A method and device for analyzing feedback of a cigarette user
By collecting and analyzing feedback data and user profiles from cigarette users, and utilizing techniques such as association rules and support vector machines, the cigarette preferences of user groups can be accurately identified. This solves the problem of inaccurate market forecasting caused by ignoring consumer feedback in existing technologies, and enables precise analysis and optimization of the cigarette market.
Patent Information
- Application Number
- CN202411805367.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-10
- Publication Date
- 2025-11-21
- Estimated Expiration
- 2044-12-10
AI Technical Summary
In existing technologies, cigarette market behavior predictions often rely on sales data and ignore consumer feedback, leading to inaccurate market forecasts.
By collecting feedback data and user profiles from cigarette users, we use association rule algorithms to mine the correspondence between user profiles and rating results, segment user groups, analyze user preferences, and combine cigarette component dispersion and support vector machine to classify users and accurately identify the true preferences of user groups.
It enables precise identification and quantification of cigarette user preferences, solves the problem of inaccurate market forecasting, and can optimize cigarette brands and ingredients based on user feedback, thereby improving the accuracy of market forecasting.
Smart Images
Figure CN119762144B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of data information analysis technology, and in particular to a method and apparatus for analyzing feedback from cigarette users. Background Technology
[0002] Cigarettes, as a special commodity circulating in the market, require strict management and restrictions on their production and sales; however, as a commodity, analyzing user feedback can effectively grasp consumer needs, improve product quality, and predict market trends, enabling the sales process to better serve consumers.
[0003] Currently, market behavior forecasts for cigarettes mostly focus on sales data, neglecting consumer feedback and leading to inaccurate predictions of the cigarette market. Summary of the Invention
[0004] Therefore, it is necessary to provide a method and apparatus for analyzing feedback from cigarette users to address the aforementioned technical problems.
[0005] Firstly, this application provides a method for analyzing feedback from cigarette users, the method comprising:
[0006] We collected feedback data from multiple cigarette users regarding various cigarette brands, as well as user profiles for each user; user profiles included smoking history, smoking frequency, and income level.
[0007] The feedback data is parsed according to preset rules to obtain the rating results of the corresponding cigarette brands, and the correspondence between user profiles and rating results is mined based on the association rule algorithm.
[0008] Based on the corresponding relationships, each cigarette user is divided into multiple user groups to determine the cigarette preferences of each user group.
[0009] In one embodiment, the step of parsing the feedback data according to preset rules to obtain the rating result of the corresponding cigarette brand includes:
[0010] Identify the types of feedback data; types of feedback data include cigarette scores and cigarette evaluations.
[0011] When the feedback data is cigarette rating, the cigarette rating is normalized to correspond to the rating result, and the normalized cigarette rating is converted into a rating result.
[0012] When the feedback data is cigarette reviews, the evaluation analysis strategy includes: comparing the cigarette reviews with preset keyword text to determine if keyword features exist; if so, extracting the contextual structure from the keyword features and evaluating and identifying the contextual structure using a pre-established recurrent neural network; the evaluation and identification results include positive reviews with decreasing probabilities of positive reviews, positive reviews with adjectives, positive reviews with contrasting probabilities, negative reviews, and negative reviews with adjectives; establishing a mapping relationship between the evaluation and identification results and the scoring results, and obtaining the scoring results based on the cigarette reviews based on the mapping relationship.
[0013] In one embodiment, cigarette brands correspond to various cigarette ingredients; the method further includes:
[0014] Based on the cigarette brands preferred by each user group, the corresponding cigarette ingredients were determined and used as target cigarette ingredients respectively.
[0015] Based on the target cigarette components, the dispersion analysis of the cigarette components of the cigarette brands on sale was carried out, and preset thresholds were set for each cigarette component.
[0016] If the dispersion of each component of any cigarette brand on sale exceeds the corresponding preset threshold compared to the dispersion of the target cigarette component, then that brand is marked as an unpopular cigarette brand.
[0017] In one embodiment, the cigarette components include tar and nicotine; the method further includes:
[0018] The dispersion of cigarette components from all currently sold cigarette brands relative to the target cigarette components was analyzed with the goal of minimizing tar and nicotine content.
[0019] Based on the traversal results, the cigarette brands with the lowest tar and nicotine content are selected as the most suitable brands for sale, in order to replace the brands with the corresponding target cigarette components.
[0020] In one embodiment, the method further includes:
[0021] Obtain the geographical distribution of sales terminals for each cigarette brand currently on sale, and randomly select user profiles from multiple pre-collected geographical distribution areas as user profiles to be identified.
[0022] Multiple feature vectors are obtained based on the user profiles of each user group, and a first preset number of feature vectors are extracted as support vectors.
[0023] The distance between each support vector and the hyperplane is calculated based on multiple feature vectors, the user group corresponding to each feature vector, and the preset hyperplane parameters.
[0024] If the minimum distance between each support vector and the hyperplane is less than a preset value, then update the hyperplane parameters; if the minimum distance between each support vector and the hyperplane is not less than a preset value, then set the support vector machine to the trained support vector machine.
[0025] The trained support vector machine is used to classify the user profile to be identified.
[0026] In one embodiment, the method further includes:
[0027] If any user profile to be identified cannot find the decision boundary in the support vector machine, then the user profile to be identified is marked.
[0028] Obtain feedback data corresponding to the labeled user profile to be identified, parse the corresponding feedback data to obtain the scoring results, determine the correspondence between the user profile to be identified and the obtained scoring results, and classify the user profile to be identified into user groups based on the correspondence.
[0029] Secondly, this application provides a feedback analysis device for cigarette users, the device comprising:
[0030] The data collection module is used to collect feedback data from multiple cigarette users on various cigarette brands, as well as a user profile for each cigarette user; the user profile includes smoking experience, smoking frequency, and income level.
[0031] The parsing module is used to parse the feedback data according to preset rules to obtain the rating results of the corresponding cigarette brands, and to mine the correspondence between user profiles and rating results based on association rule algorithms;
[0032] The segmentation module is used to divide each cigarette user into multiple user groups based on the corresponding relationship, so as to determine the cigarette preferences of each user group.
[0033] Thirdly, this application provides a computer device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method provided in the first aspect of this application.
[0034] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method provided in the first aspect of this application.
[0035] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method provided in the first aspect of this application.
[0036] The aforementioned method and apparatus for analyzing cigarette user feedback can collect feedback data from multiple cigarette users regarding various cigarette brands, as well as user profiles for each user. This allows for the accurate identification of cigarette brand preferences among users with different smoking histories, smoking frequencies, and income levels. Furthermore, it quantifies and subdivides the degree of preference using preset rules. Since users may share similarities or differences in one dimension of their user profiles while being similar or identical to other users in other dimensions, the user groups are re-divided based on the correspondence between the subdivided preference levels and smoking histories, smoking frequencies, and income levels. This accurately identifies the true preferences of different user profile groups for cigarette brands, effectively addressing the problem of inaccurate cigarette market predictions caused by ignoring consumer feedback. Attached Figure Description
[0037] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0038] Figure 1 This is a flowchart illustrating the steps involved in segmenting cigarette user groups in one embodiment;
[0039] Figure 2 This is a flowchart illustrating the steps involved in obtaining a scoring result based on feedback data in one embodiment.
[0040] Figure 3 This is a flowchart illustrating the steps involved in performing dispersion analysis on cigarette brands on the market in one embodiment.
[0041] Figure 4 A flowchart illustrating the steps of replacing the brand of the target cigarette ingredient in one embodiment;
[0042] Figure 5 This is a flowchart illustrating the steps involved in classifying a user profile to be identified in one embodiment.
[0043] Figure 6 This is a flowchart illustrating the steps involved in segmenting user groups based on a user profile to be identified, as shown in one embodiment.
[0044] Figure 7 This is a structural diagram of a feedback analysis device for cigarette users in one embodiment. Detailed Implementation
[0045] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0046] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.
[0047] In one exemplary embodiment, a feedback analysis method for cigarette users is provided, such as... Figure 1 As shown, the method includes the following steps S102 to S106. Wherein:
[0048] S102 collects feedback data from multiple cigarette users on various cigarette brands, as well as user profiles for each user; user profiles include smoking history, smoking frequency, and income level.
[0049] Specifically, since different cigarette brands have different tastes, and depending on the smoking experience and frequency, the demand for cigarettes with strong and mild flavors varies, and each cigarette brand corresponds to a different price, when the price exceeds the user's economic income and expectations, users will give different feedback based on the cost-effectiveness of the cigarette. Therefore, by analyzing the user profiles of cigarette users, we can determine the true feedback of users with different cigarette needs on any particular brand of cigarette.
[0050] Specifically, the users collected can be representative volunteer users who meet the criteria in the user profile.
[0051] S104: The feedback data is parsed according to preset rules to obtain the rating results of the corresponding cigarette brands, and the correspondence between user profiles and rating results is mined based on the association rule algorithm.
[0052] Specifically, preset rules are used to normalize different feedback types and convert them into rating results that can quantify user experience.
[0053] On the other hand, since user profiles include multiple dimensions of features such as smoking age, smoking frequency, and income capacity, and each dimension of features is actually related to the user's rating results for multiple different brands of cigarettes, even if a user gives a perfect score to some brands of cigarettes, they may also give other brands of cigarettes a score that is just below perfect. This is actually the influence of each dimension of the user profile as the user's focus on the rating results. Therefore, association rule algorithms can be used to analyze which brands users with different user profiles prefer.
[0054] S106. Based on the correspondence, each cigarette user is divided into multiple user groups to determine the cigarette preferences of each user group.
[0055] Furthermore, by analyzing which brands users with different user profiles prefer, the user groups can be segmented to determine which user groups favor each brand of cigarettes.
[0056] This application aims to provide a feedback analysis method for cigarette users, which can collect feedback data from multiple cigarette users on various cigarette brands, as well as user profiles of each cigarette user, to accurately identify the preferences of users with different smoking ages, smoking frequencies, and income levels for cigarette brands. The method quantifies and subdivides the degree of preference through preset rules. Since each user may be the same or similar in one dimension of the user profile, but the same or similar in other dimensions, the user groups are re-divided based on the correspondence between the subdivided preference degree and smoking age, smoking frequency, and income level, so as to accurately identify the true preferences of user groups with different user profiles for cigarette brands.
[0057] In one exemplary embodiment, such as Figure 2 As shown, a feedback analysis method for cigarette users further includes the following steps S1042 to S1046. Wherein:
[0058] S1042, Identify the type of feedback data; the types of feedback data include cigarette scores and cigarette evaluations.
[0059] Specifically, cigarette ratings can be direct scores given by users to cigarettes, while cigarette reviews can be verbal comments.
[0060] S1044, when the feedback data type is cigarette rating, the cigarette rating is normalized to correspond to the rating result, and the normalized cigarette rating is converted into a rating result.
[0061] Specifically, during the scoring process, since sales terminals are scattered and questionnaires or scoring sheets may not have a uniform format or scoring standards, cigarette scores can be normalized to correspond with the scoring results.
[0062] S1046, when the feedback data type is cigarette reviews, the evaluation analysis strategy includes: comparing the cigarette reviews with preset keyword text to determine whether keyword features exist; if so, extracting the contextual structure from the keyword features and evaluating and identifying the contextual structure through a pre-established recurrent neural network; the evaluation and identification results include positive reviews with adjectives, positive reviews, positive reviews with transitions, negative reviews, and negative reviews with adjectives, with the probability of positive reviews decreasing sequentially; establishing a mapping relationship between the evaluation and identification results and the scoring results, and obtaining the scoring results based on the cigarette reviews based on the mapping relationship.
[0063] Specifically, by pre-setting keyword feature templates, the positive review rate is logically sorted and judged according to positive reviews with adjectives, positive reviews with transitions, negative reviews, and negative reviews with adjectives, and then further mapped to the scoring results.
[0064] In one exemplary embodiment, such as Figure 3 As shown, the method also includes the following steps S202 to S206.
[0065] in:
[0066] S202, based on the cigarette brands preferred by each user group, determine the corresponding cigarette ingredients and use them as target cigarette ingredients respectively.
[0067] It is understandable that cigarette brands and cigarette ingredients correspond to each other. When a user has a preference for a certain cigarette brand, it is actually a preference for the cigarette ingredients. Therefore, the cigarette ingredients are used to further quantify user preferences.
[0068] S204, perform dispersion analysis on the cigarette components of the cigarette brands on sale according to the target cigarette components, and set preset thresholds corresponding to each cigarette component.
[0069] Specifically, the dispersion of each cigarette component is analyzed to quantify the difference between the cigarette's taste and the target cigarette's taste.
[0070] S206, in response to the fact that the dispersion of each cigarette component of any cigarette brand on sale exceeds the corresponding preset threshold compared to the dispersion of the target cigarette component, it is marked as an unpopular cigarette brand.
[0071] Specifically, the preset threshold can be used to measure the significant differences in cigarette taste that users can perceive. User opinions can be collected during the implementation process in the form of decreasing component proportions, and the preset threshold is given after referencing a large amount of data.
[0072] For example, ratings from multiple users can be extracted according to cigarette brands, and preset ratings can be mapped to preset thresholds related to cigarette components.
[0073] In one exemplary embodiment, the cigarette components include tar and nicotine content; such as Figure 4 As shown, the method further includes the following steps S302 to S304. Wherein:
[0074] S302, with the goal of minimizing tar and nicotine content, iterates through the dispersion of cigarette components from various brands of cigarettes on the market relative to the target cigarette components.
[0075] Specifically, tar and nicotine are the two most harmful chemical components to the body and have always been subject to mandatory control. Therefore, if the dispersion does not exceed the preset threshold, it indicates that the user can accept the taste of cigarettes. In this case, the lowest possible levels of both tar and nicotine are selected to ensure the user's health.
[0076] S304: Based on the traversal results, the cigarette brands with the lowest tar and nicotine content are selected as the most suitable brands for sale, in order to replace the brands corresponding to the target cigarette components.
[0077] Furthermore, when both tar and nicotine content are at their lowest levels, it is appropriate to promote these brands to users as alternatives to high-harm cigarettes.
[0078] In one exemplary embodiment, such as Figure 5 As shown, the method further includes the following steps S402 to S410. Wherein:
[0079] S402, obtain the geographical distribution of sales terminals for each cigarette brand on sale, and randomly select multiple pre-collected user profiles from the geographical distribution as user profiles to be identified.
[0080] Specifically, during the specific marketing campaigns of cigarette brands, the taste preferences of people in each region vary. Therefore, user profiles of local populations are randomly selected to obtain the average smoking age, smoking frequency, and income of users.
[0081] S404: Based on the user profiles of each user group, obtain multiple feature vectors, and extract a first preset number of feature vectors as support vectors.
[0082] Specifically, before classification is performed using a support vector machine, the support vector machine is trained by selecting a subset of support vectors from the feature vectors.
[0083] Furthermore, the feature vectors correspond to user groups.
[0084] S406 calculates the distance between each support vector and the hyperplane based on multiple feature vectors, the user group corresponding to each feature vector, and preset hyperplane parameters.
[0085] S408, if the minimum distance between each support vector and the hyperplane is less than the preset value, then update the hyperplane parameters; if the minimum distance between each support vector and the hyperplane is not less than the preset value, then set the support vector machine to the trained support vector machine.
[0086] S410 classifies the user profile to be identified based on the trained support vector machine.
[0087] Specifically, by constructing the optimal hyperplane, Support Vector Machines can effectively process high-dimensional data and have good generalization ability, thus performing well in user profile classification tasks.
[0088] In one exemplary embodiment, such as Figure 6 As shown, the method also includes the following steps S502 to S504.
[0089] in:
[0090] S502, if any user profile to be identified cannot find the decision boundary in the support vector machine, then mark the user profile to be identified.
[0091] Specifically, if the decision boundary cannot be found, it means that the user profiles extracted from the geographical distribution area exceed the user profiles of cigarette users collected during the user group segmentation process, and therefore it is necessary to re-segment the user groups based on the user profiles.
[0092] S504, obtain feedback data corresponding to the marked user profile to be identified, parse the corresponding feedback data to obtain the scoring result, determine the correspondence between the user profile to be identified and the obtained scoring result, and classify the user profile to be identified into user groups based on the correspondence.
[0093] Specifically, feedback data generated by user profiles in the distributed regions is collected, and the feedback data generated by user profiles in the distributed regions is re-analyzed according to the methods used in steps S102 to S106 to obtain scoring results. Since the scoring results are consistent with the scoring results of the aforementioned methods, the correspondence between the feedback data generated by user profiles in the distributed regions and the user profiles in the distributed regions is obtained again, and they are divided into corresponding user groups to finally determine the user groups of each user profile in the distributed regions. Based on the aforementioned analysis of user groups, the user groups of each user profile in the distributed regions are analyzed.
[0094] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.
[0095] Based on the same inventive concept, this application also provides a cigarette user feedback analysis device for implementing the aforementioned cigarette user feedback analysis method. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations of one or more embodiments of the cigarette user feedback analysis device provided below can be found in the limitations of the cigarette user feedback analysis method described above, and will not be repeated here.
[0096] Secondly, such as Figure 7 As shown, this application provides a feedback analysis device 700 for cigarette users, the device comprising:
[0097] The data collection module 701 is used to collect feedback data from multiple cigarette users on various cigarette brands, as well as a user profile for each cigarette user; the user profile includes smoking experience, smoking frequency, and income level.
[0098] The parsing module 702 is used to parse the feedback data according to preset rules to obtain the rating results of the corresponding cigarette brand, and to mine the correspondence between user profiles and rating results based on the association rule algorithm;
[0099] The segmentation module 703 is used to divide each cigarette user into multiple user groups according to the corresponding relationship, so as to determine the cigarette preferences of each user group.
[0100] The various modules in the aforementioned feedback analysis device for cigarette users can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.
[0101] In one exemplary embodiment, the parsing module further includes:
[0102] The identification unit is used to identify the type of feedback data; the types of feedback data include cigarette scores and cigarette evaluations.
[0103] The normalization processing unit is used to perform normalization processing on the cigarette score corresponding to the score result when the feedback data is of the type of cigarette score, and to convert the normalized cigarette score into a score result.
[0104] The evaluation and analysis strategy unit, when the feedback data is cigarette reviews, executes the following evaluation and analysis strategy: comparing the cigarette reviews with preset keyword text to determine if keyword features exist; if so, extracting the contextual structure from the keyword features and evaluating and identifying the contextual structure through a pre-established recurrent neural network; the evaluation and identification results include positive reviews with decreasing probabilities of positive reviews, positive reviews, positive reviews with transitions, negative reviews, and negative reviews with adjectives; establishing a mapping relationship between the evaluation and identification results and the scoring results, and obtaining the scoring results based on the cigarette reviews based on the mapping relationship.
[0105] In one exemplary embodiment, a cigarette user feedback analysis device further includes:
[0106] The component determination module is used to determine the corresponding cigarette components based on the cigarette brands preferred by each user group, and to use them as target cigarette components respectively;
[0107] The analysis module is used to perform dispersion analysis on the cigarette components of the cigarette brands on sale according to the target cigarette components, and to set preset thresholds corresponding to each cigarette component.
[0108] The marking module is used to mark any cigarette brand as an unpopular cigarette brand if the dispersion of each cigarette component of any brand on sale exceeds the corresponding preset threshold compared to the dispersion of the target cigarette component.
[0109] In one exemplary embodiment, the cigarette components include tar and nicotine content, and a cigarette user feedback analysis device further includes:
[0110] The traversal module is used to traverse the dispersion of cigarette components of each brand of cigarettes on sale relative to the target cigarette components, with the goal of minimizing tar and nicotine content.
[0111] The promotion module is used to select the cigarette brands with the lowest tar and nicotine content from those currently on sale based on the traversal results, and then promote them to the user groups corresponding to the target cigarette components.
[0112] In one exemplary embodiment, a cigarette user feedback analysis device further includes:
[0113] The acquisition module is used to acquire the geographical distribution of sales terminals of various cigarette brands on sale, and randomly select multiple user profiles from the pre-collected geographical distribution areas as user profiles to be identified.
[0114] The extraction module is used to obtain multiple feature vectors based on each user profile in the user group, and extract a first preset number of feature vectors as support vectors.
[0115] The calculation module is used to calculate the distance between each support vector and the hyperplane based on multiple feature vectors, the user group corresponding to each feature vector, and preset hyperplane parameters;
[0116] The update module is used to update the hyperplane parameters if the minimum distance between each support vector and the hyperplane is less than a preset value; if the minimum distance between each support vector and the hyperplane is not less than the preset value, the support vector machine is set to the trained support vector machine.
[0117] The classification module is used to classify the user profile to be identified based on the trained support vector machine.
[0118] In an exemplary embodiment, the labeling module is further configured to label the user profile to be identified if any user profile to be identified cannot find a decision boundary in the support vector machine.
[0119] The acquisition module is also used to acquire feedback data corresponding to the labeled user profile to be identified, parse the corresponding feedback data to obtain the scoring results, determine the correspondence between the user profile to be identified and the obtained scoring results, and classify the user profile to be identified into user groups based on the correspondence.
[0120] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the remote update method for the aforementioned application.
[0121] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the aforementioned automatic deployment and retrieval method for buoys.
[0122] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the aforementioned automatic deployment and retrieval method for buoys.
[0123] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.
[0124] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0125] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.
Claims
1. A method for analyzing feedback from cigarette users, characterized in that, The method includes: Feedback data from multiple cigarette users on various cigarette brands was collected, along with a user profile for each user; the user profile included smoking history, smoking frequency, and income level. The feedback data is parsed according to preset rules to obtain the rating results corresponding to the cigarette brand, and the correspondence between the user profile and the rating results is mined based on the association rule algorithm; wherein: the step of parsing the feedback data according to preset rules to obtain the rating results corresponding to the cigarette brand includes: identifying the type of feedback data; the type of feedback data includes cigarette rating and cigarette review; when the type of feedback data is cigarette rating, the cigarette rating is normalized to correspond to the rating result, and the normalized cigarette rating is converted into the rating result; when the type of feedback data is cigarette review, the evaluation analysis strategy includes: comparing the cigarette review with preset keyword text features to determine whether keyword features exist; if so, extracting the context structure in the keyword features, and evaluating and identifying the context structure through a pre-established recurrent neural network; the evaluation and identification results include positive reviews with adjectives, positive reviews, positive reviews with transitions, negative reviews, and negative reviews with adjectives, with the probability of positive reviews decreasing sequentially; establishing a mapping relationship between the evaluation and identification results and the rating results, and obtaining the rating result based on the cigarette review based on the mapping relationship; Based on the correspondence, each cigarette user is divided into multiple user groups to determine the cigarette preferences of each user group; Based on the cigarette brands preferred by each user group, the corresponding cigarette ingredients are determined and used as target cigarette ingredients respectively. According to the target cigarette components, the dispersion analysis of the cigarette components of the cigarette brands on sale is performed respectively, and preset thresholds are set for each cigarette component. If the dispersion of each cigarette component of any cigarette brand on sale exceeds the corresponding preset threshold compared to the dispersion of the target cigarette component, then the brand is marked as an unpopular cigarette brand. The cigarette components include tar content and nicotine content; the method further includes: traversing the dispersion of cigarette components of each brand of cigarettes on sale relative to the target cigarette component with the goal of minimizing the tar content and nicotine content; Based on the traversal results, the cigarette brand with the lowest tar content and the lowest nicotine content is selected as the most suitable brand for sale, in order to replace the brand corresponding to the target cigarette component. The method further includes: obtaining the geographical distribution of sales terminals of each cigarette brand on sale, and randomly selecting multiple user profiles from the pre-collected geographical distribution as user profiles to be identified. Based on the user profiles of each user in the user group, multiple feature vectors are obtained, and a first preset number of feature vectors are extracted as support vectors. The distance between each support vector and the hyperplane is calculated based on multiple feature vectors, the user group corresponding to each feature vector, and preset hyperplane parameters. If the minimum distance between each support vector and the hyperplane is less than a preset value, then the hyperplane parameters are updated; if the minimum distance between each support vector and the hyperplane is not less than the preset value, then the support vector machine is set as a trained support vector machine. The user profile to be identified is classified based on the trained support vector machine.
2. The method according to claim 1, characterized in that, The method further includes: If any user profile to be identified cannot find a decision boundary in the support vector machine, then the user profile to be identified is marked. The system acquires feedback data corresponding to the labeled user profile to be identified, parses the corresponding feedback data to obtain a rating result, determines the correspondence between the user profile to be identified and the obtained rating result, and classifies the user profile to be identified into the user group based on the correspondence.
3. A feedback analysis device for cigarette users, characterized in that, The apparatus is implemented based on the method as described in any one of claims 1-2; the apparatus includes: The data collection module is used to collect feedback data from multiple cigarette users on various cigarette brands, as well as a user profile for each cigarette user; the user profile includes smoking experience, smoking frequency, and income level. The parsing module is used to parse the feedback data according to preset rules to obtain the rating result corresponding to the cigarette brand, and to mine the correspondence between the user profile and the rating result based on the association rule algorithm; The segmentation module is used to divide each of the cigarette users into multiple user groups according to the correspondence, so as to determine the cigarette preferences of each user group.
4. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 2.
5. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 2.
6. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 2.