first group of individuals according to the crossover operator includes: determining an even number of first group individuals to be crossovered according to the crossover probability indicated by the crossover operator; randomly pairing the even number of first group individuals; and exchanging a random number of feature data in the two first group individuals that have been paired.

[0010] Optionally, performing a mutation operation on the first population individuals after the crossover operation according to the mutation operator includes: determining the first population individuals to be mutated according to the mutation probability indicated by the mutation operator; generating a second random number, wherein the second random number is a random number that conforms to a normal distribution with a mean of a preset mean and a variance of a preset variance; and randomly selecting a feature data from the first population individuals to be mutated and replacing it with the second random number.

[0011] Optionally, performing simulated annealing on the iterated first population individuals includes: determining a target temperature value; determining a first evaluation value for the iterated first population individuals based on an evaluation function, wherein the first evaluation value characterizes the quality of the first population individuals; randomly perturbing the iterated first population individuals, wherein the perturbing operation is used to change the characteristic data of the first population individuals; determining a second evaluation value for the perturbed first population individuals based on an evaluation function; determining whether to replace the unperturbed first population individuals with the perturbed first population individuals based on the first and second evaluation values; and reducing the target temperature value according to a preset temperature decrease rate; repeating the above steps until the target temperature value is not greater than a preset temperature threshold.

[0012] Optionally, determining whether to replace the unperturbed individuals of the first population with the perturbated individuals based on the first evaluation value and the second evaluation value includes: replacing the unperturbed individuals of the first population with the perturbated individuals when the first evaluation value is not greater than the second evaluation value; and determining the target replacement probability when the first evaluation value is less than the second evaluation value, and randomly determining whether to replace the unperturbed individuals of the first population with the perturbated individuals according to the target replacement probability, wherein the target replacement probability is determined based on the first evaluation value, the second evaluation value, and a preset probability generation function.

[0013] Optionally, the target parameters include at least: the clustering category of the target cluster and the frequency of word segmentation vectors in each target cluster. The preset parameter conditions include: the preset clustering category and the preset frequency threshold corresponding to the target cluster of the preset clustering category. When the target parameters of the target cluster do not meet the preset parameter conditions, determining the first group of individuals includes: when there is a target cluster that does not belong to the preset clustering category or when there is a target cluster whose word segmentation vector frequency is greater than the corresponding preset frequency threshold, determining the first group of individuals based on the target text data within a preset time period, wherein one target text data corresponds to one call voice data.

[0014] According to another aspect of the embodiments of this application, a public opinion monitoring device is also provided, comprising: a data clustering module, used to extract word segmentation vectors from target text data and cluster the word segmentation vectors to obtain multiple target clusters, wherein the target text data is converted from call voice data within a preset time period; a first judgment module, used to determine a first population individual when the target parameters of the target cluster do not meet preset parameter conditions, wherein each first population individual corresponds to one target text data, and the preset parameter conditions are the target parameters corresponding to the target cluster when the public opinion heat value of the call voice data does not exceed a preset heat threshold; a genetic iteration module, used to iterate the first population individual multiple times through selection, crossover and mutation operations, and after each iteration, perform simulated annealing operation on the iterated first population individual until the number of iterations reaches a preset number of genetic iterations to obtain a second population individual; and a second judgment module, used to send a target alarm information when the growth rate of the number of target population individuals of the second population individuals exceeds a preset growth rate threshold, wherein the number of target population individuals is the number of second population individuals corresponding to the same target cluster among the second population individuals.

[0015] According to another aspect of the embodiments of this application, an electronic device is also provided, including: a memory and a processor, the processor being configured to run a program stored in the memory, wherein the program executes a public opinion monitoring method during runtime.

[0016] According to another aspect of the embodiments of this application, a non-volatile storage medium is also provided, the non-volatile storage medium including a stored computer program, wherein the device where the non-volatile storage medium is located executes a public opinion monitoring method by running the computer program.

[0017] In this embodiment, word segmentation vectors are extracted from the target text data and clustered to obtain multiple target clusters. The target text data is converted from call voice data within a preset time period. If the target parameters of the target cluster do not meet the preset parameter conditions, a first group of individuals is determined. Each first group of individuals corresponds to one piece of target text data. The preset parameter conditions are the target parameters corresponding to the target cluster when the public opinion heat value of the call voice data does not exceed a preset heat threshold. The first group of individuals is iterated multiple times through selection, crossover, and mutation operations. After each iteration, the iterated first group of individuals undergoes simulated annealing until the number of iterations reaches a preset number of genetic iterations, resulting in a second group of individuals. If the growth rate of the number of target group individuals in the second group exceeds a preset growth rate threshold, a target alarm message is sent. The number of target group individuals is the number of second group individuals corresponding to the same target cluster. The call recording is extracted and transcribed offline. Then, NLP (Natural Language Processing) technology is used to semantically summarize the call text and classify hot words. By using genetic algorithms to predict trending topics and simulated annealing to correct data, the goal of quickly and efficiently identifying public opinion information is achieved. This solves the technical problem of poor public opinion monitoring efficiency caused by the current scattered voice agent dialogue information, which requires manual screening and analysis. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0019] Figure 1 This is a schematic diagram of a public opinion monitoring method provided according to an embodiment of this application;

[0020] Figure 2 This is a schematic diagram of a speech recognition process according to an embodiment of this application;

[0021] Figure 3 This is a schematic diagram illustrating the principle of a natural language processing technology according to an embodiment of this application;

[0022] Figure 4 This is a schematic diagram of a simulated annealing method according to an embodiment of this application;

[0023] Figure 5 This is a schematic diagram of the structure of a public opinion monitoring device according to an embodiment of this application;

[0024] Figure 6This is a hardware structure block diagram of an electronic device for implementing a method for public opinion monitoring, according to an embodiment of this application; Detailed Implementation

[0025] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0026] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0027] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0028] It should be noted that all information (including but not limited to users' voice call data, users' personal information, etc.) and data (including but not limited to data used for display, data used for analysis, etc.) involved in this disclosure are information and data authorized by users or fully authorized by all parties. For example, this system has an interface with relevant users or organizations. Before obtaining relevant information, it is necessary to send an acquisition request to the aforementioned users or organizations through the interface, and obtain the relevant information after receiving consent information from the aforementioned users or organizations.

[0029] The present invention will now be described with reference to preferred implementation steps. This application provides a method for monitoring public opinion. Figure 1 This is a schematic diagram of a public opinion monitoring method provided in an embodiment of this application, such as... Figure 1 As shown, the method includes the following steps:

[0030] Step S102: Extract word segmentation vectors from the target text data and cluster the word segmentation vectors to obtain multiple target clusters. The target text data is converted from call voice data within a preset time period.

[0031] As an optional implementation, the customer's telephone banking call recordings (i.e., the aforementioned call voice data) can be extracted using Automatic Speech Recognition (ASR), transcribed, and then categorized using Natural Language Processing (NLP) to obtain initial text data. Automatic Speech Recognition is a technology that automatically converts human speech into text. The specific speech recognition process is as follows: Figure 2 As shown; Natural Language Processing (NLP) is a technology that studies how computers process human language. Its core is to convert human natural language into instructions that computers can read, that is, the machine's ability to understand and interpret human writing and speaking. Figure 3 This is a schematic diagram illustrating the principle of a natural language processing technology according to an embodiment of this application, such as... Figure 3 As shown.

[0032] In the context of public opinion analysis, basic NLP capabilities such as word segmentation, sentence encoding, keyword extraction, and text clustering are mainly used to process the dialogue text obtained from ASR. The technology relied upon is a pre-trained model plus fine-tuning.

[0033] The initial text data obtained from the conversion of voice call data belongs to natural language data. When processing the data, the unstructured text data needs to be converted into structured data through encoding.

[0034] Specifically, based on the structural characteristics of the original sample data of telephone agent voice text, the initial text data is preliminarily cleaned and filtered. In this embodiment, at least the following filtering rules are included: Initial text data corresponding to call voice data with fewer than 5 dialogue rounds, as well as dirty data containing many incomplete dialogues, are considered invalid calls and deleted; words in the initial text data that have no actual meaning (e.g., "um," "ba," "ah," "ai," "ai," etc.) are deleted; words in the initial text data with a frequency of less than 3 in the system's preset text corpus are considered invalid words and deleted; dialogue content with a single text character length of less than 5 (not expressing effective meaning) and dialogue content with a single text character length greater than 200 (expressing lengthy information) are both considered invalid dialogue content and deleted.

[0035] After initial screening of the initial text data according to the aforementioned filtering rules, the target text data is obtained. Then, feature extraction is performed on the non-normalized target text data to transform it into normalized data. Specifically, in this embodiment, the BERT model can be used for text encoding. The BERT model converts each character in the target text data into a one-dimensional vector by querying a word vector table, which serves as the model input. The model output is the word segmentation vector corresponding to the input text, after incorporating the semantic information of the entire text.

[0036] As an optional implementation, the target text data can be divided into target text data originating from customer voice and target text data originating from customer service voice according to its source. In this embodiment, cluster analysis is mainly performed on the word segmentation vectors in the target text data originating from customer voice. Specifically, at least one of the following clustering methods can be used: Cluster analysis based on BERT-Embedding: K-means clustering is performed on the word segmentation vectors obtained after processing by the BERT model; Cluster analysis based on BERT-AVG-Embedding: the average word vectors obtained after processing by the BERT model are calculated to obtain the average word vectors, and K-means clustering is performed on the average word vectors.

[0037] After clustering the word segmentation vectors, multiple target clusters are obtained. Each target cluster contains at least one of the following features: clusterable category (e.g., pension, credit card, personal loan), word segmentation content (e.g., pension-personal pension, credit card-overdue, personal loan-mortgage, personal loan-car loan), frequency of word segmentation vectors in each target cluster (e.g., number of times word segments appear in a target cluster of a certain clustering category per hour), caller characteristics (e.g., age range, income range, call frequency), etc.

[0038] Step S104: If the target parameters of the target cluster do not meet the preset parameter conditions, determine the first type of individual, wherein each first type of individual corresponds to one target text data, and the preset parameter conditions are the target parameters corresponding to the target cluster when the public opinion heat value of the call voice data does not exceed the preset heat threshold.

[0039] In the technical solution provided in step S104, the target parameters include at least: the clustering category of the target cluster and the frequency of word vectors in each target cluster. The preset parameter conditions include: the preset clustering category and the preset frequency threshold corresponding to the target cluster of the preset clustering category.

[0040] In some embodiments of this application, when the target parameters of the target cluster do not meet the preset parameter conditions, determining the first type of individual includes the following steps: when there is a target cluster that does not belong to the preset clustering category or when the frequency of the word segmentation vector in a target cluster is greater than the corresponding preset frequency threshold, the first type of individual is determined based on the target text data within a preset time period, wherein one target text data corresponds to one voice call data.

[0041] Specifically, the system acquires the regular interval classification (i.e., the aforementioned preset clustering categories) and frequency of occurrence (i.e., the aforementioned preset frequency threshold) of the word segmentation vectors. When comparing the word segmentation vectors of the staged clustering with the reserve information (i.e., the aforementioned preset parameter conditions), and significant differences in clustering categories and high-frequency phenomena are found, it is determined that there may be a situation where the public opinion heat exceeds the heat threshold. In this case, based on the genetic algorithm and simulated annealing optimization algorithm, the target text data corresponding to the customer's call voice data within a preset time period is analyzed. By using the feature information of each target text data (e.g., the clustering category corresponding to the word segmentation vector in the target text data, the customer's age range, income range, call frequency, etc. in the call corresponding to the text data), the system tests the frequency of calls of that clustering category occurring per unit time within a future time period (e.g., 1 hour, 1 day, 1 week).

[0042] Step S106: The first population individuals are iterated multiple times through selection, crossover and mutation operations, and after each iteration, the first population individuals are simulated annealing until the number of iterations reaches the preset number of genetics, and the second population individuals are obtained.

[0043] In related technologies, genetic algorithms are used to predict public opinion. However, genetic algorithms have poor local search capabilities, are prone to premature convergence, and are easily trapped in local optima. Simulated annealing algorithms offer fast global optimization and are more tolerant of short-term non-optimal solutions, but they are slower. This application combines the two algorithms, using a genetic algorithm based on big data to quickly predict popularity and using simulated annealing for local data correction. This approach has faster convergence and optimization, enabling rapid and efficient identification of public opinion information.

[0044] In some embodiments of this application, iterating through selection, crossover, and mutation operations on the first population individuals multiple times includes the following steps: determining the first fitness of each first population individual; performing a selection operation on the first population individuals based on the first fitness and a preset selection rule; performing a crossover operation on the first population individuals after the selection operation based on the crossover operator; performing a mutation operation on the first population individuals after the crossover operation based on the mutation operator; determining the second fitness of the new generation of first population individuals obtained after the crossover operation, mutation operation, selection operation, and simulated annealing operation; repeating the crossover operation, mutation operation, and selection operation until the number of iterations is greater than the preset number of genetic iterations.

[0045] In some embodiments of this application, each individual in the population contains multiple feature data corresponding to the target text data, as shown below:

[0046] X i =(x 1i ,x 2i ,...,x ni ), i∈(1,2,...,N)

[0047] X i x represents a single individual in the first population, and there are N individuals in the population. 1i ,x 2i ,...,x ni This indicates that there are n characteristic data points for each individual in the population, where N and n are both positive integers.

[0048] Specifically, a fitting function is used as the fitness function to calculate the fitness (first fitness, second fitness, etc.) of each individual in the population at each iteration. In this embodiment, the fitting function can be expressed by the following formula:

[0049]

[0050] Where C is a constant, e1 to e n k1 to k n The fitness parameters are calculated for each feature data point, and the genetic algorithm performs selection based on the fitness value of each individual. The higher the fitness, the greater the probability of being selected and passed on to the next generation.

[0051] During the selection process, individuals with higher population fitness are chosen based on roulette wheel selection. Simultaneously, an elite preservation strategy is employed to ensure that desirable chromosomes are passed down and prevent them from being replaced by roulette wheel selection.

[0052] Specifically, the selection operation for individuals in the first population based on the first fitness and the preset selection rules includes the following steps: determining the target probability corresponding to each individual in the first population based on the first fitness, wherein the target probability is used to characterize the probability that the individual in the first population will be selected into the next generation population; determining the cumulative probability corresponding to each individual in the first population based on the target probability, wherein the cumulative probability is used to indicate the probability interval corresponding to the individual in the first population; generating a first random number within the preset value interval, and selecting the individual in the first population corresponding to the probability interval where the first random number is located into the next generation population.

[0053] To introduce better genes, this application proposes to introduce variation through a crossover operator and to avoid the genetic algorithm getting trapped in local optima through a mutation operator and simulated annealing. Specifically, the crossover operation involves exchanging one or more pieces of information (feature data) between two selected parent individuals (population individuals) according to a presumed probability (crossover probability); the mutation operation involves replacing the original feature data of the population individuals with a random number derived from a normally distributed Gaussian mutation (i.e., the second random number mentioned above).

[0054] In some embodiments of this application, performing a crossover operation on the first population individuals after selection based on the crossover operator includes the following steps: determining an even number of first population individuals to be crossovered based on the crossover probability indicated by the crossover operator; randomly pairing the even number of first population individuals; and exchanging a random number of feature data from the two paired first population individuals.

[0055] In some embodiments of this application, performing a mutation operation on the first population individuals after crossover based on the mutation operator includes: determining the first population individuals to be mutated based on the mutation probability indicated by the mutation operator; generating a second random number, wherein the second random number is a random number that conforms to a normal distribution with a mean of a preset mean and a variance of a preset variance; and randomly selecting a feature data from the first population individuals to be mutated and replacing it with the second random number.

[0056] To improve the accuracy of public opinion monitoring, this application also uses simulated annealing algorithm to penetrate the mutation of chromosomes (i.e. individuals in the first population) in the genetic algorithm, thereby increasing variability, avoiding local optima in the genetic algorithm, and obtaining more accurate predictions.

[0057] Simulated annealing mimics the process of constant cooling and finding the optimal point. The initial temperature is T (the target temperature mentioned above), at which point the solid oscillates at its maximum. As the temperature decreases, the solid gradually finds its optimal point and tends to stabilize. Starting from a relatively high temperature, and with the continuous decrease of the temperature parameter, combined with a certain probability, the jump characteristic randomly searches for the global optimum of the objective function in the solution space. That is, it can probabilistically jump out of local optima and eventually tend towards the global optimum.

[0058] In some embodiments of this application, the simulated annealing operation on the iterated first population individuals includes the following steps: determining a target temperature value; determining a first evaluation value for the iterated first population individuals based on an evaluation function, wherein the first evaluation value is used to characterize the quality of the first population individuals; randomly perturbing the iterated first population individuals, wherein the perturbing operation is used to change the characteristic data of the first population individuals; determining a second evaluation value for the perturbed first population individuals based on an evaluation function; determining whether to replace the unperturbed first population individuals with the perturbed first population individuals based on the first evaluation value and the second evaluation value; and reducing the target temperature value according to a preset temperature decrease rate; repeating the above steps until the target temperature value is not greater than a preset temperature threshold.

[0059] In some embodiments of this application, determining whether to replace the first population individual without perturbation operation with the first population individual after perturbation operation based on the first evaluation value and the second evaluation value includes the following steps: if the first evaluation value is not greater than the second evaluation value, replace the first population individual without perturbation operation with the first population individual after perturbation operation; if the first evaluation value is less than the second evaluation value, determine the target replacement probability, and randomly determine whether to replace the first population individual without perturbation operation with the first population individual after perturbation operation according to the target replacement probability, wherein the target replacement probability is determined based on the first evaluation value, the second evaluation value, and a preset probability generation function.

[0060] Figure 4 This is a schematic diagram of a simulated annealing method flow provided according to an embodiment of this application, such as... Figure 4 As shown.

[0061] The current temperature T (i.e., the target temperature value mentioned above) is compared with the threshold temperature Tmin (i.e., the preset temperature threshold mentioned above). If Tmin > T, a perturbation operation is performed on the population individuals. If the new solution (the first population individual after the perturbation operation) is better than the current solution (the first population individual without the perturbation operation), the new solution is accepted; otherwise, the Metropolis criterion is used to determine whether to accept the new solution. The target replacement probability is as follows:

[0062]

[0063] Where P represents the target replacement probability (the target replacement probability is considered to be 1 if the first evaluation value is not greater than the second evaluation value), and E t+1 E represents the second evaluation value. t The first evaluation value is represented by k, the preset temperature decrease rate is represented by T, and the target temperature value is represented by T.

[0064] For example, let T = T0, representing the initial temperature at which annealing begins, i.e., the initial target temperature value. Determine the initial solution space (e0,k0)t (i.e., the first population individual after iteration) and calculate the corresponding objective function value E(e0,k0) (i.e., the first evaluation value mentioned above, where E() is the evaluation function mentioned above). Let T = kT, where k (i.e., the temperature decrease rate mentioned above) takes a value between 0.5 and 0.9. Apply a random perturbation (perturbation operation) to the current solution (e0,k0)t, generating a new solution (e0,k0)t+1 (i.e., the first population individual after perturbation operation) in its neighborhood, and calculate the corresponding objective function value E((e0,k0)t+1) (i.e., the second evaluation value mentioned above). Calculate ΔE = E((e0,k0)t+1) - E((e0,k0)t). If ΔE < 0, accept the new solution as the current solution; otherwise, determine whether to accept the new solution according to the target replacement probability mentioned above. At temperature T, repeat the perturbation and acceptance process multiple times, and determine whether the temperature has reached the termination temperature level. If so, terminate the algorithm; otherwise, repeat the calculation.

[0065] Step S108: If the growth rate of the number of individuals in the target population of the second population exceeds the preset growth rate threshold, a target alarm message is sent, wherein the number of individuals in the target population is the number of individuals in the second population that correspond to the same target cluster.

[0066] Specifically, for the second population of individuals obtained, it is determined whether there is a situation where the number of individuals in a certain cluster category (i.e., corresponding to the same target cluster) is growing too rapidly. For populations with a growth rate exceeding a preset growth rate threshold, such as a population that doubles in number in one hour, the word segmentation vectors in the target cluster of the corresponding cluster category are given attention.

[0067] As an optional implementation method, it's also possible to filter content that focuses on men's interests, women's interests, and so on. By capturing word segmentation data and retrieving the original complete dialogue, the context and background of the focus can be quickly understood. This facilitates experts in developing scripts or allows banks to adjust strategies promptly. For example, regarding mortgage repayments, banks can determine whether to actively provide smooth channels or slow down the repayment pace based on their development plans. Similarly, for inquiries about opening personal pension accounts, convenient and clear guidance can be provided, such as through mini-programs and public accounts. By leveraging the risk of public opinion fluctuations and combining this with publicly available online information, it's possible to better capture trending topics, analyze, handle, guide, and even shut down such situations in a timely manner.

[0068] Through the above steps, the call recordings are extracted and transcribed offline. Then, NLP (Natural Language Processing) technology is used to semantically summarize the call text and categorize trending words. A genetic algorithm is used to predict trending information, and simulated annealing is used for data correction. This achieves the goal of quickly and efficiently identifying public opinion information, thus solving the technical problem of poor efficiency in public opinion monitoring caused by the current scattered nature of voice agent dialogue information, which requires manual screening and analysis.

[0069] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0070] According to an embodiment of this application, an embodiment of a public opinion monitoring device is also provided. Figure 5 This is a structural schematic diagram of a public opinion monitoring device provided according to an embodiment of this application. Figure 5 As shown, the device includes:

[0071] The data clustering module 50 is used to extract word segmentation vectors from the target text data and cluster the word segmentation vectors to obtain multiple target clusters. The target text data is converted from call voice data within a preset time period.

[0072] The first judgment module 52 is used to determine the first type of individual when the target parameters of the target cluster do not meet the preset parameter conditions. Each first type of individual corresponds to a target text data. The preset parameter conditions are the target parameters corresponding to the target cluster when the public opinion heat value of the call voice data does not exceed the preset heat threshold.

[0073] The genetic iteration module 54 is used to iterate the first population individuals multiple times through selection, crossover and mutation operations, and after each iteration, simulate annealing is performed on the first population individuals after the iteration until the number of iterations reaches the preset number of genetic iterations to obtain the second population individuals.

[0074] The second judgment module 56 is used to send a target alarm message when the growth rate of the number of individuals in the target population of the second population exceeds a preset growth rate threshold, wherein the number of individuals in the target population is the number of individuals in the second population that correspond to the same target cluster.

[0075] The public opinion monitoring device provided in this application extracts call recordings and performs offline transcription. Then, it uses NLP (Natural Language Processing) technology to semantically summarize the call text and categorize trending words. Based on a genetic algorithm, it predicts trending information and uses simulated annealing for data correction, achieving rapid and efficient identification of public opinion information. This solves the technical problem of poor public opinion monitoring efficiency caused by the current scattered nature of voice agent dialogue information, requiring manual verification and analysis.

[0076] It should be noted that each module in the aforementioned public opinion monitoring device can be a program module (e.g., a set of program instructions to implement a specific function) or a hardware module. For the latter, it can take the following forms, but is not limited to them: each of the above modules is represented by a processor, or the functions of each of the above modules are implemented by a processor.

[0077] It should be noted that the public opinion monitoring device provided in this embodiment can be used to perform... Figure 1 The public opinion monitoring method shown above is also applicable to the embodiments of this application, and will not be repeated here.

[0078] The public opinion monitoring device includes a processor and a memory. All the above-mentioned modules are stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize the corresponding functions.

[0079] The processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured; by adjusting kernel parameters, the goal of quickly and efficiently identifying public opinion information can be achieved.

[0080] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0081] This application embodiment also provides a non-volatile storage medium, which includes a stored computer program. The device containing the non-volatile storage medium executes the following public opinion monitoring method by running the computer program: extracting word segmentation vectors from target text data and clustering the word segmentation vectors to obtain multiple target clusters. The target text data is converted from call voice data within a preset time period. If the target parameters of a target cluster do not meet preset parameter conditions, a first population individual is determined. Each first population individual corresponds to one piece of target text data. The preset parameter conditions are the target parameters corresponding to the target cluster when the public opinion heat value of the call voice data does not exceed a preset heat threshold. The first population individuals are iterated multiple times through selection, crossover, and mutation operations. After each iteration, simulated annealing is performed on the iterated first population individuals until the number of iterations reaches a preset number of genetic iterations, resulting in a second population individual. If the growth rate of the target population individuals in the second population exceeds a preset growth rate threshold, a target alarm message is sent. The number of target population individuals is the number of second population individuals corresponding to the same target cluster within the second population individuals.

[0082] like Figure 6 As shown, an embodiment of the present invention provides an electronic device, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs the following steps:

[0083] Extract word segmentation vectors from the target text data and cluster them to obtain multiple target clusters. The target text data is converted from call voice data within a preset time period. If the target parameters of the target clusters do not meet the preset parameter conditions, determine the first population individuals. Each first population individual corresponds to one piece of target text data. The preset parameter conditions are the target parameters corresponding to the target clusters when the public opinion heat value of the call voice data does not exceed a preset heat threshold. Iterate the first population individuals multiple times through selection, crossover, and mutation operations. After each iteration, perform simulated annealing on the iterated first population individuals until the number of iterations reaches a preset number of inheritances to obtain the second population individuals. If the growth rate of the target population individuals in the second population exceeds a preset growth rate threshold, send a target alarm message. The number of target population individuals is the number of second population individuals corresponding to the same target cluster in the second population.

[0084] In some embodiments of this application, iterating through selection, crossover, and mutation operations on the first population individuals multiple times includes: determining the first fitness of each first population individual; performing a selection operation on the first population individuals based on the first fitness and a preset selection rule; performing a crossover operation on the first population individuals after the selection operation based on a crossover operator; performing a mutation operation on the first population individuals after the crossover operation based on a mutation operator; determining the second fitness of the new generation of first population individuals obtained after the crossover operation, mutation operation, selection operation, and simulated annealing operation; and repeating the crossover operation, mutation operation, and selection operation until the number of iterations is greater than a preset number of genetic iterations.

[0085] In some embodiments of this application, the selection operation for individuals in the first population based on a first fitness and a preset selection rule includes: determining a target probability for each individual in the first population based on the first fitness, wherein the target probability is used to characterize the probability that the individual in the first population is selected into the next generation population; determining a cumulative probability for each individual in the first population based on the target probability, wherein the cumulative probability is used to indicate the probability interval for the individual in the first population; generating a first random number within a preset value interval, and selecting the individual in the first population corresponding to the probability interval where the first random number is located into the next generation population.

[0086] In some embodiments of this application, the first group of individuals contains multiple feature data corresponding to the target text data; performing a crossover operation on the first group of individuals after selection based on the crossover operator includes: determining an even number of first group individuals to be crossovered based on the crossover probability indicated by the crossover operator; randomly pairing the even number of first group individuals; and exchanging a random number of feature data in the two first group individuals that have been paired.

[0087] In some embodiments of this application, performing a mutation operation on the first population individuals after crossover based on the mutation operator includes: determining the first population individuals to be mutated based on the mutation probability indicated by the mutation operator; generating a second random number, wherein the second random number is a random number that conforms to a normal distribution with a mean of a preset mean and a variance of a preset variance; and randomly selecting a feature data from the first population individuals to be mutated and replacing it with the second random number.

[0088] In some embodiments of this application, performing simulated annealing on the iterated first population individuals includes: determining a target temperature value; determining a first evaluation value for the iterated first population individuals based on an evaluation function, wherein the first evaluation value is used to characterize the quality of the first population individuals; randomly perturbing the iterated first population individuals, wherein the perturbing operation is used to change the characteristic data of the first population individuals; determining a second evaluation value for the perturbed first population individuals based on an evaluation function; determining whether to replace the unperturbed first population individuals with the perturbed first population individuals based on the first evaluation value and the second evaluation value; and reducing the target temperature value according to a preset temperature decrease rate; repeating the above steps until the target temperature value is not greater than a preset temperature threshold.

[0089] In some embodiments of this application, determining whether to replace the first population individual without perturbation operation with the first population individual after perturbation operation based on the first evaluation value and the second evaluation value includes: replacing the first population individual without perturbation operation with the first population individual after perturbation operation when the first evaluation value is not greater than the second evaluation value; determining the target replacement probability when the first evaluation value is less than the second evaluation value, and randomly determining whether to replace the first population individual without perturbation operation with the first population individual after perturbation operation according to the target replacement probability, wherein the target replacement probability is determined based on the first evaluation value, the second evaluation value, and a preset probability generation function.

[0090] In some embodiments of this application, the target parameters include at least: the clustering category of the target cluster and the frequency of word segmentation vectors in each target cluster. The preset parameter conditions include: a preset clustering category and a preset frequency threshold corresponding to the target cluster of the preset clustering category. When the target parameters of the target cluster do not meet the preset parameter conditions, determining the first group individual includes: when there is a target cluster that does not belong to the preset clustering category or when there is a target cluster whose word segmentation vector frequency is greater than the corresponding preset frequency threshold, determining the first group individual based on the target text data within a preset time period, wherein one target text data corresponds to one call voice data.

[0091] The devices mentioned in this article can be servers, PCs, tablets, mobile phones, etc.

[0092] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program that initializes the following method steps:

[0093] Extract word segmentation vectors from the target text data and cluster them to obtain multiple target clusters. The target text data is converted from call voice data within a preset time period. If the target parameters of the target clusters do not meet the preset parameter conditions, determine the first population individuals. Each first population individual corresponds to one piece of target text data. The preset parameter conditions are the target parameters corresponding to the target clusters when the public opinion heat value of the call voice data does not exceed a preset heat threshold. Iterate the first population individuals multiple times through selection, crossover, and mutation operations. After each iteration, perform simulated annealing on the iterated first population individuals until the number of iterations reaches a preset number of inheritances to obtain the second population individuals. If the growth rate of the target population individuals in the second population exceeds a preset growth rate threshold, send a target alarm message. The number of target population individuals is the number of second population individuals corresponding to the same target cluster in the second population.

[0094] In some embodiments of this application, iterating through selection, crossover, and mutation operations on the first population individuals multiple times includes: determining the first fitness of each first population individual; performing a selection operation on the first population individuals based on the first fitness and a preset selection rule; performing a crossover operation on the first population individuals after the selection operation based on a crossover operator; performing a mutation operation on the first population individuals after the crossover operation based on a mutation operator; determining the second fitness of the new generation of first population individuals obtained after the crossover operation, mutation operation, selection operation, and simulated annealing operation; and repeating the crossover operation, mutation operation, and selection operation until the number of iterations is greater than a preset number of genetic iterations.

[0095] In some embodiments of this application, the selection operation for individuals in the first population based on a first fitness and a preset selection rule includes: determining a target probability for each individual in the first population based on the first fitness, wherein the target probability is used to characterize the probability that the individual in the first population is selected into the next generation population; determining a cumulative probability for each individual in the first population based on the target probability, wherein the cumulative probability is used to indicate the probability interval for the individual in the first population; generating a first random number within a preset value interval, and selecting the individual in the first population corresponding to the probability interval where the first random number is located into the next generation population.

[0096] In some embodiments of this application, the first group of individuals contains multiple feature data corresponding to the target text data; performing a crossover operation on the first group of individuals after selection based on the crossover operator includes: determining an even number of first group individuals to be crossovered based on the crossover probability indicated by the crossover operator; randomly pairing the even number of first group individuals; and exchanging a random number of feature data in the two first group individuals that have been paired.

[0097] In some embodiments of this application, performing a mutation operation on the first population individuals after crossover based on the mutation operator includes: determining the first population individuals to be mutated based on the mutation probability indicated by the mutation operator; generating a second random number, wherein the second random number is a random number that conforms to a normal distribution with a mean of a preset mean and a variance of a preset variance; and randomly selecting a feature data from the first population individuals to be mutated and replacing it with the second random number.

[0098] In some embodiments of this application, performing simulated annealing on the iterated first population individuals includes: determining a target temperature value; determining a first evaluation value for the iterated first population individuals based on an evaluation function, wherein the first evaluation value is used to characterize the quality of the first population individuals; randomly perturbing the iterated first population individuals, wherein the perturbing operation is used to change the characteristic data of the first population individuals; determining a second evaluation value for the perturbed first population individuals based on an evaluation function; determining whether to replace the unperturbed first population individuals with the perturbed first population individuals based on the first evaluation value and the second evaluation value; and reducing the target temperature value according to a preset temperature decrease rate; repeating the above steps until the target temperature value is not greater than a preset temperature threshold.

[0099] In some embodiments of this application, determining whether to replace the first population individual without perturbation operation with the first population individual after perturbation operation based on the first evaluation value and the second evaluation value includes: replacing the first population individual without perturbation operation with the first population individual after perturbation operation when the first evaluation value is not greater than the second evaluation value; determining the target replacement probability when the first evaluation value is less than the second evaluation value, and randomly determining whether to replace the first population individual without perturbation operation with the first population individual after perturbation operation according to the target replacement probability, wherein the target replacement probability is determined based on the first evaluation value, the second evaluation value, and a preset probability generation function.

[0100] In some embodiments of this application, the target parameters include at least: the clustering category of the target cluster and the frequency of word segmentation vectors in each target cluster. The preset parameter conditions include: a preset clustering category and a preset frequency threshold corresponding to the target cluster of the preset clustering category. When the target parameters of the target cluster do not meet the preset parameter conditions, determining the first group individual includes: when there is a target cluster that does not belong to the preset clustering category or when there is a target cluster whose word segmentation vector frequency is greater than the corresponding preset frequency threshold, determining the first group individual based on the target text data within a preset time period, wherein one target text data corresponds to one call voice data.

[0101] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0102] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0103] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0104] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0105] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0106] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0107] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0108] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0109] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0110] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for monitoring public opinion, characterized in that, include: Extract word segmentation vectors from the target text data and cluster the word segmentation vectors to obtain multiple target clusters. The target text data is converted from call voice data within a preset time period. If the target parameters of the target cluster do not meet the preset parameter conditions, a first group individual is determined, wherein each first group individual corresponds to one target text data, and the preset parameter conditions are the target parameters corresponding to the target cluster when the public opinion heat value of the call voice data does not exceed a preset heat threshold. The target parameters include at least: the clustering category of the target cluster and the frequency of the word segmentation vector in each target cluster. The preset parameter conditions include: a preset clustering category and a preset frequency threshold corresponding to the target cluster of the preset clustering category. When the target parameters of the target cluster do not meet the preset parameter conditions, determining the first group individual includes: when there is a target cluster that does not belong to the preset clustering category or when the frequency of the word segmentation vector in the target cluster is greater than the corresponding preset frequency threshold, determining the first group individual based on the target text data within the preset time period, wherein one target text data corresponds to one call voice data. The first population of individuals is iterated multiple times through selection, crossover, and mutation operations. After each iteration, the individuals in the first population of individuals are subjected to simulated annealing until the number of iterations reaches the preset number of genetic steps, thus obtaining the second population of individuals. If the growth rate of the target population of the second population individuals exceeds a preset growth rate threshold, a target alarm message is sent, wherein the target population individuals are the number of second population individuals corresponding to the same target cluster in the second population individuals.

2. The method according to claim 1, characterized in that, Multiple iterations of the first population of individuals are performed through selection, crossover, and mutation operations, including: Determine the first fitness of each individual in the first population; Based on the first fitness and the preset selection rules, a selection operation is performed on the individuals in the first population; The crossover operation is performed on the first population individuals that have undergone the selection operation according to the crossover operator; The mutation operation is performed on the individuals in the first population after the crossover operation according to the mutation operator; Determine the second fitness of the first population individuals of the new generation obtained after the crossover, mutation, and selection operations, as well as the simulated annealing operation; Repeat the crossover, mutation, and selection operations until the number of iterations exceeds the preset number of inheritance iterations.

3. The method according to claim 2, characterized in that, The selection operation for individuals in the first population, based on the first fitness and the preset selection rule, includes: Based on the first fitness, a target probability is determined for each individual in the first population, wherein the target probability is used to characterize the probability that the individual in the first population is selected into the next generation population; Based on the target probability, the cumulative probability corresponding to each individual in the first population is determined, wherein the cumulative probability is used to indicate the probability interval corresponding to the individual in the first population; Within a preset value range, a first random number is generated, and individuals from the first population corresponding to the probability range where the first random number is located are selected for the next generation population.

4. The method according to claim 2, characterized in that, The first group of individuals contains multiple feature data corresponding to the target text data; Performing a crossover operation on the first population individuals after the selection operation according to the crossover operator includes: Based on the crossover probability indicated by the crossover operator, determine an even number of individuals from the first population to perform the crossover operation; The even number of individuals from the first population are randomly paired, and a random number of the feature data from the two paired individuals from the first population are exchanged.

5. The method according to claim 4, characterized in that, The mutation operation performed on the individuals in the first population after the crossover operation according to the mutation operator includes: Based on the mutation probability indicated by the mutation operator, determine the first population individual to perform the mutation operation; Generate a second random number, wherein the second random number is a random number that conforms to a normal distribution with a mean of a preset mean and a variance of a preset variance; In the first population individuals undergoing the mutation operation, one of the characteristic data is randomly selected and replaced with the second random number.

6. The method according to claim 1, characterized in that, The simulated annealing operation on the first population individuals after iteration includes: Determine the target temperature value; Based on the evaluation function, the first evaluation value of the first population individual after iteration is determined, wherein the first evaluation value is used to characterize the quality of the first population individual; The first population individuals after iteration are randomly perturbed, wherein the perturbing operation is used to change the feature data of the first population individuals; Based on the evaluation function, determine the second evaluation value of the first population individual after the perturbation operation; Based on the first evaluation value and the second evaluation value, determine whether to replace the first population individuals that have not undergone the perturbation operation with the first population individuals that have undergone the perturbation operation; and, The target temperature value is reduced according to a preset temperature decrease rate; Repeat the above steps until the target temperature value is not greater than the preset temperature threshold.

7. The method according to claim 6, characterized in that, Determining whether to replace individuals in the first population that have not undergone the perturbation operation with individuals from the first population that have undergone the perturbation operation, based on the first evaluation value and the second evaluation value, includes: If the first evaluation value is not greater than the second evaluation value, the first population individual that has not undergone the perturbation operation is replaced by the first population individual that has undergone the perturbation operation; If the first evaluation value is less than the second evaluation value, a target replacement probability is determined, and a random determination is made according to the target replacement probability to replace the first population individual that has not undergone the perturbation operation with the first population individual that has undergone the perturbation operation. The target replacement probability is determined based on the first evaluation value, the second evaluation value and a preset probability generation function.

8. A public opinion monitoring device, characterized in that, include: The data clustering module is used to extract word segmentation vectors from target text data and cluster the word segmentation vectors to obtain multiple target clusters. The target text data is converted from call voice data within a preset time period. The first judgment module is used to determine a first group individual when the target parameters of the target cluster do not meet the preset parameter conditions, wherein each first group individual corresponds to one target text data, and the preset parameter conditions are the target parameters corresponding to the target cluster when the public opinion heat value of the call voice data does not exceed the preset heat threshold. The target parameters include at least: the clustering category of the target cluster and the frequency of the word segmentation vector in each target cluster. The preset parameter conditions include: a preset clustering category and a preset frequency threshold corresponding to the target cluster of the preset clustering category. When the target parameters of the target cluster do not meet the preset parameter conditions, determining the first group individual includes: when there is a target cluster that does not belong to the preset clustering category or when the frequency of the word segmentation vector in the target cluster is greater than the corresponding preset frequency threshold, determining the first group individual based on the target text data within the preset time period, wherein one target text data corresponds to one call voice data. The genetic iteration module is used to iterate the first population individuals multiple times through selection, crossover and mutation operations, and after each iteration, simulate annealing is performed on the first population individuals after the iteration until the number of iterations reaches the preset number of genetic iterations to obtain the second population individuals; The second judgment module is used to send a target alarm message when the growth rate of the number of individuals in the target population of the second population exceeds a preset growth rate threshold, wherein the number of individuals in the target population is the number of individuals in the second population that correspond to the same target cluster.

9. An electronic device, characterized in that, include: A memory and a processor, the processor being configured to run a program stored in the memory, wherein the program, when running, executes the public opinion monitoring method according to any one of claims 1 to 7.

10. A non-volatile storage medium, characterized in that, The non-volatile storage medium includes a stored computer program, wherein the device containing the non-volatile storage medium executes the public opinion monitoring method according to any one of claims 1 to 7 by running the computer program.

Images