Method for determining tobacco product brand packaging and promotional strategies based on data mining

Abstract

The application discloses a tobacco product brand packaging and propaganda strategy determination method based on data mining, and comprises the following steps: analyzing a first influence factor of tobacco product brand packaging according to industry association announcements and news; predicting a second influence factor of tobacco product brand packaging and propaganda strategy caused by tax and industry requirement changes; analyzing consumer acceptance degree information of tobacco product brand packaging and health warnings; mining consumer preference degree information of different brand packaging design elements; determining diversified direction information of tobacco product brand packaging; analyzing market brand packaging trends and competition situations; determining a relationship curve between sales performance and different brand packaging and different propaganda strategies; evaluating a third influence factor of tobacco product brand packaging and propaganda strategy on market sales; and determining target brand packaging and target propaganda strategy of the tobacco product. The application improves the accuracy and reliability of tobacco product brand packaging and propaganda strategy determination, and can be applied to the field of information technology.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of information technology, and particularly to a method for determining tobacco product brand packaging and promotion strategies based on data mining. BACKGROUND

[0002] With the development of the tobacco industry and the increasing competition, cross-unit joint development and product brand packaging and promotion strategies are crucial to the market competitiveness of enterprises. However, there are some problems in the current tobacco industry in terms of "industry and commerce cooperation" and product brand packaging and promotion, which need to be analyzed and improved. First of all, the current product brand packaging and promotion strategies lack in-depth understanding of consumer needs and preferences. Tobacco companies need to strengthen "industry and commerce cooperation" to promote development together, and at the same time, understand consumer preferences for different product brand packaging design elements and their acceptance of product brand packaging and health warnings. Only by deeply understanding the needs and preferences of consumers, can enterprises develop product brand packaging and promotion strategies that meet market demand. Secondly, the current product brand packaging and promotion strategies lack diversity and individualization. Tobacco companies usually only provide a few flavors, qualities and types of product brand packaging, which cannot meet the diverse needs of consumers. At the same time, the lack of individualized product brand packaging design also makes the product lack of differentiated competitiveness in the market. Therefore, tobacco companies need to optimize product brand packaging and promotion strategies according to consumer needs and market trends, and achieve diversity and individualization. In addition, the current product brand packaging and promotion strategies lack in-depth analysis and understanding of competitors. Tobacco companies need to monitor competitors' product brand packaging design and promotion strategies, as well as analyze popular product brand packaging trends and competitive situations in the market. Only by deeply understanding competitors' strategies, can enterprises develop more competitive product brand packaging and promotion strategies. Finally, the current product brand packaging and promotion strategies lack evaluation of the specific impact on market sales, and ignore the method of designing tobacco display shelves to arrange and combine tobacco products to improve tobacco sales, which leads to the fact that tobacco companies cannot provide specific service guidance for cigarette retailers, and form a friendly relationship with customers. SUMMARY

[0003] The present application aims to at least partially solve one of the problems in the prior art.

[0004] To this end, it is an object of embodiments of the present application to provide a method for determining tobacco product brand packaging and promotion strategies based on data mining, which improves the accuracy and reliability of the determination of tobacco product brand packaging and promotion strategies, thereby improving the market competitiveness of tobacco companies and tobacco products.

[0005] Another object of the embodiments of the present application is to provide a tobacco product brand packaging and promotion strategy determination system based on data mining.

[0006] To achieve the above technical purposes, the technical solutions adopted by the embodiments of the present application include:

[0007] In a first aspect, the embodiments of the present application provide a tobacco product brand packaging and promotion strategy determination method based on data mining, including the following steps:

[0008] According to the announcements and news of industry associations, negative evaluation information of the cigarette industry is crawled, key information related to brand packaging requirements is extracted according to the negative evaluation information, and a first influence factor of the tobacco product brand packaging is analyzed according to the key information;

[0009] Tobacco tax and tobacco control regulations are analyzed, and a second influence factor of the tobacco product brand packaging and promotion strategy is predicted by the changes of tax and industry requirements;

[0010] Using public opinion analysis tools, according to the discussions on social media platforms and professional health websites, the acceptance degree information of consumers to the tobacco product brand packaging and health warning is analyzed;

[0011] According to the online feedback and comments of consumers on the tobacco product brand packaging, using sentiment analysis technology, the preference degree information of consumers to different brand packaging design elements is mined;

[0012] According to market sales data, the demand of consumers for different quality and type of brand packaging is analyzed to determine the diversification direction information of the tobacco product brand packaging;

[0013] The brand packaging design data of competitors is crawled, the convolutional neural network is used to identify brand packaging images, and the market brand packaging trend and competitive situation are analyzed;

[0014] According to the advertising activities, the tobacco product brand packaging design and the sales data of the tobacco product, the relationship curve between the sales performance and different brand packaging and different promotion strategies is determined;

[0015] According to the promotion effect and market feedback of the tobacco product brand packaging, using multiple regression analysis and prediction model, the third influence factor of the tobacco product brand packaging and promotion strategy on market sales is evaluated;

[0016] According to the acceptance degree information, the preference degree information, the diversification direction information, the market brand packaging trend and competitive situation, the relationship curve, the first influence factor, the second influence factor and the third influence factor, the target brand packaging and the target promotion strategy of the tobacco product are determined.

[0017] Further, in one embodiment of the present application, the negative evaluation information of the tobacco industry is crawled according to the industry association announcements and news, the key information related to the brand packaging requirements is extracted according to the negative evaluation information, and the first influence factor on the brand packaging of tobacco products is analyzed according to the key information, which specifically includes:

[0018] The first text data of the industry association announcements and news reports is crawled, the first text data is denoised, segmented and stop words are removed to obtain a word sequence;

[0019] The first entity about tobacco products and tobacco enterprises in the word sequence is recognized by using a named entity recognition algorithm;

[0020] According to the first entity, the key information related to the industry regulation changes and brand packaging requirements is extracted from the first text data by using keyword extraction or a topic model;

[0021] The importance and frequency of occurrence of each word are determined by performing word frequency analysis on the key information;

[0022] According to the sorting and position of the words in the text, the relevance and context of each word are determined;

[0023] The proportion of positive sentiment, negative sentiment and neutral sentiment is obtained by performing sentiment analysis on the key information, and the sentiment attitude and sentiment change are determined;

[0024] The theme distribution of the key information is determined, the theme words and theme proportions are determined, the association graph of the theme is constructed, and the association relationship between the themes is analyzed;

[0025] The specific content of the industry association announcement changes and the first influence factor on the brand packaging of tobacco products are determined.

[0026] Further, in one embodiment of the present application, the second influence factor of the tax and industry requirements changes on the brand packaging and promotion strategy of tobacco products is predicted by analyzing the tobacco tax and tobacco control regulations, which specifically includes:

[0027] Attribute data of product brand packaging specifications, brand packaging design features, printing information, brand packaging materials, promotion channels and promotion content are crawled;

[0028] According to the second text data of the tobacco tax and tobacco control regulations, the influence of the brand packaging and promotion strategy of tobacco products is taken as a result label;

[0029] A random forest algorithm is used to construct a prediction model of the influence of changes in tax regulations on brand packaging promotion, and to determine the changes in packaging specifications, design features, materials, channels, and content of brand packaging of tobacco products under different tobacco tax and smoking control regulations;

[0030] According to the prediction results, the second influence factor of tobacco tax and smoking control regulations on tobacco product brand packaging and promotion strategies is determined.

[0031] Further, in an embodiment of the present application, the public opinion analysis tool is used to analyze the consumer acceptance of tobacco product brand packaging and health warnings based on discussions on social media platforms and professional health websites, which specifically includes:

[0032] Crawl comment text on social media platforms and professional health websites, and collect consumer evaluation text on the appearance design, information transmission effect, and health warning acceptance of tobacco product brand packaging;

[0033] TF-IDF algorithm is used to extract keywords related to the attractiveness of tobacco product brand packaging, the fit of brand image, and the consistency of design style and personal aesthetics;

[0034] TextRank algorithm is used to determine the acceptance of different brand packaging and health warnings by consumers;

[0035] Sentiment analysis algorithm is used to analyze the degree of influence of health warnings on smoking behavior and the degree of concern for personal health attributes considered by consumers;

[0036] The frequency of keywords in consumer comments on the environmental friendliness of tobacco product brand packaging is counted, and the degree of recognition of the environmental friendliness of brand packaging materials and the acceptance of environmentally friendly brand packaging by consumers are analyzed.

[0037] Further, in an embodiment of the present application, the consumer preference for different brand packaging design elements is mined based on online feedback and comments on tobacco product brand packaging using sentiment analysis technology, which specifically includes:

[0038] According to the online feedback and comments of consumers, data containing evaluations and emotional expressions of brand packaging colors are obtained;

[0039] Sentiment analysis algorithm is used to analyze the degree of preference and emotional response of consumers to different colors, and to determine the preference information of consumers for different color packaging design;

[0040] According to the online feedback and comments of consumers, data containing evaluations and emotional expressions of brand packaging images and patterns are obtained;

[0041] adopting the sentiment analysis algorithm to analyze the preference degree and emotional response of the consumers to different images and patterns, and determining the preference degree information of the consumers to the packaging design of different images and patterns;

[0042] extracting data containing evaluation and emotional expression on the font and layout of the brand packaging according to the online feedback and comments of the consumers;

[0043] adopting the sentiment analysis algorithm to analyze the preference degree and emotional response of the consumers to different fonts and layouts, and determining the preference degree information of the consumers to the packaging design of different fonts and layouts;

[0044] extracting data containing evaluation and emotional expression on the material and texture of the brand packaging according to the online feedback and comments of the consumers;

[0045] adopting the sentiment analysis algorithm to analyze the preference degree and emotional response of the consumers to different materials and textures, and determining the preference degree information of the consumers to the packaging design of different materials and textures.

[0046] Further, in an embodiment of the present application, the market sales data is used to analyze the demand of the consumers for different quality and type of brand packaging, and determine the diversification direction information of the brand packaging of the tobacco product, which specifically includes:

[0047] obtaining the purchase history data, preference data and evaluation data of the consumers for purchasing the tobacco product;

[0048] obtaining the demand of the consumers for different quality and type of brand packaging according to the purchase history data and preference data of the consumers;

[0049] obtaining the demand of the consumers for the brand packaging of the tobacco product characteristics and brand image according to the tobacco product purchased by the consumers and the corresponding brand packaging characteristics;

[0050] adopting the collaborative filtering algorithm to analyze the preference and correlation of the consumers for different brand packaging attributes of the tobacco product according to the purchase history data and evaluation data of different consumers, and determining the preference degree and requirement of the consumers for the safety, sustainability, visual appeal and convenience brand packaging attributes;

[0051] adopting the collaborative filtering algorithm to analyze the brand packaging attributes of the tobacco product that are most valued by the consumers according to the preference and correlation of the consumers for the brand packaging attributes of the tobacco product, and determining the diversification direction information of the brand packaging of the tobacco product.

[0052] Further, in an embodiment of the present application, the brand packaging design data of the competitors is crawled, and the convolutional neural network is used to identify the brand packaging images, and analyze the market brand packaging trend and competitive situation, which specifically includes:

[0053] Obtain the brand packaging design image dataset of the competitor and label it to obtain label data containing brand packaging color, shape, pattern and text;

[0054] According to the obtained brand packaging design image dataset and label data, a convolutional neural network algorithm is used to construct a brand packaging image recognition model to determine the brand packaging color, shape, pattern and text attributes of the new brand packaging image;

[0055] The brand packaging design images of the competitors are divided into several clusters, each cluster representing a brand packaging trend or competitive situation, and a K-means clustering algorithm is used to construct a brand packaging feature clustering analysis model;

[0056] Based on the feature attributes of the clusters, the brand packaging design trends in the market and the brand packaging strategies of the competitors are determined.

[0057] Further, in an embodiment of the present application, the determination of the relationship curve between sales performance and different brand packaging and different promotion strategies according to advertising activities, tobacco product brand packaging design and sales data of tobacco products specifically includes:

[0058] Obtain sales data and brand packaging type information, including brand packaging type, sales performance and brand;

[0059] Through data preprocessing, the sales data and brand packaging type are encoded and associated rule mining is performed;

[0060] Frequent item sets of different brand packaging types and sales performance are constructed, and the frequent item sets represent the association relationship between different brand packaging and sales performance;

[0061] Construct association rules, and extract association rules from the frequent item sets according to the threshold of confidence;

[0062] According to the support and confidence, the importance of the association rules is evaluated, and the association rules with practical significance are screened out;

[0063] According to the obtained association rules, the relationship curve between sales performance and different brand packaging and different promotion strategies is determined.

[0064] Further, in an embodiment of the present application, the third influence factor of the tobacco product brand packaging and promotion strategy on market sales is evaluated according to the promotion effect of the tobacco product brand packaging and market feedback, using a multiple regression analysis and prediction model, which specifically includes:

[0065] Obtain tobacco product brand packaging design and color-related data, including sales data, market research, and consumer feedback comment text for different design and color schemes;

[0066] Determine the first impact coefficient of brand packaging design and color on market sales volume using multiple regression analysis method;

[0067] Obtain relevant data on different brand packaging materials and quality, including consumer evaluations of brand packaging materials and quality and product protection performance;

[0068] Determine the second impact coefficient of brand packaging materials and quality on market sales volume using multiple regression analysis method;

[0069] Obtain relevant data on different promotion channels and content, including exposure of different channels and content and consumer response to promotion content;

[0070] Determine the third impact coefficient of promotion channels and content on market sales volume using multiple regression analysis method;

[0071] Obtain target audience reaction data on brand packaging and promotion strategy, including consumer purchase intention and brand awareness;

[0072] Determine the fourth impact coefficient of target audience reaction on market sales volume using multiple regression analysis method;

[0073] Determine the third impact factor of tobacco product brand packaging and promotion strategy on market sales based on the first impact coefficient, the second impact coefficient, the third impact coefficient, and the fourth impact coefficient.

[0074] In a second aspect, an embodiment of the present application provides a tobacco product brand packaging and promotion strategy determination system based on data mining, comprising:

[0075] A first impact factor determination module is configured to crawl negative evaluation information of the cigarette industry according to industry association announcements and news, extract key information related to brand packaging requirements based on the negative evaluation information, and analyze the first impact factor of tobacco product brand packaging based on the key information;

[0076] A second impact factor determination module is configured to analyze tobacco tax and smoking control regulations, and predict the second impact factor of tax and industry requirements changes on tobacco product brand packaging and promotion strategy;

[0077] An acceptance information determination module is configured to use public opinion analysis tools to analyze consumer acceptance information of tobacco product brand packaging and health warnings based on discussions on social media platforms and professional health websites;

[0078] The preference degree information determination module is configured to determine the preference degree information of the consumers for different brand packaging design elements of the tobacco product by using sentiment analysis technology based on online feedback and comments of the consumers on the brand packaging of the tobacco product.

[0079] The diversification direction information determination module is configured to determine the diversification direction information of the brand packaging of the tobacco product by analyzing the demand of the consumers for different quality and type of brand packaging based on market sales data.

[0080] The market brand packaging trend and competitive situation determination module is configured to crawl brand packaging design data of competitors, recognize brand packaging images by using a convolutional neural network, and analyze market brand packaging trends and competitive situations.

[0081] The relationship curve determination module is configured to determine the relationship curve between sales performance and different brand packaging and different promotion strategies based on advertising activities, brand packaging design of the tobacco product, and sales data of the tobacco product.

[0082] The third influence factor determination module is configured to evaluate the third influence factor of the brand packaging and the promotion strategy of the tobacco product on market sales by using multiple regression analysis and prediction models based on the promotion effect and market feedback of the brand packaging of the tobacco product.

[0083] The brand packaging and promotion strategy determination module is configured to determine the target brand packaging and the target promotion strategy of the tobacco product based on the acceptance degree information, the preference degree information, the diversification direction information, the market brand packaging trend and competitive situation, the relationship curve, the first influence factor, the second influence factor, and the third influence factor.

[0084] In a third aspect, an embodiment of the present application provides a data mining-based determination apparatus for brand packaging and promotion strategy of a tobacco product, comprising:

[0085] at least one processor;

[0086] at least one memory configured to store at least one program;

[0087] When the at least one program is executed by the at least one processor, the at least one processor is caused to implement the data mining-based determination method for brand packaging and promotion strategy of a tobacco product.

[0088] In a fourth aspect, an embodiment of the present application further provides a computer readable storage medium, wherein a processor executable program is stored, and the processor executable program is configured to execute the data mining-based determination method for brand packaging and promotion strategy of a tobacco product when executed by a processor.

[0089] The advantages and beneficial effects of the present application will be partially given in the following description, partially will become apparent from the following description, or will be appreciated by the practice of the present application.

[0090] According to the industry association announcement and news, the negative evaluation information of the tobacco industry is crawled, the key information related to the brand packaging requirements is extracted according to the negative evaluation information, the first influence factor of the tobacco product brand packaging is analyzed according to the key information, the tobacco tax and the control of tobacco regulations are analyzed, the second influence factor of the tobacco product brand packaging and the propaganda strategy is predicted according to the change of the tax and the industry requirement, the public opinion analysis tool is used, the acceptance degree information of the consumer to the tobacco product brand packaging and the health warning is analyzed according to the discussion on the social media platform and the professional health website, the sentiment analysis technology is used, the preference degree information of the consumer to different brand packaging design elements is mined according to the online feedback and the comment of the consumer to the tobacco product brand packaging, the market sales data is analyzed, the demand of the consumer to different quality and type of brand packaging is analyzed, the diversification direction information of the tobacco product brand packaging is determined, the brand packaging design data of the competitor is crawled, the convolutional neural network is used to identify the brand packaging image, the market brand packaging trend and the competition situation are analyzed, the relationship curve of the sales performance and different brand packaging and different propaganda strategy is determined according to the advertising propaganda activity, the tobacco product brand packaging design and the sales data of the tobacco product, the third influence factor of the tobacco product brand packaging and the propaganda strategy to the market sales is evaluated according to the propaganda effect and the market feedback of the tobacco product brand packaging, the multiple regression analysis and the prediction model are used, the target brand packaging and the target propaganda strategy of the tobacco product are determined according to the acceptance degree information, the preference degree information, the diversification direction information, the market brand packaging trend and the competition situation, the relationship curve, the first influence factor, the second influence factor and the third influence factor. The accuracy and the reliability of the tobacco product brand packaging and the propaganda strategy are improved, so that the market competitiveness of the tobacco enterprise and the tobacco product is improved. BRIEF DESCRIPTION OF DRAWINGS

[0091] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following introduces the drawings needed to be used in the embodiments of the present application. It should be understood that the drawings introduced in the following are only for facilitating the clear expression of part of the embodiments in the technical solutions of the present application, and other drawings can be obtained by those skilled in the art without paying creative labor on the premise.

[0092] Figure 1 A step flow chart of a tobacco product brand packaging and propaganda strategy determination method based on data mining provided by the embodiments of the present application is provided.

[0093] Figure 2A structural block diagram of a tobacco product brand packaging and promotion strategy determination system based on data mining provided by an embodiment of the present application is provided.

[0094] Figure 3 A structural block diagram of a tobacco product brand packaging and promotion strategy determination device based on data mining provided by an embodiment of the present application is provided. DETAILED DESCRIPTION

[0095] The embodiments of the present application are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be understood as a limitation of the present application. For the step numbers in the following embodiments, they are only set for the convenience of explanation, and the order between the steps is not limited in any way, and the execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0096] In the description of the present application, the meaning of multiple is two or more, and if the first, the second is described, it is only for the purpose of distinguishing technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features or implicitly indicating the sequence of indicated technical features. In addition, unless otherwise defined, all technical and scientific terms used herein have the same meaning as understood by those skilled in the art.

[0097] Reference Figure 1 , the embodiment of the present application provides a tobacco product brand packaging and promotion strategy determination method based on data mining, which specifically comprises the following steps:

[0098] S101, according to the announcement and news of the industry association, the negative evaluation information of the cigarette industry is crawled, the key information related to the brand packaging requirement is extracted according to the negative evaluation information, and the first influence factor of the tobacco product brand packaging is analyzed according to the key information.

[0099] Further as an optional implementation, according to the announcement and news of the industry association, the negative evaluation information of the cigarette industry is crawled, the key information related to the brand packaging requirement is extracted according to the negative evaluation information, and the first influence factor of the tobacco product brand packaging is analyzed according to the key information, which specifically comprises:

[0100] S1011, the first text data of the announcement and news of the industry association is crawled, the first text data is denoised, segmented and stop words are removed to obtain a word sequence;

[0101] S1012, the first entity about tobacco products and tobacco enterprises in the word sequence is recognized by using a named entity recognition algorithm.

[0102] S1013, extracting, from the first text data, key information related to the industry regulation change and brand packaging requirement according to the first entity by using keyword extraction or topic modeling;

[0103] S1014, performing word frequency analysis on the key information to determine the importance and frequency of occurrence of each word;

[0104] S1015, determining the relevance and context of each word according to the ranking of the word and the position of the word in the text;

[0105] S1016, performing sentiment analysis on the key information to obtain the proportion of positive sentiment, negative sentiment and neutral sentiment, and determine the sentiment attitude and sentiment change;

[0106] S1017, performing topic distribution judgment on the key information to determine the topic word and topic proportion, constructing the association graph of the topic, and analyzing the association relationship between the topics;

[0107] S1018, determining the specific content of the industry association announcement change and the first influence factor on the tobacco product brand packaging.

[0108] Specifically, the industry association announcement and news report text data are crawled, and text preprocessing is performed on the crawled text data, including removing noise, word segmentation and removing stop words. The entities related to the tobacco industry are identified by using the named entity recognition algorithm, including tobacco enterprises and cigarette brands. After identifying the related entities, the key information related to the industry regulation change and product brand packaging requirement is extracted from the text by using keyword extraction or topic modeling. The extracted key information is subjected to word frequency analysis, and the word frequency information is obtained by frequency statistics of the words to determine the importance and frequency of occurrence of the words. According to the ranking of the words and the position of the words in the text, the relevance and context of the words are determined. The extracted key information is subjected to sentiment analysis to obtain the proportion of positive sentiment, negative sentiment and neutral sentiment, and determine the sentiment attitude and sentiment change. The extracted key information is subjected to topic distribution judgment to determine the topic word and topic proportion, construct the association graph of the topic, and analyze the association relationship between the topics. The specific content of the association announcement change, the timeliness, the adjustment of the product brand packaging requirement and the influence on the market demand are determined.

[0109] For example, 1000 industry announcements and news report text data are crawled. HTML tags, special characters and non-Chinese characters are removed to obtain clean text data. Jieba word segmentation tool is used to segment each text to obtain a list of words. The stop word list is used to filter the segmentation results to remove common words without actual meaning, such as "of", "is", etc. For example, using named entity recognition algorithm, using LTP and other tools to perform named entity recognition on the segmentation results, and identifying entities related to the tobacco industry, such as tobacco companies and cigarette brands. Use the TF-IDF algorithm to extract keywords from the text to obtain key information related to industry regulation changes and product brand packaging requirements, such as "product brand packaging requirements". Word frequency analysis, such as counting the frequency of keywords in the text to obtain word frequency information. For example, the announcement change appears 200 times, and the product brand packaging requirement appears 150 times. Sentiment analysis, such as using sentiment analysis algorithm to analyze the sentiment of the key information. Get the proportion of positive sentiment, negative sentiment and neutral sentiment, such as positive sentiment accounts for 30%, negative sentiment accounts for 10%, and neutral sentiment accounts for 60%. Topic distribution determination, such as using topic model algorithm to determine the topic distribution of the key information. Determine the theme word and theme proportion, construct the association graph of the theme, and analyze the association relationship between the themes. According to the word frequency information and the topic distribution analysis result, the specific industry regulation change content and product brand packaging requirement adjustment are determined, and the association relationship between the themes is analyzed to determine the timeliness and market demand influence. For example, according to the keywords "prohibit advertising" and the theme "health warning", it is inferred that the industry regulation change content is about the prohibition of cigarette advertising, which will have a certain negative impact on the market demand of tobacco companies in the business aspect, such as sharp decline in sales, and product oversupply.

[0110] S102, analyze tobacco tax and tobacco control regulations to predict the second influence factor of tax and industry requirement changes on tobacco product brand packaging and promotion strategies.

[0111] Further as an optional implementation, the tobacco tax and tobacco control regulations are analyzed to predict the second influence factor of tax and industry requirement changes on tobacco product brand packaging and promotion strategies, which specifically includes:

[0112] S1021, crawling attribute data of product brand packaging specifications, brand packaging design features, printing information, brand packaging materials, promotion channels and promotion content;

[0113] S1022, according to the second text data of tobacco tax and tobacco control regulations, the influence on tobacco product brand packaging and promotion strategy is taken as a result label;

[0114] S1023, using a random forest algorithm, a prediction model of the influence of changes in tax regulations on brand packaging promotion is constructed, and the changes in packaging specifications, design features, materials, promotion channels and promotion content of tobacco product brand packaging under different tobacco tax and tobacco control regulations are determined;

[0115] S1024, according to the prediction results, the second influence factor of tobacco tax and tobacco control regulations on tobacco product brand packaging and promotion strategy is determined.

[0116] Specifically, the attribute data of product brand packaging specifications, product brand packaging design features, printed information, product brand packaging materials, promotion channels and promotion content are crawled. According to the text of tax and tobacco control regulations, the influence of cigarette product brand packaging and promotion is taken as the result label. Using a random forest algorithm, a prediction model of the influence of changes in tax regulations on product brand packaging promotion is constructed, and the changes in design features, materials, promotion channels and content of cigarette product brand packaging and promotion under different tax and tobacco control regulations are determined. According to the prediction results, the specific influence of tax and tobacco control regulations on cigarette product brand packaging and promotion is determined, including the changes in design features, the changes in materials, the adjustment of promotion channels and the changes in promotion content.

[0117] For example, assume that the product brand packaging specifications attributes of 100 cigarette brands are crawled, and the following data is obtained. Brand A has attributes of length 10 cm, width 5 cm, and height 2 cm; brand B has attributes of length 8 cm, width 4 cm, and height 2 cm; and brand Z has attributes of length 12 cm, width 6 cm, and height 3 cm. According to industry regulations, the impact of cigarette product brand packaging and promotion is taken as the result label. For example, assume that the impact of tax rate changes on product brand packaging and promotion can be divided into 3 levels, high, medium, and low. 10 brands are randomly selected, and their product brand packaging and promotion impact levels are manually labeled. Brand A is medium, brand B is low, and brand Z is high. Using the random forest algorithm, a prediction model of the impact of industry regulation changes on product brand packaging and promotion is constructed. The product brand packaging specifications, product brand packaging design features, printing information, product brand packaging materials, promotion channels, and promotion content are used as features, and the impact level of product brand packaging and promotion is used as the label. Through training the model, the impact of other brands' product brand packaging and promotion under different tax rates and industry regulations can be predicted. For example, using the random forest algorithm to train the model and make predictions, the following impact level prediction results are obtained. Brand K is high, brand L is medium, and brand Z is low. According to the prediction results, the changes in design features, materials, promotion channels, and content of cigarette product brand packaging and promotion under different tax rates and industry requirements can be determined. For example, in the industrial and commercial aspects, if the tax rate and industry regulation changes result in an increase in the prediction result of "high" brands, it can be inferred that in this case, the design features of cigarette product brand packaging may be more unique, the materials may be more high-end, the promotion channels may use more high-end media, and the promotion content may emphasize quality and uniqueness more.

[0118] Using the random forest algorithm, the impact of tobacco tax rates and tobacco tax revenue on cigarette product brand packaging is determined, and the trend of changes in manufacturer product brand packaging cost pressure and investment under different tax rates is judged.

[0119] According to the tobacco tax regulations and the cigarette product brand packaging attributes, obtain the tobacco tax rate and the cigarette product brand packaging attribute data. Clean and process the data, including handling missing values, outliers, and duplicates, to ensure the quality and accuracy of the data. Visualize the data, perform descriptive statistics and exploratory analysis, and qualitatively analyze the relationships and trends between the attributes. Based on the tobacco tax rate and the cigarette product brand packaging attribute data, use the random forest algorithm to build a tobacco tax revenue prediction model. According to the cigarette product brand packaging cost data, predict the tobacco tax revenue. According to the cigarette sales volume and cigarette brand data, predict the impact of the cigarette product brand packaging attributes on the tobacco tax revenue. According to the cigarette product brand packaging cost data, predict the trend of the change in tobacco tax revenue under different tax rates. According to the model results, determine the trend of changes in product brand packaging cost pressure and investment of manufacturers under different tax rates.

[0120] For example, according to the tobacco tax regulations and the cigarette product brand packaging attributes, obtain the tobacco tax rate and the cigarette product brand packaging attribute data. Obtain the cigarette product brand packaging attribute data and the corresponding tobacco tax rate data of 100 brands from relevant government departments. Clean and process the data, including handling missing values, outliers, and duplicates, to ensure the quality and accuracy of the data. Find that the cigarette product brand packaging attribute data of 10 brands has missing values, which can be supplemented by interpolation method. Visualize the data, perform descriptive statistics and exploratory analysis, and qualitatively analyze the relationships and trends between the attributes. Draw a histogram and box plot of the tobacco tax rate to understand the distribution of the tax rate and the existence of outliers. Use the random forest algorithm to build a tobacco tax revenue prediction model. Use the cigarette product brand packaging attributes as feature variables and the tobacco tax revenue as target variable to train the random forest model and obtain the prediction model. According to the cigarette product brand packaging cost data, predict the tobacco tax revenue. According to the cigarette product brand packaging cost data and the established prediction model, input different product brand packaging cost data to obtain the corresponding predicted tax revenue. According to the cigarette sales volume and cigarette brand data, predict the impact of the cigarette product brand packaging attributes on the tobacco tax revenue. By analyzing the relationship between the cigarette sales volume, cigarette brand data, and the tobacco tax revenue, the impact of different product brand packaging attributes on the tax revenue can be evaluated. According to the cigarette product brand packaging cost data, predict the trend of the change in tobacco tax revenue under different tax rates. For example, by predicting the tobacco tax revenue under different tax rates through the model, a trend curve is obtained, which can observe the impact of tax rate increase or decrease on tax revenue. According to the model results, determine the trend of changes in product brand packaging cost pressure and investment of manufacturers under different tax rates. By analyzing the model results, it can be found that the product brand packaging cost pressure of manufacturers is larger under high tax rate, and the investment needs to be increased to reduce the cost.

[0121] S103, using public opinion analysis tools, analyzing the acceptance of tobacco product brand packaging and health warnings by consumers according to discussions on social media platforms and professional health websites.

[0122] Further as an optional implementation, using public opinion analysis tools, analyzing the acceptance of tobacco product brand packaging and health warnings by consumers according to discussions on social media platforms and professional health websites, which specifically includes:

[0123] S1031, crawling comment text on social media platforms and professional health websites, and collecting evaluation text of consumers on the appearance design, information transmission effect and acceptance of health warnings of tobacco product brand packaging;

[0124] S1032, using TF-IDF algorithm, extracting keywords related to the attractiveness of tobacco product brand packaging, the fit of brand image, and the consistency of design style and personal aesthetics;

[0125] S1033, using TextRank algorithm, determining the acceptance of different brand packaging and health warnings by consumers;

[0126] S1034, using sentiment analysis algorithm, analyzing the influence of health warnings on smoking behavior and the concern for personal health attributes considered by consumers;

[0127] S1035, counting the keyword frequency of comments on the environmental friendliness of tobacco product brand packaging, analyzing the recognition of consumers on the environmental friendliness of brand packaging materials and the acceptance of environmentally friendly brand packaging.

[0128] Specifically, comment text on social media platforms is crawled, and evaluation text of consumers on the appearance design, information transmission effect, and acceptance of health warnings of cigarette product brand packaging is collected. Using TF-IDF algorithm, keywords related to the attractiveness of cigarette product brand packaging, the fit of brand image, and the consistency of design style and personal aesthetics are extracted. Comment text of consumers on different brand cigarette product brand packaging on social media is crawled, and TextRank algorithm is used to determine the preference and evaluation of consumers on different product brand packaging, including appearance design, information transmission effect, and acceptance of health warnings. Discussion text on social media and professional health websites is crawled, and the recognition of consumers on health warnings is analyzed using sentiment analysis algorithm. The influence of health warnings on smoking behavior and the concern for personal health attributes considered by consumers are analyzed. The keyword frequency of comments on the environmental friendliness of cigarette product brand packaging is counted, and the recognition of consumers on the environmental friendliness of product brand packaging materials and the acceptance of environmentally friendly product brand packaging are analyzed.

[0129] For example, 1000 comments on cigarette product brand packaging were crawled from microblogs, of which 500 comments thought that the appearance design of a certain brand of cigarette product brand packaging was attractive. 400 comments thought that the brand packaging of a certain brand of cigarette product brand was in line with the brand image, and 100 comments thought that the design style of a certain brand of cigarette product brand packaging was in line with personal aesthetics. For example, using the TF-IDF algorithm, keywords related to consumer attraction to cigarette product brand packaging, compatibility with brand image, and design style in line with personal aesthetics are extracted. Key words such as "exquisite", "fashionable", "unique" are extracted from the comment text. 1000 comments were crawled from microblogs, BBS and other platforms, of which 500 comments thought that the appearance design of a certain brand of cigarette product brand packaging was attractive. 300 comments thought that the information transmission effect of a certain brand of cigarette product brand packaging was good, and 200 comments thought that the health warning of a certain brand of cigarette product brand packaging was highly accepted. For example, using the TextRank algorithm, the preferences and evaluations of consumers for different product brand packaging are determined. Key words such as "beautiful", "attractive", "simple and clear" are extracted from the comment text. The comment texts on social media and professional health websites are crawled, for example, 1000 texts are crawled from microblogs and health home websites. Of which 400 comments thought that the health warning had an impact on smoking behavior, and 600 comments thought that the health warning had a high degree of concern for their own health. Using sentiment analysis algorithms, such as using sentiment dictionary method, the degree of influence of health warning on smoking behavior and the degree of concern for their own health are analyzed. For example, through sentiment analysis, it is concluded that the proportion of comments that think health warning has a positive impact on smoking behavior is 60%. The keyword frequency of consumer comments on the environmental friendliness of cigarette product brand packaging is counted, for example, the keyword frequency of "environmentally friendly", "degradable", "green product brand packaging" is counted from the comment text, and "environmentally friendly" appears 300 times. Analyze the degree of recognition of the environmental friendliness of product brand packaging materials and the acceptance of environmentally friendly product brand packaging. For example, through keyword frequency statistics, it is concluded that the proportion of comments that think product brand packaging materials are environmentally friendly is 30%.

[0130] S104, according to the online feedback and comments of consumers on tobacco product brand packaging, using sentiment analysis technology, the preference degree information of consumers for different brand packaging design elements is mined.

[0131] Further, as an optional implementation, according to the online feedback and comments of consumers on tobacco product brand packaging, using sentiment analysis technology, the preference degree information of consumers for different brand packaging design elements is mined, which specifically includes:

[0132] S1041, according to the online feedback and comments of consumers, data containing evaluations and emotional expressions of brand packaging colors are obtained;

[0133] S1042, using an emotional analysis algorithm, analyzing the degree of preference and emotional response of consumers to different colors, and determining the preference information of consumers for different color packaging designs;

[0134] S1043, according to the online feedback and comments of consumers, data containing evaluations and emotional expressions of brand packaging images and patterns are obtained;

[0135] S1044, using an emotional analysis algorithm, analyzing the degree of preference and emotional response of consumers to different images and patterns, and determining the preference information of consumers for different image and pattern packaging designs;

[0136] S1045, according to the online feedback and comments of consumers, data containing evaluations and emotional expressions of brand packaging fonts and layouts are extracted;

[0137] S1046, using an emotional analysis algorithm, analyzing the degree of preference and emotional response of consumers to different fonts and layouts, and determining the preference information of consumers for different font and layout packaging designs;

[0138] S1047, according to the online feedback and comments of consumers, data containing evaluations and emotional expressions of brand packaging materials and textures are extracted;

[0139] S1048, using an emotional analysis algorithm, analyzing the degree of preference and emotional response of consumers to different materials and textures, and determining the preference information of consumers for different material and texture packaging designs.

[0140] Specifically, online feedback and reviews of consumers are crawled to obtain data containing evaluations and emotional expressions of product brand packaging colors. Using sentiment analysis algorithms, the degree of consumer preference and emotional response to different colors is determined to find out which colors can evoke joy and excitement in consumers and which colors can bring calmness or comfort to consumers. The degree of consumer preference for product brand packaging images and patterns is obtained. Based on online feedback and reviews of users, data containing evaluations and emotional expressions of product brand packaging images and patterns are obtained. Using sentiment analysis algorithms, the degree of consumer preference for different images and patterns is determined to find out which images and patterns can attract the interest and attention of consumers and which images and patterns can bring good emotional experiences to consumers. Online feedback and reviews of consumers are crawled to extract data containing evaluations and emotional expressions of product brand packaging fonts and layout designs. Using sentiment analysis, the degree of consumer preference for different fonts and layout designs is determined to find out which fonts and layout designs can convey specific emotions and styles and which fonts and layout designs can cause consumers to feel confused or uncomfortable. Based on online feedback and reviews of consumers, data containing evaluations and emotional expressions of product brand packaging materials and textures are extracted. Using sentiment analysis algorithms, the degree of consumer preference and response to different materials and textures is determined to find out which materials and textures can bring comfort and quality to consumers and which materials and textures can make consumers feel cheap or uncomfortable.

[0141] For example, according to the crawled consumer online feedback and comment data, the following values can be obtained. The degree of preference and emotional response of product brand packaging color, 30% of consumers express their preference for red product brand packaging, and they think that red can arouse their excitement and pleasure. 40% of consumers express their preference for blue product brand packaging, and they think that blue can bring them calmness and comfort. 30% of consumers express their preference for green product brand packaging, and they think that green can bring them relaxation and naturalness. The degree of preference of product brand packaging image and pattern, 55% of consumers like product brand packaging with animal patterns, and they think that such patterns can arouse their interest and attention. 30% of consumers like product brand packaging with flower patterns, and they think that such patterns can bring them good emotional experience. 15% of consumers like simple geometric patterns, and they think that such patterns can convey specific emotions and styles. The degree of preference of product brand packaging font and layout design, 55% of consumers like to use round and readable fonts, and they think that such fonts can convey friendly and professional emotions. 30% of consumers express their preference for bold fonts, and they think that such fonts can bring them power and determination. 15% of consumers express their preference for handwriting style fonts, and they think that such fonts can convey creativity and unique emotions. The degree of preference and response of product brand packaging material and texture, 60% of consumers like smooth and textured product brand packaging materials, and they think that such materials can bring them comfort and quality. 40% of consumers express their preference for metal textured product brand packaging, and they think that such texture can make the product more high-end and luxurious. By analyzing and counting these data, it can be understood that in the business aspect, consumers' preference and emotional response to different colors, images, fonts and materials can help to design product brand packaging that meets consumers' needs and expectations.

[0142] S105, according to market sales data, analyzing consumer demand for different quality and type of brand packaging, and determining the diversification direction information of tobacco product brand packaging.

[0143] Further as an optional implementation, according to market sales data, analyzing consumer demand for different quality and type of brand packaging, and determining the diversification direction information of tobacco product brand packaging, which specifically includes:

[0144] S1051, obtaining consumer purchase history data, preference data and evaluation data of tobacco products;

[0145] S1052, according to the consumer's purchase history data and preference data, obtaining the consumer's demand for different quality and type of brand packaging;

[0146] S1053, according to the tobacco products purchased by the consumer and the corresponding brand packaging characteristics, obtaining the brand packaging demand of the consumer for the tobacco product characteristics and brand image;

[0147] S1054, according to the purchase history data and evaluation data of different consumers, using a collaborative filtering algorithm to analyze the preference and correlation of different tobacco product brand packaging attributes of the consumer, and determining the preference degree and requirement of the consumer for the safety, sustainability, visual appeal and convenience brand packaging attributes;

[0148] S1055, according to the preference and correlation of the consumer for each tobacco product brand packaging attribute, using a collaborative filtering algorithm to analyze the tobacco product brand packaging attributes that the consumer values most, and determining the diversification direction information of the tobacco product brand packaging.

[0149] Specifically, the purchase history, preference and evaluation data of the consumer purchasing tobacco products are obtained. According to the purchase history and preference data of the consumer, the data of the consumer's demand for different quality and type of tobacco product brand packaging is obtained. According to the tobacco products purchased by the consumer and the corresponding product brand packaging characteristics, the product brand packaging demand of the consumer for the tobacco product characteristics and brand image is obtained. According to the purchase history and evaluation data of different consumers, using a collaborative filtering algorithm to analyze the preference and correlation of different tobacco product brand packaging attributes of the consumer, and determining the preference degree and requirement of the consumer for the safety, sustainability, visual appeal and convenience product brand packaging attributes. According to the preference and correlation of the consumer for each tobacco product brand packaging attribute, using a collaborative filtering algorithm to determine the tobacco product brand packaging attributes that the consumer values most, and determining the focus and characteristics of the tobacco product brand packaging.

[0150] For example, assume a consumer has purchased the following tobacco products. Brand A cigarettes, with a purchase history of 1 pack in January, 2 packs in February, and 3 packs in March. Brand B cigarettes, with a purchase history of 2 packs in January, 2 packs in February, and 2 packs in March. Brand C cigarettes, with a purchase history of 3 packs in January, 3 packs in February, and 3 packs in March. From this consumer's purchase history, the degree of preference for different brands of tobacco products can be calculated. By calculating the average purchase amount for each brand, the following results can be obtained. The average purchase amount for Brand A is 2 packs, the average purchase amount for Brand B is 2 packs, and the average purchase amount for Brand C is 3 packs. From the average purchase amount, it can be seen that the consumer has a higher preference for Brand C tobacco products. In addition, assume that a survey was conducted to ask the consumer about the degree of preference for different quality and type of tobacco product brand packaging. By counting the survey results, the following data is obtained. In terms of safety in industry, the consumer gives a score of 8 out of 10. In terms of sustainability in industry and commerce, the consumer gives a score of 6. In terms of visual appeal in commerce, the consumer gives a score of 9. In terms of convenience in commerce, the consumer gives a score of 7. From the consumer's scores for product brand packaging attributes, it can be determined that he has a higher preference for visual appeal and safety product brand packaging attributes. Further, using a collaborative filtering algorithm, combined with the purchase history and evaluation data of other consumers, the consumer's preferences and correlations for different tobacco product brand packaging attributes can be analyzed. Other consumers also like Brand C tobacco products and have a higher score for visual appeal, so it can be inferred that the consumer also has a higher preference for visual appeal product brand packaging attributes. Based on the consumer's preferences and correlations for each tobacco product brand packaging attribute, a collaborative filtering algorithm can be used to determine the tobacco product brand packaging attribute that the consumer values most. If the consumer has a higher preference for safety and visual appeal product brand packaging attributes, it can be determined that the focus of the tobacco product brand packaging should be on safety and visual appeal.

[0151] Using the K-means clustering algorithm, the design of cigarette product brand packaging at different prices is determined for different consumer groups based on consumption levels.

[0152] The consumption level data of different consumer groups and the attribute data related to the packaging design of cigarette product brands are obtained, the consumption level data of different consumer groups includes consumption amount and purchase frequency, and the attribute data related to the packaging design of cigarette product brands includes color, shape, text and pattern. The obtained consumption level data of different consumer groups and the attribute data related to the packaging design of cigarette product brands are data cleaned to remove outliers and missing values, and feature selection is performed, and the selected features are scaled to ensure that the differences between different attributes will not affect the clustering results. The consumption level data of different consumer groups and the attribute data related to the packaging design of cigarette product brands are analyzed by using K-means clustering algorithm, and the number of clusters is determined. Through iterative calculation, the cluster to which each data point belongs is obtained. Analyze the clustering results, and according to the center point and member distribution of the cluster, judge the consumption level of different consumer groups and the preference and preference for the packaging design of cigarette product brands. Through comparison and verification with actual market data and consumer feedback, the reliability of the clustering results is determined. If it is found that the clustering results do not meet the expectations, adjust the feature selection or K value parameter, and run the clustering algorithm again for analysis.

[0153] For example, there are three different consumer groups A, B and C. Their consumption level data and attributes related to cigarette product brand packaging design have been collected. The consumption level data is that the consumption amount of consumer group A is 200 yuan, and the purchase frequency is once a week. The consumption amount of consumer group B is 500 yuan, and the purchase frequency is three times a week. The consumption amount of consumer group C is 1000 yuan, and the purchase frequency is five times a week. The attributes related to cigarette product brand packaging design are that consumer group A prefers light color product brand packaging design, prefers square shape, and prefers concise text. Consumer group B prefers medium color product brand packaging design, prefers rectangular shape, and prefers descriptive text. Consumer group C prefers dark color product brand packaging design, prefers circular shape, and prefers fashionable text. During data cleaning, check if there are outliers in the consumption amount data, such as negative or excessively large amounts. At the same time, check if there are missing values in the purchase frequency data, such as blanks or unreasonable frequencies. In feature selection, you can select the most relevant features for clustering analysis based on correlation, variance or other statistical indicators. You can select consumption amount and purchase frequency as features of consumption level data, and select color, shape, text and pattern as features of product brand packaging design attributes. Feature scaling is to normalize the values of different attributes to ensure that their differences do not affect the clustering results. You can scale the values of consumption amount and purchase frequency to the range of 0 to 1. When using the K-means clustering algorithm, you need to select the number of clusters K. You can manually select the K value and run the algorithm multiple times, then select the best K value based on the stability and effectiveness of the clustering results. You can try K=2, K=3 and K=4, then analyze the consumption level and preferences of different consumer groups for product brand packaging design based on the center points and member distributions of the clustering clusters. Finally, you can compare and verify the clustering results with actual market data and consumer feedback. If the clustering results are not as expected, you can adjust the feature selection or K value parameters and run the clustering algorithm again for analysis. You may find that the clustering results are not clear or accurate enough, and you may need to adjust the feature selection, such as adding more features related to consumer behavior, or adjusting the K value to better capture the differences between different consumer groups.

[0154] S106, crawling the brand packaging design data of competitors, using convolutional neural network to identify brand packaging images, analyzing market brand packaging trends and competitive situation.

[0155] Further, as an optional implementation, crawling the brand packaging design data of competitors, using convolutional neural network to identify brand packaging images, analyzing market brand packaging trends and competitive situation, which specifically includes:

[0156] S1061, obtain a competitor's brand packaging design image dataset and label, to obtain a label data containing brand packaging color, shape, pattern and text;

[0157] S1062, according to the obtained brand packaging design image dataset and label data, adopt convolutional neural network algorithm, build brand packaging image recognition model, determine the brand packaging color, shape, pattern and text attribute of new brand packaging image;

[0158] S1063, the brand packaging design image of the competitor is divided into several clusters, each cluster represents a brand packaging trend or competitive situation, and a K-means clustering algorithm is used to build a brand packaging feature clustering analysis model;

[0159] S1064, based on the feature attribute of the cluster, determine the brand packaging design trend in the market and the brand packaging strategy of the competitor.

[0160] Specifically, the product brand packaging design image dataset of the competitor is obtained and labeled, including product brand packaging color, shape, pattern, text attribute. According to the obtained product brand packaging design image dataset and label, adopt convolutional neural network algorithm, build product brand packaging image recognition model, determine the product brand packaging color, shape, pattern, text attribute of new product brand packaging image. The product brand packaging design image of the competitor is divided into several clusters, each cluster represents a product brand packaging trend or competitive situation, and a K-means clustering algorithm is used to build a product brand packaging feature clustering analysis model. Based on the feature attribute of the cluster, determine the product brand packaging design trend in the market and the product brand packaging strategy of the competitor. The model features include the following attributes. Product brand packaging color, including color type, lightness, saturation attribute. Product brand packaging shape, including square, circle, rectangle shape attribute. Pattern, including pattern, icon, stripe pattern attribute. Text, including brand name, product characteristics, usage instruction text attribute. The features of the image are extracted by the convolution layer of the convolutional neural network model, and the specific image shape and texture features are identified. Adjust the weight of the convolutional neural network model, and adjust the structure of the convolutional neural network model to optimize the feature recognition accuracy.

[0161] For example, assume that a dataset of 1000 images of competitor product brand packaging designs is obtained and labeled, including product brand packaging color, shape, pattern, and text attributes. First, a convolutional neural network algorithm can be used to build a product brand packaging image recognition model. The dataset is divided into a training set and a test set, and the training set is used to train the model, and then the test set is used to evaluate the accuracy of the model. Assuming that through training, a product brand packaging image recognition model with an accuracy of 90% is obtained. Next, the competitor's product brand packaging design images can be divided into several clusters, each representing a product brand packaging trend or competitive situation. The K-means clustering algorithm can be used to divide the images into 10 clusters. Each cluster represents a different product brand packaging design trend or competitor product brand packaging strategy. For example, assume that cluster 1 has the following characteristic attributes: red, square, patterned, and brand name text attributes; cluster 2 has the following characteristic attributes: blue, circular, icon pattern, and product feature text attributes. By analyzing the characteristic attributes of each cluster, the product brand packaging design trends and competitor product brand packaging strategies in the market can be determined. Further, the accuracy of feature recognition can be optimized by adjusting the weights and structure of the convolutional neural network model. For example, the number of convolutional layers can be increased or the size of the convolutional kernel can be adjusted to improve the accuracy of the model. Assuming that by adjusting the structure and weights of the model, the accuracy of the product brand packaging image recognition model is improved to 95%.

[0162] According to the product brand packaging design images of competitors, the convolutional neural network is used to recognize the product brand packaging design style, color and pattern preferences, and font and logo.

[0163] A dataset of competitor product brand packaging design images is obtained. According to the product brand packaging design image dataset, a convolutional neural network model is trained to label different product brand packaging design styles. According to the trained model, new product brand packaging design images are predicted to determine their product brand packaging design style. According to the prediction results, the attributes of product brand packaging design style are determined, including simple, modern, and retro. According to the product brand packaging design image dataset, a convolutional neural network model is used to label different color and pattern preferences of the image dataset. According to the trained model, new product brand packaging design images are predicted to determine their color and pattern preferences. According to the prediction results, the color and pattern attributes of product brand packaging design are determined, including bright colors, geometric patterns, and patterns. According to the product brand packaging design image dataset, a convolutional neural network model is used to label different fonts and logos. According to the trained model, new product brand packaging design images are predicted to determine their font and logo, including different font styles, trademarks or logos. According to the recognition results, the product brand packaging design strategy of competitors is determined.

[0164] For example, a dataset of competitor product brand packaging design images is obtained, containing 1000 images of different styles of product brand packaging designs. A convolutional neural network model is trained using this dataset to be able to predict the product brand packaging design style of a new product brand packaging design image. Before training, the dataset needs to be labeled, i.e. each image is labeled with the corresponding label. Suppose the product brand packaging design style is divided into three categories: minimalist, modern, and retro. Each image is assigned a label, for example, images of minimalist style are labeled 0, images of modern style are labeled 1, and images of retro style are labeled 2. After labeling, the convolutional neural network model is trained. A classic convolutional neural network structure with convolutional layers, pooling layers, and fully connected layers is selected. Suppose during training, 80% of the image data is used for training, and the remaining 20% is used to verify the accuracy of the model. After multiple iterations of training, a model with an accuracy of 85% is obtained. This model is used to predict the product brand packaging design style of a new product brand packaging design image. For example, a new product brand packaging design image is obtained, and the model predicts that it belongs to the modern style. According to this prediction result, the style attribute of this product brand packaging design is determined to be modern. Similarly, a convolutional neural network model can be trained using the same dataset to predict the color and pattern preferences of product brand packaging design. Suppose the color and pattern preferences are divided into three categories: bright colors, geometric patterns, and patterns. Each image in the dataset is assigned a corresponding label. For example, images of bright colors are labeled 0, images of geometric patterns are labeled 1, and images of patterns are labeled 2. Similarly, 80% of the data is used for training, and the remaining 20% is used to verify the accuracy of the model. After multiple iterations of training, a model with an accuracy of 90% is obtained. This model is used to predict the color and pattern preferences of a new product brand packaging image. Suppose after prediction, the color preference is bright colors and the pattern preference is geometric patterns. According to this prediction result, the color attribute of this product brand packaging design is determined to be bright colors, and the pattern attribute is determined to be geometric patterns. Finally, the selection of fonts and logos in the product brand packaging design strategy of competitors can be analyzed. Similarly, a convolutional neural network model can be trained and predicted to determine the font and logo used in the product brand packaging design. Suppose the font and logo are divided into three categories: different font styles, trademarks, or logos. Each image in the dataset is assigned a corresponding label, for example: font style is 0, trademark or logo is 1. Similarly, 80% of the data is used for training, and the remaining 20% is used to verify the accuracy of the model. After multiple iterations of training, a model with an accuracy of 95% is obtained. Now, when a new product brand packaging design image is obtained, this model is used to predict its font and logo.Assuming that the prediction is made, the font style is 0, and the trademark or logo is 1. According to the prediction result, it is determined that the competitor has selected the font style with the number 0 and the trademark or logo with the number 1 in the product brand packaging design.

[0165] The K-means clustering algorithm is used to cluster the identified attributes and determine the popular product brand packaging trends in the market, to judge the mainstream style and characteristics of the current market product brand packaging design.

[0166] According to the color, pattern, font, and material attributes in product brand packaging design, K cluster centers are selected as initial values. The initial cluster centers are determined by random selection or using genetic algorithm. According to the distance between each sample and each cluster center, the sample is assigned to the cluster with the closest distance. The Euclidean distance or Manhattan distance between the sample and the cluster center is calculated to determine the cluster assignment method of the sample. For each cluster, the average value of all samples is calculated as the new cluster center. According to the attribute values of the samples in the cluster, the average value is calculated to update the cluster center. The genetic algorithm is used to optimize the cluster center until the cluster center and cluster division no longer change, or the preset iteration number is reached. According to the new cluster center, the sample is re-assigned to the cluster, and the cluster center is updated. According to the color, pattern, font, and material attributes in product brand packaging design, the final clustering analysis result is determined, including the specific cluster division and the cluster center of each cluster. According to the common characteristics of the samples in the cluster, the popular product brand packaging trends in the market are determined, including popular colors, prominent patterns, and popular fonts.

[0167] For example, assume there are 100 product brand packaging samples, each with color, pattern, font, material attributes, and we want to divide these samples into K clusters and determine the cluster center of each cluster. First, we need to select K initial cluster centers, which can be determined using random selection or genetic algorithm. Suppose 3 initial cluster centers are selected, then the distance of each sample from these cluster centers is calculated. Suppose sample 1 has a Euclidean distance of 5 from cluster center 1, a Euclidean distance of 3 from cluster center 2, and a Euclidean distance of 6 from cluster center 3. According to the nearest cluster center, sample 1 is assigned to cluster 2. Next, the average of all samples in each cluster is calculated as the new cluster center. Suppose cluster 2 has samples 1, 2, and 3, and their color attributes are red, green, and blue, respectively. Then the new cluster center is (red + green + blue) / 3, i.e. (1 + 2 + 3) / 3 = 2. Then the cluster centers are optimized using genetic algorithm. Through iterative calculation, until the cluster center and cluster division no longer change, or the preset number of iterations is reached. Finally, according to the new cluster center, the samples are re-assigned to the clusters, and the cluster centers are updated, obtaining the specific cluster division and cluster center of each cluster. According to the common characteristics of the samples in the cluster, the popular product brand packaging trend in the market is determined. The samples in cluster 1 all have bright colors, bold patterns, and fashionable fonts, so it can be considered that the popular product brand packaging trend in the market is bright colors, bold patterns, and fashionable fonts.

[0168] S107、According to the advertising campaign, tobacco product brand packaging design and sales data of tobacco products, determine the relationship curve of sales performance and different brand packaging and different promotion strategy.

[0169] Further as an optional implementation, according to the advertising campaign, tobacco product brand packaging design and sales data of tobacco products, determine the relationship curve of sales performance and different brand packaging and different promotion strategy, which specifically includes:

[0170] S1071, obtain sales data and brand packaging type information, including brand packaging type, sales performance and brand;

[0171] S1072, through data preprocessing, encode sales data and brand packaging type, and perform association rule mining;

[0172] S1073, construct a frequent item set of different brand packaging types and sales performance, the frequent item set representing the association relationship between different brand packaging and sales performance;

[0173] S1074, construct association rules, and extract association rules from the frequent item set according to a confidence threshold;

[0174] S1075、According to the support degree and the confidence degree, the importance of the association rule is evaluated, and the association rule with practical significance is screened out.

[0175] S1076、According to the obtained association rule, the relationship curve between the sales performance and different brand packaging and different promotion strategies is determined.

[0176] Specifically, sales data and product brand packaging type information are obtained, including product brand packaging type, sales performance, and brand. Through data preprocessing, the sales data and product brand packaging type are encoded, and association rule mining is performed. A frequent item set is constructed, and the frequent item set of different product brand packaging types and sales performance is calculated according to a support degree threshold. The frequent item set represents the association relationship between different product brand packaging and sales performance. An association rule is constructed, and the association rule is extracted from the frequent item set according to a confidence degree threshold. According to the support degree and the confidence degree, the importance of the association rule is evaluated, and the association rule with practical significance is screened out. According to the obtained association rule, the relationship between different product brand packaging and sales performance is determined.

[0177] For example, in the sales data of a cigarette company, there are different brands of cigarette products, different product brand packaging types, and corresponding sales performance. The following steps are performed for association rule mining: determine relevant attributes: product brand packaging type, sales performance, brand. Crawl the sales data of cigarette products, including sales volume, sales amount, and other indicators, as well as the corresponding product brand packaging type and brand of each product. Encode the sales data and product brand packaging type, and convert them into a format suitable for association rule mining. For example, use one-hot encoding to convert the product brand packaging type into a binary feature vector. Build frequent itemsets: according to the set support threshold, calculate the frequent itemsets of different product brand packaging types and sales performance. Frequent itemsets are combinations of product brand packaging types and sales performance that frequently appear in sales data, reflecting their association. Build association rules: extract association rules from frequent itemsets according to the set confidence threshold. Association rules are a formal expression of the relationship between product brand packaging types and sales performance, such as "if a product uses high-end product brand packaging, the likelihood of improved sales performance is greater". Evaluate the importance of association rules: evaluate the importance of association rules through support and confidence. Support reflects the frequency of association rules in the entire sales data, and confidence indicates the likelihood of sales performance given a product brand packaging type. Filter out association rules with practical significance: according to the evaluation results, filter out association rules with practical significance. For example, select rules with high support and confidence, or rules related to business goals. In business, if it is found from association rules that high-end product brand packaging and sales performance have strong association, it can be considered to use high-end product brand packaging in products to improve sales performance.

[0178] In some optional embodiments, through association rule mining, it is determined whether there is a certain synergistic effect between advertising activities and product brand packaging design to jointly promote the improvement of sales performance.

[0179] Specifically, data of advertising campaigns and product brand packaging design are acquired, including advertising channels, advertising types, advertising content, advertising duration, advertising investment, product brand packaging appearance, product brand packaging material, product brand packaging logo, product brand packaging size, product brand packaging characteristic attributes. The data is cleaned, de-duplicated, and missing value processed. An association rule mining Apriori algorithm is used to mine association rules from the preprocessed data. Appropriate minimum support and minimum confidence thresholds are set to filter out association rules that meet the conditions. The association rules mined are evaluated, and the confidence and support evaluation indicators of each rule are calculated. The association rules with the highest confidence and support are selected as the rules of focus. According to the mining results and evaluation indicators, texts explaining the significance of the association rules and the underlying association relationships are generated.

[0180] For example, suppose that data on advertising campaigns and product brand packaging designs over a period of time is obtained from a company's marketing department. The following data is obtained: advertising channels include television, radio, outdoor advertising, the Internet, magazines, newspapers, etc. Types include product promotion, brand promotion, promotional activities, public service advertising, etc. Advertising content is specific advertising copy or a description of the visuals. Advertising length is in seconds, e.g., 30 seconds, 60 seconds, etc. Advertising investment is in ten thousand yuan, indicating the amount of investment the company made in the advertisement. Product brand packaging appearance is the form of the product brand packaging, e.g., box, bottle, bag, etc. Product brand packaging material is the material used for the product brand packaging, e.g., paper box, plastic bottle, iron can, etc. Product brand packaging identification is the relevant information marked on the product brand packaging, e.g., product name, brand identification, production date, etc. Product brand packaging size is the size of the product brand packaging, e.g., specific values for length, width, and height. Product brand packaging characteristic attributes: characteristics or attributes of the product brand packaging, e.g., moisture-proof, shockproof, easy to open, etc. The data is cleaned to remove duplicate data and handle missing values. For example, for the advertising investment item, if a piece of data is missing the investment amount, the average value can be used to fill in the missing value or the piece of data can be discarded. The Apriori algorithm for association rule mining is used to analyze the data. Suppose the minimum support is set to 3 and the minimum confidence is set to 8. It is hoped that frequent item sets and association rules will be mined to understand the relationship between advertising campaigns and product brand packaging designs. After algorithm operation, some association rules are obtained. For example, the rule "high advertising investment -> advertising channel television" has a confidence of 9 and a support of 4. This means that when the advertising investment is high, it is more likely to choose television as the advertising channel. The rule "product brand packaging material paper box -> product brand packaging appearance box" has a confidence of 8 and a support of 3. This means that when the product brand packaging material is paper box, it is more likely to choose box as the appearance form. Through analysis of the mining results and evaluation indicators, it can be concluded that in the case of high advertising investment, television is more likely to be chosen as the advertising channel, and when paper box is chosen as the product brand packaging material, box is more likely to be chosen as the appearance form.

[0181] In some optional embodiments, the Apriori algorithm is used to design different price tobacco product arrangement combinations for tobacco display counters to maximize tobacco sales.

[0182] Specifically, the sales data of tobacco products are acquired, including product price, brand, category, product brand packaging, and promotion information. Through data cleaning and removing duplicate data, the accuracy and consistency of the data are ensured. The Apriori algorithm is used to analyze the sales data to find frequent item sets and association rules. According to the support and confidence thresholds, the importance of frequent item sets and association rules is determined. According to the sales maximization target, several highest sales association rules are selected. According to the selected association rules, the arrangement and combination of different price tobacco products are designed. The effect of the designed tobacco display counter is evaluated, and adjustments and optimizations are made according to consumer feedback and changes in sales data.

[0183] For example, the following sales data is obtained from a tobacco shop: product prices are [10 yuan, 20 yuan, 15 yuan, 25 yuan, 30 yuan, 20 yuan], brands are [A brand, B brand, C brand, A brand, B brand, C brand], types are [cigarettes, cigars, cigarettes, cigars, cigarettes, cigars], product brand packaging is [box, box, bag, bag, box, box], and promotional activities are [no activity, no activity, discount, discount, no activity, discount]. First, the data needs to be cleaned and de-duplicated. After cleaning, the data is as follows: product prices are [10 yuan, 20 yuan, 15 yuan, 25 yuan, 30 yuan], brands are [A brand, B brand, C brand], types are [cigarettes, cigars], product brand packaging is [box, bag], and promotional activities are [no activity, discount]. Next, use the Apriori algorithm to analyze and find frequent itemsets and association rules. Set the threshold to support 2 and confidence 5, and the resulting frequent itemsets and association rules are as follows: frequent itemset is {product price: 20 yuan, brand: A brand, type: cigarette, product brand packaging: box}, {product price: 20 yuan, type: cigarette, product brand packaging: box, promotional activity: no activity}, association rule is {product price: 20 yuan, brand: A brand} -> {type: cigarette, product brand packaging: box}, {product price: 20 yuan, type: cigarette} -> {product brand packaging: box, promotional activity: no activity}. According to the goal of maximizing sales, the highest sales association rule can be filtered out. The formula for calculating sales is sales = product price * sales quantity. According to the sales data and association rules, the following sales can be calculated: association rule 1 sales = 20 yuan * 2 = 40 yuan, association rule 2 sales = 20 yuan * 2 = 40 yuan. Therefore, association rule 1 and association rule 2 are both the highest sales rules. According to the filtered association rules, different price tobacco product arrangements can be designed. The product price in association rule 1 can be set to 25 yuan and the product price in association rule 2 can be set to 30 yuan to increase sales. Finally, evaluate the effectiveness of the designed tobacco display counter. Through consumer feedback and changes in sales data, the prices, brands, types, product brand packaging, and promotional activities of the products can be adjusted and optimized to increase sales and meet consumer demand

[0184] S108, according to the promotional effect of tobacco product brand packaging and market feedback, using multiple regression analysis and prediction model, evaluate the third influence factor of tobacco product brand packaging and promotion strategy on market sales.

[0185] Further as an optional implementation, according to the promotional effect of tobacco product brand packaging and market feedback, using multiple regression analysis and prediction model, evaluate the third influence factor of tobacco product brand packaging and promotion strategy on market sales, which specifically includes:

[0186] S1081, obtain tobacco product brand packaging design and color related data, including sales data, market research and consumer feedback comment text of different design and color schemes;

[0187] S1082, determine the first influence coefficient of brand packaging design and color on market sales volume by using multiple regression analysis method;

[0188] S1083, obtain related data of different brand packaging materials and quality, including consumer evaluation and product protection performance of brand packaging materials and quality;

[0189] S1084, determine the second influence coefficient of brand packaging materials and quality on market sales volume by using multiple regression analysis method;

[0190] S1085, obtain related data of different propaganda channels and propaganda content, including exposure of different channels and content and consumer response to propaganda content;

[0191] S1086, determine the third influence coefficient of propaganda channels and propaganda content on market sales volume by using multiple regression analysis method;

[0192] S1087, obtain target audience's reaction data to brand packaging and propaganda strategy, including consumer's purchase willingness and brand awareness;

[0193] S1088, determine the fourth influence coefficient of target audience's reaction on market sales volume by using multiple regression analysis method;

[0194] S1089, determine the third influence factor of tobacco product brand packaging and propaganda strategy on market sales according to the first influence coefficient, the second influence coefficient, the third influence coefficient and the fourth influence coefficient.

[0195] Specifically, the cigarette product brand packaging design and color related data are acquired, including the sales data, market research and consumer feedback comment text of different design and color schemes. According to the product brand packaging design and color data, a multiple regression analysis method is used to determine the market sales. According to the significance test and explanatory power of the regression coefficient, the specific influence of the product brand packaging design and color on the market sales is determined. The relevant data of different product brand packaging materials and quality are acquired, including the consumer's evaluation of the product brand packaging materials and quality, and the product protection performance. A multiple regression analysis method is used to determine the influence coefficient of the product brand packaging materials and quality on the market sales. Through the significance test and explanatory power of the regression coefficient, the influence coefficient of the product brand packaging materials and quality on the market sales is determined. The relevant data of different propaganda channels and content are acquired, including the exposure of different channels and content, and the response of consumers to the propaganda content. Through multiple regression analysis, the influence coefficient of the propaganda channel and content on the market sales is determined. Through the significance test and explanatory power of the regression coefficient, the real and specific influence of the propaganda channel and content on the market sales is determined. The reaction data of the target audience to the product brand packaging and propaganda strategy are acquired, including the purchase willingness and brand awareness of consumers. Through multiple regression analysis method, the specific influence coefficient of the target audience's reaction on the market sales is determined. Through the significance test and explanatory power of the regression coefficient, the real and specific influence of the target audience's reaction on the market sales is determined.

[0196] For example, assume there are two different product brand packaging designs and color schemes, Scheme A and Scheme B. Conduct a regression analysis on the sales data of both schemes, with product brand packaging design and color as independent variables, and market sales as the dependent variable. Through the significance test of regression coefficients and the degree of explanation, the specific impact of product brand packaging design and color on market sales can be determined. Assume the results of the regression analysis show that the impact of product brand packaging design and color of Scheme A on market sales is more significant, with a regression coefficient of 8, while the regression coefficient of Scheme B is only 3. This means that the product brand packaging design and color of Scheme A are more popular with consumers and have a greater impact on market sales. Similarly, by collecting relevant data on different product brand packaging materials and quality, a multiple regression analysis can be conducted to determine the specific impact of product brand packaging materials and quality on market sales. If the analysis results show that high-quality product brand packaging materials have a significant positive impact on market sales, with a regression coefficient of 11, while low-quality product brand packaging materials have a regression coefficient of only 2, it can be concluded that high-quality product brand packaging materials have a greater impact on market sales. Similarly, by collecting relevant data on different promotion channels and content, a multiple regression analysis can be conducted to determine the specific impact of promotion channels and content on market sales. Assume the analysis results show that under the television advertising channel, promotion content A has the greatest impact on market sales, with a regression coefficient of 9, while under the social media channel, promotion content B has a regression coefficient of 6, it can be concluded that the impact of promotion channels and content on market sales varies. In addition, by conducting surveys and analysis targeting different target audiences, reaction data of target audiences to product brand packaging and promotion strategies can be collected to determine the impact of target audience reactions on market sales. In business terms, if the analysis results show that young people are more interested in Scheme A product brand packaging design and promotion content, with a regression coefficient of 7, while middle-aged people are more interested in Scheme B, with a regression coefficient of 5, it can be concluded that the reactions of different target audiences have different impacts on market sales. If product brand packaging design and color have a greater impact on market sales, it is recommended to optimize product brand packaging design and color schemes. If the impact of promotion channels and content on market sales varies, it is recommended to adjust the promotion strategy to better attract different target audiences.

[0197] In some optional embodiments, support vector machine algorithm is used to automatically identify and mine hot topics related to the tobacco industry, as well as keywords of product brand packaging and promotion strategies.

[0198] Specifically, a text dataset related to the tobacco industry is obtained, including news reports, social media comments, and industry reports. Ensure that each sample in the dataset has a corresponding label indicating the sample's relevance to the tobacco industry. Use TF-IDF weights to convert the text dataset into feature vectors that can be used to train a support vector machine algorithm. Standardize or normalize the extracted feature vectors to ensure that each feature has a similar scale. Divide the dataset into a training set and a test set, and use the support vector machine algorithm to train the labeled training data. Adjust the hyperparameters and use cross-validation methods to optimize the model performance. Evaluate the trained model using the test dataset, calculate the accuracy and recall rate indicators. Based on the model training results, use information gain or chi-square test methods to obtain several features that have the most impact on the tobacco industry-related topics, product brand packaging, and promotion strategies. Use the trained model to predict new text datasets, determine their relevance to the tobacco industry, and identify popular topics and keywords related to product brand packaging and promotion strategies.

[0199] For example, a dataset containing 1000 text samples can be used, where each sample has a label indicating its relevance to the tobacco industry. These samples can come from news articles, social media comments, and industry reports. Next, the text dataset can be converted into feature vectors using TF-IDF weights. By calculating the TF-IDF weight of each word in each text sample, the text data can be converted into a numerical feature vector. Then, the extracted feature vectors can be standardized or normalized to ensure that each feature has a similar scale. Next, the dataset is divided into a training set and a test set. The training set is used for model training of the support vector machine algorithm, while the test set is used to evaluate the performance of the trained model. Then, the labeled training data can be used to train the model using the support vector machine algorithm. During model training, the performance of the model can be optimized by adjusting hyperparameters and using cross-validation methods. Different kernel functions, regularization parameters, and penalty parameters can be tried, and cross-validation can be used to select the optimal combination of parameters. Then, the trained model can be evaluated using the test dataset. Accuracy and recall can be calculated to measure the performance of the model. The accuracy of the model on the test dataset is 80%, and the recall is 85%. According to the model training results, information gain or chi-square test methods can be used to obtain keywords related to the tobacco industry and product brand packaging and promotion strategies. These keywords can help understand the hot topics and key strategies of the tobacco industry. Using the chi-square test method, the keywords "health risks" and "market marketing" are obtained. Finally, the trained model can be used to predict new text datasets and determine their relevance to the tobacco industry, and identify keywords related to hot topics and key strategies. A new text sample can be input into the trained model, and the model will output the relevance of the sample to the tobacco industry, and keywords related to hot topics and key strategies can be obtained according to the keyword extraction method.

[0200] S109, according to the acceptance degree information, the preference degree information, the diversification direction information, the market brand packaging trend and the competition situation, the relationship curve, the first influence factor, the second influence factor and the third influence factor, determine the target brand packaging and the target promotion strategy of the tobacco product.

[0201] Specifically, according to market research and consumer preference data, the preference data of consumers for the appearance, material, function, and convenience of tobacco products is obtained. According to the consumer preference data, the tobacco product brand packaging design with the highest similarity to the preference data is determined. The product brand packaging is made of environmentally friendly materials, emphasizing the sustainability of the product to meet the needs of consumers for environmentally friendly products. According to the industry regulation text, the content and format of the label of the product brand packaging material in the comparison scheme are compared. Unique product brand packaging design and promotional language are used to highlight the differentiation of the product scheme from competitors. According to the differentiation result, the tobacco product promotion strategy is optimized to highlight the value and benefits of the product and convey the characteristics and functions of the product to consumers. The final optimized tobacco product brand packaging design and promotion method scheme is determined.

[0202] For example, according to market research data, it is found that 50% of consumers prefer simple design, 30% prefer luxurious design, and 20% prefer innovative design for the appearance of tobacco products. According to the consumer preference data, the tobacco product brand packaging design with the highest similarity to the preference data can be determined as simple design. In terms of product brand packaging materials, according to environmental requirements, sustainable materials are used to make product brand packaging, such as using degradable carton materials. According to the industry regulation text, the label content of the product brand packaging material should include product name, supplier information, product ingredients, health warnings, etc. The format of the label should comply with the requirements of font, size, and color. For differentiation design and promotion methods, assuming that the product brand packaging design of competitors is biased towards luxury and innovation, the product brand packaging design of the product chooses a simple style. Bold font and color can be used on the product brand packaging to highlight the simplicity and high quality of the product. At the same time, unique promotional language can be used to emphasize the uniqueness of the product and its differentiation from competitors. For example, the promotional language can be "simple is not simple, unique taste" to highlight the simple style and unique characteristics of the product. Through optimization of the tobacco product promotion strategy, the value and benefits of the product can be conveyed to consumers, such as reducing environmental pollution and providing a comfortable smoking experience. In the mode of industry synergy, the final optimized tobacco product brand packaging design and promotion method scheme can be to use simple product brand packaging design, use environmentally friendly materials to make product brand packaging, and the label content and format comply with the requirements of the industry regulations. At the same time, through unique promotional language and differentiation design, the uniqueness and value of the product are highlighted to meet the needs of consumers for environmentally friendly products.

[0203] The method steps of the embodiments of the present application are described above. It can be recognized that the embodiments of the present application crawl text data such as industry association announcements and news reports, extract attributes related to cigarette product brand packaging requirements using text mining technology, and analyze the specific influence of these attributes on product brand packaging requirements; through analysis of tobacco tax and tobacco control regulations, the influence of tax and industry regulations on cigarette product brand packaging and promotion is predicted using a random forest algorithm; using public opinion analysis tools, the acceptance of consumers to cigarette product brand packaging and health warnings is analyzed through discussions on social media and professional health websites; through sentiment analysis of online feedback and comments of consumers on product brand packaging, the preferences of consumers for different product brand packaging design elements are mined; through market sales data and using collaborative filtering algorithm, the demand of consumers for different taste, quality and type of product brand packaging is analyzed to determine the diversification direction of product brand packaging; by monitoring the product brand packaging design and promotion strategy of competitors, the product brand packaging design image is recognized using convolutional neural network, and the popular product brand packaging trend and competitive situation in the market are analyzed through K-means clustering algorithm; the relationship between advertisements, sales performance and different product brand packaging types is analyzed through association rule mining, and the specific influence of product brand packaging and promotion strategy on market sales is evaluated through multiple regression analysis and prediction model, and the Apriori algorithm is used to design different price tobacco product arrangement combinations of tobacco display counter to maximize tobacco sales; according to the preferences of consumers, market trends and industry requirements, the product brand packaging and promotion strategy is optimized, the accuracy and reliability of the product brand packaging and promotion strategy are improved, and the market competitiveness of tobacco enterprises and tobacco products is improved. The embodiments of the present application improve the income of cigarette retail customers and the satisfaction of consumers from the unique perspective of analyzing the product brand packaging and promotion strategy of the cigarette industry in a business-collaborative mode, and further promote tobacco tax and improve the image of tobacco enterprises in the society.

[0204] Reference Figure 2 The embodiments of the present application provide a tobacco product brand packaging and promotion strategy determination system based on data mining, comprising:

[0205] A first influence factor determination module is configured to crawl negative evaluation information of the cigarette industry according to industry association announcements and news, extract key information related to brand packaging requirements according to the negative evaluation information, and analyze a first influence factor on the tobacco product brand packaging according to the key information.

[0206] A second influence factor determination module is configured to analyze tobacco tax and tobacco control regulations, and predict a second influence factor of tax and industry requirements change on the tobacco product brand packaging and promotion strategy.

[0207] an acceptance degree information determining module configured to analyze the acceptance degree information of consumers on the tobacco product brand packaging and health warnings based on discussions on social media platforms and professional health websites by using an opinion analysis tool;

[0208] a preference degree information determining module configured to mine the preference degree information of consumers on different brand packaging design elements based on online feedback and comments of consumers on the tobacco product brand packaging by using sentiment analysis technology;

[0209] a diversification direction information determining module configured to analyze the demand of consumers for brand packaging of different qualities and types based on market sales data to determine the diversification direction information of the tobacco product brand packaging;

[0210] a market brand packaging trend and competitive situation determining module configured to crawl brand packaging design data of competitors, recognize brand packaging images by using a convolutional neural network, and analyze the market brand packaging trend and competitive situation;

[0211] a relationship curve determining module configured to determine the relationship curve of sales performance and different brand packaging and different promotion strategies based on advertising activities, tobacco product brand packaging design, and sales data of the tobacco product;

[0212] a third influence factor determining module configured to evaluate the third influence factor of the tobacco product brand packaging and promotion strategy on market sales based on the promotion effect and market feedback of the tobacco product brand packaging by using multiple regression analysis and a prediction model;

[0213] a brand packaging and promotion strategy determining module configured to determine the target brand packaging and target promotion strategy of the tobacco product based on the acceptance degree information, the preference degree information, the diversification direction information, the market brand packaging trend and competitive situation, the relationship curve, the first influence factor, the second influence factor, and the third influence factor.

[0214] The content in the method embodiments described above is applicable to the system embodiments, the system embodiments specifically implement the same functions as the method embodiments, and achieve the same beneficial effects as the method embodiments.

[0215] With reference to Figure 3 The embodiment of the present application provides a tobacco product brand packaging and promotion strategy determining device based on data mining, which comprises:

[0216] at least one processor;

[0217] at least one memory for storing at least one program;

[0218] When the at least one program is executed by the at least one processor, the at least one processor implements the data mining-based tobacco product brand packaging and promotion strategy determination method.

[0219] The contents in the method embodiments are applicable to the device embodiments, the device embodiments specifically implement the functions of the method embodiments, and achieve the same beneficial effects as the method embodiments.

[0220] The embodiment of the present application further provides a computer readable storage medium, wherein a processor executable program is stored, and the processor executable program is used for executing the data mining-based tobacco product brand packaging and promotion strategy determination method when executed by a processor.

[0221] The computer readable storage medium of the embodiment of the present application can execute the data mining-based tobacco product brand packaging and promotion strategy determination method provided by the method embodiments of the present application, execute the steps of any combination of the method embodiments, and has the corresponding functions and beneficial effects of the method.

[0222] The embodiment of the present application further discloses a computer program product or a computer program, which comprises computer instructions stored in a computer readable storage medium. A processor of a computer device can read the computer instructions from the computer readable storage medium, and the processor executes the computer instructions, so that the computer device executes the method shown in the embodiment. Figure 1 The method shown in the embodiment.

[0223] In some alternative embodiments, the functions / operations mentioned in the block diagram can not occur in the order mentioned in the operation diagram. For example, depending on the functions / operations involved, two blocks shown in succession can actually be executed substantially simultaneously or the blocks mentioned above can be executed in reverse order at times. In addition, the embodiments presented and described in the flowcharts of the present application are provided by way of example, and the purpose is to provide a more comprehensive understanding of the technology. The disclosed method is not limited to the operations and logical flows presented herein. Alternative embodiments are contemplated in which the order of various operations is changed and in which sub-operations described as part of larger operations are independently executed.

[0224] Furthermore, although the present application is described in the context of functional modules, it is to be understood that one or more of the functions and / or features described above can be integrated in a single physical device and / or software module, or one or more functions and / or features can be implemented in separate physical devices or software modules. It will also be appreciated that detailed discussion of the actual implementation of each module is unnecessary to an understanding of the present application. Rather, the actual implementation of the modules, in conjunction with their attributes, functions, and internal relationships, are to be understood within the context of the devices disclosed herein. Thus, those skilled in the art with access to the teachings presented herein will be able to devise suitable implementations of the present application without undue experimentation. It is also to be understood that the particular concepts disclosed are merely illustrative and are not intended to limit the scope of the present application, which is defined by the appended claims and equivalents thereof.

[0225] If the above functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or the part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0226] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a list of executable instructions for implementing logic functions, which can be specifically embodied in any computer readable medium for use by an instruction execution system, device or apparatus, such as a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, device or apparatus, or in conjunction with these instructions execution system, device or apparatus. For the purpose of this specification, "computer readable medium" can be any device that can contain, store, communicate, propagate or transport programs for use by an instruction execution system, device or apparatus, or in conjunction with these instruction execution system, device or apparatus.

[0227] More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection (electronic) having one or more wires, a portable computer diskette (magnetic), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can also be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for example, via optical scanning of the paper or other medium, then compiled, interpreted, or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.

[0228] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above described embodiments, various steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any of the following technologies, known in the art, or combinations thereof, can be used: a discrete logic circuit having logic gates for implementing logic functions upon data signals, an application specific integrated circuit having appropriate combinational logic gates, a programmable gate array (PGA), a field programmable gate array (FPGA), and / or the like.

[0229] In the above description of the present specification, the description referring to the terms "one embodiment", "another embodiment", or "certain embodiments" or the like means that a specific feature, structure, material or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present application. The illustrative expressions do not necessarily refer to the same embodiment or example throughout the present specification. Also, the specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0230] Although embodiments of the present application have been shown and described, it would be recognized by those of ordinary skill in the art that various changes, modifications, alternatives, and variations can be made to the embodiments without departing from the spirit and scope of the application, which is defined by the claims and their equivalents.

[0231] The above is a specific description of the preferred embodiments of the present application, but the present application is not limited to the above-described embodiments, and those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present application, and these equivalent modifications or substitutions are included in the scope defined by the claims of the present application.

Claims

1. A data mining-based tobacco product brand packaging and promotion strategy determination method, characterized by, The method comprises the following steps: According to the industry association announcement and news, the negative evaluation information of the tobacco industry is crawled, the key information related to the brand packaging requirements is extracted according to the negative evaluation information, and the first influence factor of the tobacco product brand packaging is analyzed according to the key information; The tobacco tax and tobacco control regulations are analyzed, and the second influence factor of the change of tobacco tax and industry requirements on tobacco product brand packaging and promotion strategy is predicted; Using public opinion analysis tools, according to the discussion on social media platforms and professional health websites, the acceptance degree information of consumers to tobacco product brand packaging and health warning is analyzed; According to the online feedback and comments of consumers on tobacco product brand packaging, sentiment analysis technology is used to mine the preference degree information of consumers to different brand packaging design elements; According to the market sales data, the demand of consumers for different quality and type of brand packaging is analyzed to determine the diversification direction information of tobacco product brand packaging; The brand packaging design data of competitors is crawled, and convolutional neural network is used to identify brand packaging images to analyze market brand packaging trends and competitive situation; According to the advertising activities, tobacco product brand packaging design and sales data of tobacco products, the relationship curve between sales performance and different brand packaging and different promotion strategies is determined; According to the promotion effect and market feedback of tobacco product brand packaging, multiple regression analysis and prediction model are used to evaluate the third influence factor of tobacco product brand packaging and promotion strategy on market sales; According to the acceptance degree information, the preference degree information, the diversification direction information, the market brand packaging trend and competitive situation, the relationship curve, the first influence factor, the second influence factor and the third influence factor, the target brand packaging and the target promotion strategy of the tobacco product are determined; The analysis of tobacco tax and tobacco control regulations to predict the second influence factor of the change of tobacco tax and industry requirements on tobacco product brand packaging and promotion strategy specifically includes: Crawl the attribute data of product brand packaging specifications, brand packaging design features, printing information, brand packaging materials, promotion channels and promotion content; According to the second text data of tobacco tax and tobacco control regulations, the influence of tobacco product brand packaging and promotion strategy is taken as the result label; Using random forest algorithm, a prediction model of the influence of tax regulation change on brand packaging promotion is constructed to determine the changes of packaging specifications, design features, materials, promotion channels and promotion content of tobacco product brand packaging under different tobacco tax and tobacco control regulations; According to the prediction results, the second influence factor of tobacco tax and tobacco control regulations on tobacco product brand packaging and promotion strategy is determined; The analysis of market sales data to analyze the demand of consumers for different quality and type of brand packaging to determine the diversification direction information of tobacco product brand packaging specifically includes: Obtain the purchase history data, preference data and evaluation data of consumers purchasing tobacco products; According to the purchase history data and preference data of consumers, the demand of consumers for different quality and type of brand packaging is obtained; According to the tobacco products purchased by consumers and the corresponding brand packaging characteristics, the brand packaging requirements of consumers for the characteristics and brand image of tobacco products are obtained; According to the purchase history data and evaluation data of different consumers, the collaborative filtering algorithm is used to analyze the preference and correlation of consumers for different tobacco product brand packaging attributes, and the preference degree and requirement of consumers for the safety, sustainability, visual appeal and convenience brand packaging attributes are determined; According to the preference and correlation of consumers for each tobacco product brand packaging attribute, the collaborative filtering algorithm is used to analyze the tobacco product brand packaging attributes that consumers pay most attention to, and the diversification direction information of the tobacco product brand packaging is determined.

2. The data mining based tobacco product brand packaging and promotion strategy determination method according to claim 1, characterized in that, According to the negative evaluation information of the tobacco industry crawled from the announcements and news of industry associations, the key information related to brand packaging requirements is extracted according to the negative evaluation information, and the first influence factor of the tobacco product brand packaging is analyzed according to the key information, which specifically includes: The first text data of industry association announcements and news reports is crawled, and the word sequence is obtained by denoising, word segmentation and removing stop words on the first text data; The first entity about tobacco products and tobacco enterprises in the word sequence is recognized by using named entity recognition algorithm; According to the first entity, the key information related to industry regulation changes and brand packaging requirements is extracted from the first text data by using keyword extraction or topic model; The importance and frequency of each word are determined by word frequency analysis on the key information; According to the sorting and position of the words in the text, the relevance and context of each word are determined; The proportion of positive, negative and neutral sentiment is obtained by sentiment analysis on the key information, and the emotional attitude and emotional change are determined; The theme distribution of the key information is judged to determine the theme words and theme proportion, and the association graph of the theme is constructed to analyze the association relationship between the themes; The specific content of the industry association announcement change and the first influence factor of the tobacco product brand packaging are determined.

3. The data mining based tobacco product brand packaging and promotion strategy determination method according to claim 1, characterized in that, According to the discussion on social media platforms and professional health websites, the acceptance degree information of consumers for tobacco product brand packaging and health warning is analyzed by using public opinion analysis tool, which specifically includes: The comment text on social media platforms and professional health websites is crawled, and the evaluation text of consumers on the appearance design, information transmission effect and health warning acceptance degree of tobacco product brand packaging is collected; The keywords related to the attractiveness of tobacco product brand packaging, the consistency of brand image, and the consistency of design style and personal aesthetics are extracted by using TF-IDF algorithm; The acceptance degree information of consumers for different brand packaging and health warning is determined by using TextRank algorithm; The influence degree of health warning on smoking behavior and the attention degree of consumers on their own health are analyzed by using sentiment analysis algorithm; The keyword frequency of consumers' comments on the environmental friendliness of tobacco product brand packaging is counted, and the recognition degree information of consumers on the environmental friendliness of brand packaging materials and the acceptance degree information of consumers on the environment-friendly brand packaging are analyzed.

4. The data mining based tobacco product brand packaging and promotion strategy determination method according to claim 1, characterized in that, According to the online feedback and comments of consumers on tobacco product brand packaging, the emotional analysis technology is used to mine the preference degree information of consumers on different brand packaging design elements, which specifically includes: According to the online feedback and comments of consumers, data containing evaluation and emotional expression of brand packaging color are obtained; Using emotional analysis algorithm, the preference degree and emotional response of consumers to different colors are analyzed to determine the preference degree information of consumers for different color packaging design; According to the online feedback and comments of consumers, data containing evaluation and emotional expression of brand packaging image and pattern are obtained; Using emotional analysis algorithm, the preference degree and emotional response of consumers to different images and patterns are analyzed to determine the preference degree information of consumers for different image and pattern packaging design; According to the online feedback and comments of consumers, data containing evaluation and emotional expression of brand packaging font and layout are extracted; Using emotional analysis algorithm, the preference degree and emotional response of consumers to different fonts and layouts are analyzed to determine the preference degree information of consumers for different font and layout packaging design; According to the online feedback and comments of consumers, data containing evaluation and emotional expression of brand packaging material and texture are extracted; Using emotional analysis algorithm, the preference degree and emotional response of consumers to different materials and textures are analyzed to determine the preference degree information of consumers for different material and texture packaging design.

5. The data mining based tobacco product brand packaging and promotion strategy determination method according to claim 1, characterized in that, The brand packaging design data of competitors is crawled, and the convolutional neural network is used to identify brand packaging images to analyze market brand packaging trends and competitive situation, which specifically includes: The brand packaging design image dataset of competitors is obtained and labeled to obtain label data containing brand packaging color, shape, pattern and text; According to the obtained brand packaging design image dataset and label data, a convolutional neural network algorithm is used to build a brand packaging image recognition model to determine the brand packaging color, shape, pattern and text attributes of new brand packaging images; The brand packaging design images of competitors are divided into several clusters, each cluster representing a brand packaging trend or competitive situation, and a K-means clustering algorithm is used to build a brand packaging feature clustering analysis model; Based on the feature attributes of the clusters, the brand packaging design trends in the market and the brand packaging strategies of competitors are determined.

6. The data mining based tobacco product brand packaging and promotion strategy determination method according to claim 1, characterized in that, According to the advertising activities, tobacco product brand packaging design and sales data of tobacco products, the relationship curve between sales performance and different brand packaging and different promotion strategies is determined, which specifically includes: Obtain sales data and brand packaging type information, including brand packaging type, sales performance and brand; Through data preprocessing, sales data and brand packaging type are encoded for association rule mining; Frequent item sets of different brand packaging types and sales performance are constructed, which represent the association relationship between different brand packaging and sales performance; Association rules are constructed, and association rules are extracted from frequent item sets according to the threshold of confidence; According to the support and confidence, the importance of association rules is evaluated, and the association rules with practical significance are screened out; According to the obtained association rules, the relationship curve of sales performance and different brand packaging and different promotion strategies is determined.

7. The data mining based tobacco product brand packaging and promotion strategy determination method according to claim 1, characterized in that, According to the promotion effect and market feedback of the tobacco product brand packaging, a multiple regression analysis and a prediction model are used to evaluate the third influence factor of the tobacco product brand packaging and the promotion strategy on the market sales, which specifically includes: Obtain the data related to the design and color of the tobacco product brand packaging, including the sales data, market research and consumer feedback comment text of different design and color schemes; Using a multiple regression analysis method, the first influence coefficient of brand packaging design and color on market sales is determined; Obtain the relevant data of different brand packaging materials and quality, including consumer evaluation of brand packaging materials and quality and product protection performance; Using a multiple regression analysis method, the second influence coefficient of brand packaging materials and quality on market sales is determined; Obtain the relevant data of different promotion channels and promotion content, including the exposure of different channels and content and the response of consumers to the promotion content; Using a multiple regression analysis method, the third influence coefficient of promotion channels and promotion content on market sales is determined; Obtain the reaction data of the target audience to the brand packaging and promotion strategy, including the purchase willingness and brand awareness of consumers; Using a multiple regression analysis method, the fourth influence coefficient of the target audience's reaction on market sales is determined; According to the first influence coefficient, the second influence coefficient, the third influence coefficient and the fourth influence coefficient, the third influence factor of the tobacco product brand packaging and the promotion strategy on the market sales is determined.

8. A data mining based tobacco product brand packaging and promotion strategy determination system characterized by, A data mining-based tobacco product brand packaging and promotion strategy determination method as claimed in any one of claims 1 to 7 is implemented, comprising: A first influence factor determination module is configured to crawl negative evaluation information of the cigarette industry according to industry association announcements and news, extract key information related to brand packaging requirements from the negative evaluation information, and analyze the first influence factor of the tobacco product brand packaging according to the key information; A second influence factor determination module is configured to analyze tobacco tax and tobacco control regulations, and predict the second influence factor of tax and industry requirement changes on the tobacco product brand packaging and promotion strategy; An acceptance information determination module is configured to use public opinion analysis tools to analyze the acceptance information of consumers to the tobacco product brand packaging and health warnings based on discussions on social media platforms and professional health websites; A preference information determination module is configured to use sentiment analysis technology to mine the preference information of consumers for different brand packaging design elements based on online feedback and comments of consumers on the tobacco product brand packaging; A diversification direction information determination module is configured to analyze the demand of consumers for different quality and type of brand packaging based on market sales data, and determine the diversification direction information of the tobacco product brand packaging; A market brand packaging trend and competitive situation determination module is configured to crawl the brand packaging design data of competitors, use convolutional neural networks to recognize brand packaging images, and analyze the market brand packaging trend and competitive situation; a relationship curve determining module, configured to determine a relationship curve between sales performance and different brand packaging and different promotion strategies according to advertising activities, tobacco product brand packaging designs, and sales data of the tobacco products; a third influence factor determining module, configured to evaluate a third influence factor of the tobacco product brand packaging and the promotion strategies on market sales by using multiple regression analysis and a prediction model according to promotion effects of the tobacco product brand packaging and market feedback; a brand packaging and promotion strategy determining module, configured to determine target brand packaging and target promotion strategies of the tobacco products according to the acceptance degree information, the preference degree information, the diversification direction information, the market brand packaging trends and competitive situations, the relationship curve, the first influence factor, the second influence factor, and the third influence factor.

Images