Method, device and storage medium for analyzing net recommendation value

Abstract

Embodiments of the present application provide a method, device and storage medium for analyzing net recommendation value. The method comprises: obtaining a historical net recommendation dataset and a target questionnaire filled by a user, wherein the target questionnaire comprises a target survey item and corresponding survey data, the historical net recommendation dataset comprises a historical net recommendation value score corresponding to the target survey item and market environment data; performing sentiment analysis on the target questionnaire by using a natural language processing model to obtain a sentiment analysis result; analyzing the historical net recommendation dataset by using a deep learning model to obtain a predicted net recommendation value in a preset time period; and determining a net recommendation value score corresponding to the target survey item according to the sentiment analysis result and the predicted net recommendation value. The embodiments of the present application aim to quickly and accurately determine the net recommendation value of the target questionnaire by using the above method.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, in particular to a net recommendation value analysis method, a net recommendation value analysis device, a computer device and a computer readable storage medium. BACKGROUND

[0002] The net recommendation value in market research is an index for measuring the willingness of a customer to recommend a product or service to others. The net recommendation value is a customer satisfaction measure that directly reflects customer loyalty to a company. By analyzing the net recommendation value, the satisfaction of users at different times and in different regions can be seen, and the questionnaire distributor can improve subsequent production and operation methods based on the overall satisfaction obtained from the analysis.

[0003] Currently, although there are various data analysis methods for the net recommendation value, none of them can satisfy the questionnaire distributor or the enterprise. Due to the limited number of sample recoveries in questionnaire research, the influence of research time, research language, and other factors, it is impossible to guarantee the quality of sample recovery, and the recovery of research takes time, which leads to low accuracy of the net recommendation value obtained from the analysis and cannot accurately reflect user loyalty. Therefore, it is necessary to propose a net recommendation value analysis method that can quickly and accurately determine the net recommendation value of the target questionnaire. SUMMARY

[0004] The present application provides a net recommendation value analysis method, a net recommendation value analysis device, a computer device and a computer readable storage medium, which aims to quickly and accurately determine the net recommendation value of the target questionnaire through the above method.

[0005] To achieve the above purpose, the present application also provides a net recommendation value analysis method, comprising:

[0006] obtaining a historical net recommendation data set and a target questionnaire filled by a user, wherein the target questionnaire comprises a target survey item and corresponding survey data, and the historical net recommendation data set comprises a historical net recommendation value score corresponding to the target survey item and market environment data;

[0007] performing sentiment analysis on the target questionnaire by a natural language processing model to obtain a sentiment analysis result; and

[0008] analyzing the historical net recommendation data set by a deep learning model to obtain a predicted net recommendation value in a preset time period;

[0009] determining a net recommendation value score corresponding to the target survey item according to the sentiment analysis result and the predicted net recommendation value.

[0010] To achieve the above object, the present application further provides a net recommendation value analysis device, comprising:

[0011] an acquisition module, configured to acquire a historical net recommendation dataset and a target survey questionnaire filled by a user, wherein the target survey questionnaire comprises a target survey item and corresponding survey data, and the historical net recommendation dataset comprises a historical net recommendation value score corresponding to the target survey item and market environment data;

[0012] an analysis module, configured to perform sentiment analysis on the target survey questionnaire by using a natural language processing model to obtain a sentiment analysis result; and

[0013] a deep learning model is used to analyze the historical net recommendation dataset to obtain a predicted net recommendation value in a preset time period;

[0014] a determination module, configured to determine a net recommendation value score corresponding to the target survey item according to the sentiment analysis result and the predicted net recommendation value.

[0015] In addition, to achieve the above object, the present application further provides a computer device, comprising a memory and a processor; the memory is configured to store a computer program; the processor is configured to execute the computer program and implement the steps of the net recommendation value analysis method provided by any one of the embodiments of the present application when the computer program is executed.

[0016] In addition, to achieve the above object, the present application further provides a computer readable storage medium, which stores a computer program, and the computer program makes the processor implement the steps of the net recommendation value analysis method provided by any one of the embodiments of the present application when the computer program is executed by the processor.

[0017] The net recommendation value analysis method, the net recommendation value analysis device, the computer device and the computer readable storage medium are disclosed. The method comprises obtaining a historical net recommendation data set and a target questionnaire filled by a user. The target questionnaire comprises target survey items and corresponding survey data. The historical net recommendation data set comprises historical net recommendation value scores corresponding to the target survey items and market environment data. Further, the target questionnaire is subjected to sentiment analysis by a natural language processing model to obtain a sentiment analysis result. The historical net recommendation data set is analyzed by a deep learning model to obtain a predicted net recommendation value in a preset time period. Thus, the net recommendation value score corresponding to the target survey items can be determined according to the sentiment analysis result and the predicted net recommendation value. The natural language processing model and the deep learning model are combined to accurately identify the sentiment analysis result, and the future net recommendation value is predicted based on the historical net recommendation data set, thereby effectively supporting the scoring and optimization decision of the target survey items. The method not only improves the depth and accuracy of net recommendation value analysis, but also provides comprehensive sentiment insight and trend prediction, thereby providing a reliable basis for improving user satisfaction and optimizing market strategy. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0019] Figure 1 is a scene schematic diagram of a net recommendation value analysis method provided by an embodiment of the present application;

[0020] Figure 2 is a flow schematic diagram of a net recommendation value analysis method provided by an embodiment of the present application;

[0021] Figure 3 is a schematic block diagram of a net recommendation value analysis device provided by an embodiment of the present application;

[0022] Figure 4 is a schematic block diagram of a computer device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0023] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0024] The flowcharts shown in the drawings are merely illustrative and do not necessarily include all the contents and operations / steps, nor are they necessarily executed in the order described. For example, some operations / steps can be further divided, combined or partially merged, so the actual execution order can be changed according to actual conditions. In addition, although the functional modules are divided in the device schematic diagram, in some cases, the modules can be divided differently from the device schematic diagram.

[0025] The term “and / or” used in the present application and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0026] Some embodiments of the present application will be described in detail below with reference to the accompanying drawings. The following examples and features in the examples can be combined with each other without conflict.

[0027] As Figure 1 indicated, the net recommendation value analysis method provided by the embodiments of the present application can be applied to an application environment as Figure 1 indicated. The application environment includes a terminal device 110 and a server 120, wherein the terminal device 110 can communicate with the server 120 through a network. Specifically, the server 120 can obtain a historical net recommendation data set and a target survey questionnaire filled by a user, wherein the target survey questionnaire includes a target survey item and corresponding survey data, and the historical net recommendation data set includes a historical net recommendation value score corresponding to the target survey item and market environment data. Further, the target survey questionnaire can be analyzed by a natural language processing model to obtain an emotional analysis result, and the historical net recommendation data set can be analyzed by a deep learning model to obtain a predicted net recommendation value in a preset time period. Thus, the net recommendation value score corresponding to the target survey item can be determined according to the emotional analysis result and the predicted net recommendation value, and the net recommendation value score is sent to the terminal device 110. The server 120 can be an independent server, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms, etc. The terminal device 110 can be a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart speaker, a smart watch, etc., but is not limited thereto. The terminal and the server can be directly or indirectly connected through wired or wireless communication, which is not limited in the present application.

[0028] Please refer to Figure 2 , Figure 2is a flowchart of a method for analyzing a net recommendation value provided by an embodiment of the present application. As shown in Figure 2 The method for analyzing a net recommendation value can be implemented by steps S11 to S14.

[0029] Step S11: Obtain a historical net recommendation dataset and a target survey questionnaire filled out by a user.

[0030] The target survey questionnaire includes target survey items and corresponding survey data, and the historical net recommendation dataset includes historical net recommendation value scores corresponding to the target survey items and market environment data.

[0031] For example, the target survey questionnaire is data filled out by a user, including but not limited to target survey items, i.e., survey topics for specific scenarios or products. For example, a user satisfaction survey or a service experience survey for a certain product; survey data, i.e., specific feedback information filled out by a user, such as score data, text feedback, and basic information, etc., which provides specific inputs of a current user for subsequent sentiment analysis and prediction.

[0032] It should be noted that the historical net recommendation dataset is a basic dataset for modeling and analysis, including historical net recommendation value scores, i.e., historical score data related to the target survey items. For example, score data of similar survey items by different user groups at different times or in different time periods, which provides a reference basis for predicting future trends; market environment data, used to record external influencing factors related to the target survey items, such as seasonal factors, promotional activities, etc., which are used to capture the macro background affecting user feedback, so that the model learns more accurate prediction patterns.

[0033] Further, the present application does not limit the acquisition method of the target survey questionnaire, for example, including real-time acquisition, i.e., collecting user feedback through online forms, mobile applications, social media questionnaires, etc., or through formatted storage, i.e., converting survey data into structured data (such as JSON, CSV format) for subsequent processing.

[0034] In the embodiments of the present application, by obtaining the above data, high-quality inputs are provided for subsequent sentiment analysis and prediction, laying a solid foundation for the entire analysis process.

[0035] Step S12: Perform sentiment analysis on the target survey questionnaire by a natural language processing model to obtain a sentiment analysis result.

[0036] For example, the sentiment analysis result includes sentiment tendency, i.e., positive sentiment, such as very satisfied, product is very useful, service is very in place; negative sentiment, such as not satisfied, operation is very complicated, waiting time is too long; neutral sentiment, such as no feeling, etc. The sentiment analysis result can also include key factors affecting user sentiment, such as product function, service quality, price level, use experience, etc., which are not limited by the present application.

[0037] For example, the target questionnaire can be subjected to sentiment analysis by a natural language processing model (such as BERT, GPT, RoBERTa), such as classifying each text (positive, neutral, negative), and extracting topics or keywords in the target questionnaire, such as logistics, price, service attitude, etc., to obtain the sentiment analysis result.

[0038] The above implementation reduces the complexity of manual analysis by means of automatic sentiment analysis, can efficiently process a large amount of user feedback, and the sentiment analysis result can help enterprises understand user demand from the detail level and improve the scientificity and pertinence of decision-making.

[0039] Step S13: Analyzing the historical net recommendation data set by a deep learning model to obtain a predicted net recommendation value in a preset time period.

[0040] For example, the historical net recommendation data set can be analyzed by a deep learning model (such as LSTM or Transformer) to generate a predicted net recommendation value in a preset time period by learning the time series pattern and potential causal relationship in the data. The deep learning model can comprehensively consider various variables, including historical net recommendation value scores, market environment data, seasonal factors, etc., to mine long-term trends and short-term fluctuations of the data and output a predicted net recommendation value in a preset time period. The preset time period can be one month in the future, one week in the future, etc., which are not limited by the present application.

[0041] The above implementation can significantly improve the accuracy of net recommendation value trend analysis by using a deep learning model for prediction, helping enterprises to predict the changing trend of user satisfaction in advance, not only providing data support for formulating targeted market strategies, but also assisting in identifying key factors that may affect future scores, thereby optimizing products or services, improving user experience, and enhancing market competitiveness.

[0042] Step S14: Determining the net recommendation value score corresponding to the target survey project according to the sentiment analysis result and the predicted net recommendation value.

[0043] For example, the target survey project can be comprehensively evaluated in combination with the sentiment analysis result and the predicted net recommendation value predicted by the deep learning model. Since the sentiment analysis result reveals the key influencing factors of user satisfaction, and the predicted net recommendation value provides a predicted score in the future period of time, these information can be associated with the target survey project to correct or complete the net recommendation value score of the project, ensuring that the score is more comprehensive and representative.

[0044] The above embodiments can more accurately reflect the actual influence of the target survey project by combining sentiment analysis results and predicted scores, providing more insightful scoring basis for enterprises, not only improving the reliability of data analysis, but also better helping enterprises to position user needs, optimize services or products, and ultimately improve user satisfaction and market competitiveness.

[0045] The net recommendation value analysis method disclosed by the embodiments of the present application includes obtaining historical net recommendation data set and target survey questionnaire filled by users, wherein the target survey questionnaire includes target survey items and corresponding survey data, and the historical net recommendation data set includes historical net recommendation value scores corresponding to the target survey items and market environment data. Further, sentiment analysis of the target survey questionnaire can be performed by a natural language processing model to obtain sentiment analysis results, and analysis of the historical net recommendation data set can be performed by a deep learning model to obtain predicted net recommendation values in a preset time period. Thus, the net recommendation value score corresponding to the target survey project can be determined according to the sentiment analysis results and the predicted net recommendation values. By combining the natural language processing model and the deep learning model, the embodiments of the present application can accurately identify sentiment analysis results and predict future net recommendation values based on historical net recommendation data sets, thereby effectively supporting scoring and optimization decisions for target survey projects. This method not only improves the depth and accuracy of net recommendation value analysis, but also provides comprehensive sentiment insights and trend predictions, providing reliable basis for improving user satisfaction and optimizing market strategies.

[0046] Optionally, sentiment analysis of the target survey questionnaire is performed by a natural language processing model to obtain sentiment analysis results, including: performing preprocessing operations on the target survey questionnaire to obtain preprocessed target survey questionnaire; wherein the preprocessing operations include denoising operations, word segmentation operations and part-of-speech restoration operations; performing feature extraction operations on the preprocessed target survey questionnaire to obtain target feature vectors; analyzing the target feature vectors by the natural language processing model to obtain sentiment score information corresponding to the target survey questionnaire, and determining the sentiment analysis results based on the sentiment score information, wherein the sentiment analysis results include positive sentiment, negative sentiment and neutral sentiment.

[0047] For example, when performing sentiment analysis on the target questionnaire by the natural language processing model, the target questionnaire can be preprocessed first. The preprocessing includes denoising operation (such as removing special characters and irrelevant information), word segmentation operation (splitting continuous text into independent words or phrases), and part-of-speech reduction operation (reducing words to their basic forms), etc., which are not limited in the present application. In this way, the data can be cleaned and standardized to ensure the accuracy of subsequent analysis.

[0048] For example, the preprocessed target questionnaire can be converted into a target feature vector for analysis. The natural language processing model can analyze the target feature vector to generate sentiment score information. The sentiment score information quantifies the sentiment tendency contained in the target questionnaire, including the proportion of positive sentiment, negative sentiment, and neutral sentiment. Based on the sentiment score information, a sentiment analysis result can be further generated to identify the sentiment type and intensity of the user for a specific survey item, providing support for subsequent decision-making.

[0049] The above embodiments can automatically identify the sentiment tendency in user feedback through sentiment analysis of the target questionnaire, providing a direct basis for enterprises to quickly understand customer satisfaction. The sentiment analysis result not only supplements the shortcomings of quantitative scoring, but also helps enterprises to locate specific problems or advantages behind positive and negative feedback, so as to more targetedly optimize services or products and improve user experience and brand loyalty.

[0050] Based on the above embodiments, the sentiment analysis result is determined based on the sentiment score information, including: in response to determining that the sentiment score information is greater than a preset sentiment threshold, determining that the sentiment analysis result is positive sentiment; or, in response to determining that the sentiment score information is less than the preset sentiment threshold, determining that the sentiment analysis result is negative sentiment; or, in response to determining that the sentiment score information is the same as the preset sentiment threshold, determining that the sentiment analysis result is neutral sentiment.

[0051] For example, positive sentiment is used to reflect the optimistic sentiment of the target questionnaire; negative sentiment is used to reflect the pessimistic sentiment of the target questionnaire; and neutral sentiment is used to reflect content that is neutral or cannot be classified.

[0052] It should be noted that the present application does not limit the preset threshold, for example, the preset threshold is 50%, 60%, etc. The present application takes 50% as an example for illustration. In response to determining that the sentiment score information is greater than 50%, the sentiment analysis result is determined to be positive sentiment; or, in response to determining that the sentiment score information is less than 50%, the sentiment analysis result is determined to be negative sentiment; or, in response to determining that the sentiment score information is the same as 50%, the sentiment analysis result is determined to be neutral sentiment.

[0053] The above embodiments convert complex emotional information into clear positive, negative and neutral emotional categories through quantitative analysis, providing an efficient tool for enterprises to quickly interpret user feedback emotions. Positive emotional results help enterprises identify successes and further strengthen strengths; negative emotional results reveal problems and guide improvement directions; neutral emotional results provide basic information support for ambiguous feedback, avoiding ignoring potential problems. Overall, the method improves the accuracy and practicality of sentiment analysis, providing stronger data support for optimizing decisions.

[0054] Optionally, the natural language processing model is obtained by: obtaining a training data set and a pre-trained model, wherein the training data set includes a plurality of questionnaires; performing sentiment label annotation on the training data set to obtain an annotation result, wherein the annotation result includes a sentiment analysis result corresponding to each questionnaire; and training the pre-trained model based on the training data set and the annotation result to obtain the natural language processing model.

[0055] It should be noted that the natural language processing model is not limited in the present application, for example, including BERT, BiLSTM, CRF, GPT and ELMo, etc. The present application takes CRF as an example for illustration.

[0056] The CRF (conditional random field algorithm) model is a kind of probability graph model, which is often used in sequence labeling or sentiment analysis tasks. The CRF model is a kind of undirected graph model, which is used to model the conditional probability distribution between the label sequence and the input sequence. The main idea of CRF model is to use the context information to jointly label the elements in the sequence. It considers the dependence between adjacent elements, as well as the conditional dependence between the features of the current element and the label. By learning the relationship between features and labels in the training data, the CRF model can predict the most likely label for each input element, thus completing the sentiment analysis task. In addition, the CRF model considers the context information and conditional dependence in the sentiment analysis task, which improves the performance of the model and makes the sentiment analysis result more accurate.

[0057] Specifically, a training data set containing a plurality of questionnaires can be collected for training. For example, the training data set can be obtained by manual collection, web crawler or public data set, etc. The present application does not limit this.

[0058] Further, each survey questionnaire can be labeled to obtain a sentiment analysis result corresponding to each survey questionnaire. Then, the sentiment analysis result is taken as a label of the group of input data, and each group of training data set with the label is input into the CRF model for supervised learning. When a training end condition is met, such as the number of training times reaches a threshold or the output accuracy of the model reaches an accuracy threshold, the training is ended, and a trained natural language processing model is obtained.

[0059] In the embodiments of the present application, the training data set and the labeled result are input into the CRF model for supervised learning, and then a natural language processing model is trained. Thus, the sentiment analysis result can be output based on the natural language processing model.

[0060] The above embodiments can enhance the data quality and diversity during the training of the natural language processing model, and improve the generalization ability and actual application effect of the model.

[0061] On the basis of the above embodiments, after the pre-training model is trained based on the training data set and the labeled result to obtain the natural language processing model, the following steps are included: the natural language processing model is iteratively trained based on the training data set and the labeled result to extract data features and calculate a loss function; the loss function is iteratively trained by using a preset method to reduce the value of the loss function until a desired threshold condition is met; and the iteratively trained natural language processing model is obtained based on the iteratively trained loss function.

[0062] It can be understood that, in order to train a natural language processing model with higher accuracy, the natural language processing model can be iteratively trained in a manner that reduces the loss function until the loss function meets the desired threshold condition, and then a more accurate sentiment analysis result can be obtained based on the iteratively trained natural language processing model.

[0063] It should be noted that the present application does not limit the above-mentioned preset method and desired threshold, for example, the preset method can be a gradient descent algorithm, a batch gradient descent algorithm, a stochastic gradient descent algorithm, etc., and the present application takes the gradient descent algorithm as an example for illustration.

[0064] The purpose of the gradient descent algorithm is to find the minimum value of the loss function by iteration, or to converge to the minimum value. Geometrically, the gradient descent algorithm is to find the minimum value of the function by moving in the opposite direction of the vector with the fastest increase in the function, so that the gradient decreases the fastest and the minimum value of the function is easier to find. Therefore, in the embodiments of the present application, the gradient descent algorithm is used to iteratively train the natural language processing model to reduce the loss function, thereby reducing the error of the calculation result.

[0065] In the embodiments of the present application, the gradient descent algorithm is used to iteratively train the natural language processing model, so that the loss function is continuously reduced to obtain the iterated natural language processing model, and then a more accurate sentiment analysis result can be obtained based on the iterated natural language processing model.

[0066] Optionally, the historical net recommendation data set is analyzed by the deep learning model to obtain a predicted net recommendation value in a preset time period, including: constructing time series data based on historical net recommendation value scores and market environment data; processing the time series data by the deep learning model to obtain the predicted net recommendation value in the preset time period.

[0067] For example, when the historical net recommendation data set is analyzed by the deep learning model, the historical net recommendation value scores and the market environment data can be organized as time series data to show the trend and pattern over time. Then, the time series data is processed by the deep learning model (such as LSTM or Transformer) to mine the correlation and potential rules in the historical data, and finally generate the predicted net recommendation value in the preset time period, reflecting the trend of net recommendation score in the future period.

[0068] The above embodiments use historical data and market environment change information by the deep learning model to accurately predict the trend of future net recommendation value, providing the enterprise with the ability to understand the changes in customer satisfaction in advance. Therefore, the enterprise can optimize resource allocation and develop targeted customer strategies to improve customer experience and enhance market competitiveness, while reducing potential risks caused by fluctuations in satisfaction.

[0069] Optionally, after determining the net recommendation value score corresponding to the target survey project according to the sentiment analysis result and the predicted net recommendation value, including: generating a net recommendation value analysis report based on the net recommendation value score, and displaying the net recommendation value analysis report through a display page; wherein the net recommendation value analysis report includes the net recommendation value score, the performance information of the submarket, and the improvement suggestion information.

[0070] For example, after determining the net recommendation value score of the target survey project according to the sentiment analysis result and the predicted net recommendation value, a net recommendation value analysis report can be further generated. The net recommendation value analysis report can summarize the net recommendation value score, the performance information of the submarket (such as the net recommendation value trend of different customer groups or regions), and the improvement suggestion (optimization strategy based on sentiment analysis and score result). Finally, the report can be intuitively presented through a display page, facilitating users to quickly understand the core information.

[0071] The above implementation methods, by generating and displaying Net Promoter Score (NPS) analysis reports, enable a comprehensive understanding of historical customer satisfaction performance and specific market segment conditions, thereby identifying potential problems and opportunities. Furthermore, by incorporating improvement suggestions, targeted optimization measures can be developed to enhance customer experience, strengthen customer relationship management, and gain a competitive edge in the fierce market.

[0072] Please see Figure 3 , Figure 3 This is a schematic block diagram of a net recommender value (NPV) analysis apparatus provided in an embodiment of this application. The NPV analysis apparatus can be configured in a server to execute the aforementioned NPV analysis method.

[0073] like Figure 3 As shown, the net recommendation value analysis device 200 includes: an acquisition module 201, an analysis module 202, and a determination module 203.

[0074] The acquisition module 201 is used to acquire the historical net promoter dataset and the target survey questionnaire filled out by the user. The target survey questionnaire includes target survey items and corresponding survey data, and the historical net promoter dataset includes the historical net promoter score and market environment data corresponding to the target survey items.

[0075] Analysis module 202 is used to perform sentiment analysis on the target questionnaire using a natural language processing model to obtain sentiment analysis results; and,

[0076] The historical net recommender dataset is analyzed using a deep learning model to obtain the predicted net recommender value for a preset time period.

[0077] The determination module 203 is used to determine the net promoter score corresponding to the target survey item based on the sentiment analysis results and the predicted net promoter score.

[0078] The analysis module 202 is further configured to perform preprocessing operations on the target questionnaire to obtain a preprocessed target questionnaire; wherein the preprocessing operations include denoising, word segmentation, and part-of-speech tagging; perform feature extraction on the preprocessed target questionnaire to obtain a target feature vector; analyze the target feature vector through the natural language processing model to obtain the sentiment score information corresponding to the target questionnaire, and determine the sentiment analysis result based on the sentiment score information, wherein the sentiment analysis result includes positive sentiment, negative sentiment, and neutral sentiment.

[0079] The analysis module 202 is further configured to determine that the sentiment analysis result is the positive sentiment in response to determining that the sentiment score information is greater than a preset sentiment threshold; or determine that the sentiment analysis result is the negative sentiment in response to determining that the sentiment score information is less than the preset sentiment threshold; or determine that the sentiment analysis result is the neutral sentiment in response to determining that the sentiment score information is the same as the preset sentiment threshold.

[0080] The acquisition module 201 is further configured to acquire a training data set and a pre-trained model, wherein the training data set includes a plurality of questionnaires; perform sentiment label annotation on the training data set to obtain an annotation result, wherein the annotation result includes a sentiment analysis result corresponding to each questionnaire; and train the pre-trained model based on the training data set and the annotation result to obtain the natural language processing model.

[0081] The acquisition module 201 is further configured to perform iterative training on the natural language processing model based on the training data set and the annotation result to extract data features and calculate a loss function; perform iterative training on the loss function by using a preset method for the purpose of reducing the loss function value until a preset threshold condition is met; and obtain an iterated natural language processing model based on the iterated loss function.

[0082] The analysis module 202 is further configured to construct time series data based on the historical net recommendation value score and the market environment data; and process the time series data by using the deep learning model to obtain a predicted net recommendation value of the preset time period.

[0083] The determination module 203 is further configured to generate a net recommendation value analysis report based on the net recommendation value score, and display the net recommendation value analysis report by using a display page; wherein the net recommendation value analysis report includes the net recommendation value score, market segment performance information and improvement suggestion information.

[0084] It should be noted that, for the convenience and brevity of description, the specific working process of the above-described device and each module and unit can refer to the corresponding process in the foregoing method embodiments, which will not be described herein.

[0085] The method and device of the present application can be used in many general or special-purpose computing system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and the like.

[0086] Exemplarily, the method and the device described above can be implemented as a computer program in a form of a computer program product, which can run on the computer device as shown in the figures. Figure 4

[0087] Please refer to Figure 4 , Figure 4 is a schematic diagram of a computer device provided by an embodiment of the present application. The computer device can be a server.

[0088] As shown in the figure, the computer device includes a processor, a memory and a network interface connected through a system bus, wherein the memory can include a volatile storage medium, a non-volatile storage medium and an internal memory. Figure 4

[0089] The non-volatile storage medium can store an operating system and a computer program. The computer program includes program instructions, which, when executed, can cause the processor to perform any kind of analysis method of the net recommendation value.

[0090] The processor is used to provide computing and control capabilities to support the operation of the entire computer device.

[0091] The internal memory provides an environment for the execution of the computer program in the non-volatile storage medium, which, when executed by the processor, can cause the processor to perform any kind of analysis method of the net recommendation value.

[0092] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art can understand that the structure of the computer device is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0093] It should be understood that the processor can be a central processing unit (CPU), and the processor can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0094] ​​In some embodiments, the processor is configured to obtain a historical net recommendation dataset and a target survey questionnaire filled by a user, wherein the target survey questionnaire comprises target survey items and corresponding survey data, the historical net recommendation dataset comprises historical net recommendation value scores corresponding to the target survey items and market environment data; perform sentiment analysis on the target survey questionnaire by using a natural language processing model to obtain a sentiment analysis result; and analyze the historical net recommendation dataset by using a deep learning model to obtain a predicted net recommendation value for a preset time period; and determine a net recommendation value score corresponding to the target survey items based on the sentiment analysis result and the predicted net recommendation value.

[0095] In some embodiments, the processor is further configured to perform a preprocessing operation on the target survey questionnaire to obtain a preprocessed target survey questionnaire, wherein the preprocessing operation comprises a denoising operation, a word segmentation operation, and a part-of-speech restoration operation; perform a feature extraction operation on the preprocessed target survey questionnaire to obtain a target feature vector; analyze the target feature vector by using the natural language processing model to obtain sentiment score information corresponding to the target survey questionnaire, and determine the sentiment analysis result based on the sentiment score information, wherein the sentiment analysis result comprises positive sentiment, negative sentiment, and neutral sentiment.

[0096] In some embodiments, the processor is further configured to determine that the sentiment analysis result is the positive sentiment in response to determining that the sentiment score information is greater than a preset sentiment threshold; or determine that the sentiment analysis result is the negative sentiment in response to determining that the sentiment score information is less than the preset sentiment threshold; or determine that the sentiment analysis result is the neutral sentiment in response to determining that the sentiment score information is the same as the preset sentiment threshold.

[0097] In some embodiments, the processor is further configured to obtain a training dataset and a pre-trained model, wherein the training dataset comprises a plurality of survey questionnaires; perform sentiment label annotation on the training dataset to obtain an annotation result, wherein the annotation result comprises a sentiment analysis result corresponding to each survey questionnaire; and train the pre-trained model by using the training dataset and the annotation result to obtain the natural language processing model.

[0098] In some embodiments, the processor is further configured to iteratively train the natural language processing model based on the training dataset and the annotation result to extract data features and calculate a loss function; iteratively train the loss function by using a preset method until a desired threshold condition is met, with the purpose of reducing the value of the loss function; and obtain an iteratively trained natural language processing model based on the iteratively trained loss function.

[0099] In some embodiments, the processor is further configured to construct time series data based on the historical net recommendation value score and the market environment data; and process the time series data by the deep learning model to obtain the predicted net recommendation value of the preset time period.

[0100] In some embodiments, the processor is further configured to generate a net recommendation value analysis report based on the net recommendation value score, and display the net recommendation value analysis report through a display page; wherein the net recommendation value analysis report comprises the net recommendation value score, market segment performance information, and improvement suggestion information.

[0101] The embodiments of the present application further provide a computer readable storage medium, and the computer readable storage medium stores a computer program. The computer program comprises program instructions, and the program instructions are executed to implement any of the analysis methods of the net recommendation value provided by the embodiments of the present application.

[0102] The computer readable storage medium can be an internal storage unit of the computer device, for example, a hard disk or a memory of the computer device. The computer readable storage medium can also be an external storage device of the computer device, for example, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc.

[0103] Further, the computer readable storage medium can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required by a function, etc.

[0104] The above merely illustrates the specific implementation of the present application, but the protection scope of the present application is not limited to this. Any person skilled in the art can easily think of various equivalent modifications or replacements within the technical range disclosed by the present application, and these modifications or replacements should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method of analyzing net promoter scores, the method comprising: The method comprises: obtaining a historical net recommendation data set and a target survey questionnaire filled by a user, wherein the target survey questionnaire comprises a target survey item and corresponding survey data, the historical net recommendation data set comprises a historical net recommendation value score corresponding to the target survey item and market environment data; performing sentiment analysis on the target survey questionnaire by a natural language processing model to obtain a sentiment analysis result; and analyzing the historical net recommendation data set by a deep learning model to obtain a predicted net recommendation value in a preset time period; determining a net recommendation value score corresponding to the target survey item according to the sentiment analysis result and the predicted net recommendation value.

2. The method of claim 1, wherein, The sentiment analysis on the target survey questionnaire by the natural language processing model to obtain a sentiment analysis result comprises: performing a preprocessing operation on the target survey questionnaire to obtain a preprocessed target survey questionnaire; wherein the preprocessing operation comprises a denoising operation, a word segmentation operation and a part-of-speech restoration operation; performing a feature extraction operation on the preprocessed target survey questionnaire to obtain a target feature vector; analyzing the target feature vector by the natural language processing model to obtain sentiment score information corresponding to the target survey questionnaire, and determining the sentiment analysis result based on the sentiment score information, wherein the sentiment analysis result comprises positive sentiment, negative sentiment and neutral sentiment.

3. The method of claim 2, wherein, The determination of the sentiment analysis result based on the sentiment score information comprises: in response to determining that the sentiment score information is greater than a preset sentiment threshold, determining that the sentiment analysis result is the positive sentiment; or in response to determining that the sentiment score information is less than the preset sentiment threshold, determining that the sentiment analysis result is the negative sentiment; or in response to determining that the sentiment score information is the same as the preset sentiment threshold, determining that the sentiment analysis result is the neutral sentiment.

4. The method of claim 2, wherein, The natural language processing model is obtained by: obtaining a training data set and a pre-training model, wherein the training data set comprises a plurality of survey questionnaires; performing sentiment label annotation on the training data set to obtain an annotation result, wherein the annotation result comprises a sentiment analysis result corresponding to each survey questionnaire; training the pre-training model by the training data set and the annotation result to obtain the natural language processing model.

5. The method of claim 4, wherein, After the training of the pre-training model by the training data set and the annotation result to obtain the natural language processing model, the following steps are performed: based on the training data set and the annotation result, iteratively training the natural language processing model to extract data features and calculating a loss function; iteratively training the loss function by a preset method to reduce the loss function value until a desired threshold is met; based on the iteratively trained loss function, obtaining an iteratively trained natural language processing model.

6. The method of claim 1, wherein, The analysis of the historical net recommendation data set by the deep learning model to obtain a predicted net recommendation value in a preset time period comprises: constructing time series data based on the historical net recommendation value score and the market environment data; The time series data is processed by the deep learning model to obtain a predicted net recommendation value of the preset time period.

7. The method of claim 1, wherein, After determining the net recommendation value score corresponding to the target survey item according to the sentiment analysis result and the predicted net recommendation value, the method further includes: generating a net recommendation value analysis report based on the net recommendation value score and displaying the net recommendation value analysis report through a display page, wherein the net recommendation value analysis report includes the net recommendation value score, market segment performance information, and improvement suggestion information.

8. An analysis device of a net recommendation value, characterized by The analysis device includes: an acquisition module configured to acquire a historical net recommendation dataset and a target survey questionnaire filled by a user, wherein the target survey questionnaire includes target survey items and corresponding survey data, and the historical net recommendation dataset includes historical net recommendation value scores corresponding to the target survey items and market environment data; an analysis module configured to perform sentiment analysis on the target survey questionnaire through a natural language processing model to obtain a sentiment analysis result; and a deep learning model configured to analyze the historical net recommendation dataset to obtain a predicted net recommendation value of a preset time period; a determination module configured to determine a net recommendation value score corresponding to the target survey item according to the sentiment analysis result and the predicted net recommendation value.

9. A computer device, comprising: includes: a memory and a processor; wherein the memory is connected to the processor and is used to store programs, and the processor is used to realize the steps of the net recommendation value analysis method according to any one of claims 1-7 by running the programs stored in the memory.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program is executed by the processor to make the processor realize the steps of the net recommendation value analysis method according to any one of claims 1-7.

Images