Cigarette sensory smoker tendency mining and characterization method based on association rules

Abstract

The invention discloses a cigarette sensory smoker tendency mining and characterization method based on association rules. The method mainly comprises the following steps: collecting and preprocessing cigarette sensory evaluation data; carrying out tendency quantitative identification by utilizing a preset evaluation criterion; establishing a tendency association rule base of the smoke panel test personnel based on the identification result; utilizing the association rule base to mine tendency association relationships of smoke panel test personnel on different types and brands of cigarettes and evaluation indexes; according to the tendency incidence relation, a smoker tendency characterization model is constructed, and the model is used for predicting the individual evaluation tendency of the smoker in the evaluation process of the unknown cigarette sample; and verifying and evaluating the smoke panel tendency characterization model by using a preset cigarette smoke panel sample set. According to the method, the objectivity and accuracy of the evaluation result can be remarkably improved, the product research and development direction can be guided, and the product formula and process can be optimized.

Description

Technical Field

[0001] The invention relates to the field of cigarette manufacturing, and in particular to a method for mining and characterizing the tendency of cigarette sensory evaluators based on association rules. Background Art

[0002] The sensory quality of cigarettes is an important basis for consumers to choose, and sensory evaluation is a key link in evaluating the quality of cigarettes. The accuracy and reliability of the results directly affect the market competitiveness of the product. Sensory evaluation is a method of evaluating the aroma, impurities, irritation, aftertaste and other characteristics of cigarettes through the visual, olfactory and taste sensory perceptions of the evaluators. During the evaluation process, the evaluators will score and describe the various sensory characteristics of cigarettes based on their own sensory experience and professional knowledge. However, sensory evaluation mostly relies on the subjective judgment of the evaluators, and has high requirements for sensitivity, accuracy and reproducibility. It has certain limitations and is easily affected by adverse physiological conditions, wrong guidance, abnormal psychology, etc., making it difficult to objectively and scientifically describe and evaluate the sensory characteristics of cigarettes.

[0003] When evaluating cigarettes, if the reviewers have personal tendencies, such as a preference for a certain type of tobacco leaf, cigarette brand or specific flavor, this subjective bias may unconsciously penetrate into the evaluation process, causing the evaluation results to deviate from objective facts. For example, reviewers may over-evaluate the aroma and taste of a certain brand of cigarettes, while ignoring its possible defects or deficiencies. In addition, the reviewers' tendencies are also reflected in the inconsistency of evaluation standards. Different reviewers have different understandings and grasps of the evaluation standards for cigarettes based on their respective experiences, preferences and cognitions. This difference in evaluation standards may not only lead to very different evaluation results for the same cigarette among different reviewers, but may also affect fair competition and consumer choice in the cigarette market.

[0004] The existing related technologies are mainly based on the basic principles of sensory evaluation, combined with simple comparative analysis and statistical methods, to quantify the differences in sensory responses of the smokers to reveal their tendency characteristics. The key technical point is that each smoker first quantifies the sensory characteristics of the cigarettes, for example, using a 0-9 point system or a percentage system, and then compares and analyzes the smoker's scoring results with the standard score or the scores of other smokers to obtain the score differences and tendency characteristics between the smokers.

[0005] As mentioned above, the existing methods have many disadvantages:

[0006] (1) Highly subjective

[0007] In the implementation of comparative analysis, whether it is the selection of comparative objects, the dimensions of comparison or the final interpretation of conclusions, it is easily influenced by subjective thinking. This subjectivity may cause the analysis results to deviate from objective facts and affect the accuracy and reliability of the conclusions.

[0008] (2) Limited scope of application

[0009] Comparative analysis is usually applicable to relatively simple data relationships or scenarios. It can only compare single features horizontally, and it is difficult to fully capture the complex relationship between evaluators and feature indicators. In addition, comparative analysis relies more on the subjective judgment and assumptions of researchers, and is insufficient in mining hidden relationships and potential patterns between data.

[0010] (3) The prediction is not reliable

[0011] Comparative analysis has certain limitations in terms of prediction. It is often based on researchers’ understanding and assumptions about data relationships, and these assumptions may not be completely accurate or reliable. Therefore, the prediction results of comparative analysis may be subject to the risk of overestimation or underestimation bias, resulting in low credibility of the prediction.

[0012] (4) Low efficiency

[0013] When dealing with large-scale data sets, comparative analysis may require researchers to perform a lot of manual operations and data comparison work, resulting in inefficient analysis process.

[0014] In summary, although comparative analysis has certain application value in the characterization of sensory tendencies of tasters, its shortcomings, such as strong subjectivity, limited application scope, low prediction credibility, difficulty in exploring hidden relationships, and low efficiency, limit its breadth and depth in practical applications. Summary of the invention

[0015] In view of the above, the present invention aims to provide a method for mining and characterizing the tendency of cigarette sensory evaluators based on association rules to solve the above-mentioned technical problems.

[0016] The technical solution adopted by the present invention is as follows:

[0017] The present invention provides a method for mining and characterizing the tendency of cigarette sensory evaluators based on association rules, which includes:

[0018] Collect and preprocess cigarette sensory evaluation data;

[0019] Use the preset evaluation benchmark to quantitatively identify the tendency;

[0020] Establish the tendency association rule base of the assessors based on the identification results;

[0021] By using the association rule library, the tendency association relationship of the smokers on different types and brands of cigarettes and evaluation indicators is mined;

[0022] According to the tendency association relationship, a smoking reviewer tendency representation model is constructed, wherein the model is used to predict the individual evaluation tendency of the smoking reviewers in the process of evaluating unknown cigarette samples;

[0023] The cigarette smoking tester tendency characterization model is verified and evaluated using a preset cigarette smoking test sample set.

[0024] In at least one possible implementation, the collecting and preprocessing of cigarette sensory evaluation data includes:

[0025] Cigarette samples of different types and brands are used as evaluation objects. Multiple smokers conduct sensory evaluation on the cigarette samples in advance and record the evaluation data of each smoker on each indicator.

[0026] Clean and standardize the collected evaluation data;

[0027] In addition, the cigarette type, cigarette brand, evaluator, and evaluation index of the evaluation data are coded.

[0028] In at least one possible implementation, the quantitative identification of the tendency using a preset evaluation benchmark includes:

[0029] The evaluation results of pre-formed benchmarking experts are used as the evaluation benchmark for cigarette samples;

[0030] The evaluation data of each evaluator is compared with the evaluation benchmark to obtain a quantitative expression of the evaluator's tendency towards each evaluation indicator.

[0031] In at least one possible implementation, the method of constructing the tendency association rule library includes: extracting all frequent item sets in the preset tendency transaction database of the smoking reviewers, and forming association rules for the smoking reviewers on different types and brands of cigarettes and evaluation indicators.

[0032] In at least one possible implementation, the method of mining the tendency association relationship includes:

[0033] Based on the association rule library of the smokers, the lift of each association rule data is calculated, and the association rules with lift greater than the given threshold are screened out to obtain the strong correlation between the smokers' preferences for different types and brands of cigarettes and evaluation indicators.

[0034] In at least one possible implementation manner, the cigarette sensory evaluation data includes at least one or more of the following combinations: gloss, aroma, harmony, miscellaneous smell, irritation, aftertaste, and strength.

[0035] Compared with the prior art, the main design concept of the present invention is to collect sensory evaluation data of smokers and select experienced and sensitive smokers as a benchmark expert group, use their evaluation results as a comparison benchmark, quantify and identify the sample tendencies of all smokers; then, use the association rule method to construct an association rule library and a tendency characterization model for smokers, explore the intrinsic relationship between smokers of different types and brands of cigarettes and evaluation indicators, and characterize personal tendencies in the evaluation process, so as to improve the objectivity and accuracy of the sensory evaluation of cigarettes.

[0036] The present invention can significantly improve the objectivity and accuracy of the evaluation results. The main advantages are as follows:

[0037] First, reduce subjective interference: This technology reduces the influence of individual subjective factors on the evaluation results by quantifying the sensory experience of the evaluator. During the evaluation process, the evaluator may be influenced by factors such as personal preferences and emotional state, and this technology can weaken these interference factors to a certain extent, making the evaluation results more objective.

[0038] Second, improve evaluation consistency: by characterizing the tendency of the reviewers, the evaluation results of different reviewers on the same cigarette product can be made closer, thus improving the consistency of the evaluation. This helps companies to understand the true sensory quality of the product more accurately.

[0039] The present invention can also guide the direction of product research and development, optimize product formula and process, and its main advantages are as follows:

[0040] First, by collecting and analyzing a large amount of sensory evaluation data, companies can understand consumers' preferences and needs for cigarette products, and then guide the direction of product development. For example, based on consumers' feedback on aroma, taste, etc., companies can adjust product formulas and processes to meet market demand.

[0041] Second, the technology can also be used to evaluate the impact of different formulas and processes on the sensory quality of cigarette products. By comparing the sensory evaluation data of products under different formulas and processes, companies can screen out the optimal formula and process combination to improve product quality.

[0042] The present invention can also enhance market competitiveness, and its main advantages are as follows:

[0043] First, improve product quality: By continuously optimizing product formulas and processes, companies can produce cigarette products that better meet consumer demand. This helps improve product market competitiveness and attract more consumers to buy.

[0044] Second, shaping the brand image: high-quality cigarette products can win the trust and love of consumers, and then shape a good brand image, which is of great significance for enterprises to stand out in the fiercely competitive market.

[0045] The present invention can also promote the technological progress of the industry. The main advantages are as follows:

[0046] First, promote technological innovation: The development and application of cigarette sensory evaluation tendency characterization technology has promoted technological innovation in the field of sensory evaluation in the tobacco industry. By continuously exploring new evaluation methods and technical means, we can have a deeper understanding of the inherent characteristics and changing laws of cigarette products.

[0047] Second, improve the overall level of the industry: With the popularization and application of this technology, the overall level of the tobacco industry will also be improved. Enterprises will pay more attention to the sensory quality of products, strengthen technological research and development and innovation, and promote the development of the entire industry to a higher level. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be further described below with reference to the accompanying drawings, in which:

[0049] Figure 1 A schematic diagram of a method for mining and characterizing the tendency of cigarette sensory evaluators based on association rules provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0050] Embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention.

[0051] The present invention proposes an embodiment of a method for mining and characterizing the tendency of cigarette sensory evaluators based on association rules. Specifically, Figure 1 shown, including:

[0052] Step S1, collecting and preprocessing cigarette sensory evaluation data;

[0053] For different types and brands of cigarette samples as evaluation objects, multiple tasters can be organized in advance to conduct sensory evaluation of the cigarette samples, and the evaluation data of each taster on indicators such as aroma, odor, irritation, and aftertaste can be recorded.

[0054] The cigarette sensory evaluation data may specifically include at least one or more of the following combinations:

[0055] 1) Gloss: refers to the brightness of the surface of tobacco or cigarettes. Oily tobacco usually has better gloss. The brightness of the gloss can reflect the smoking quality of the cigarette.

[0056] 2) Aroma: refers to the unique aroma of tobacco inherent in cigarette smoke, and is one of the main indicators for evaluating cigarette quality. Aroma has both qualitative and quantitative meanings. High-quality cigarettes usually have a strong, pure, and lasting aroma.

[0057] 3) Harmony: refers to the harmony among the various components of cigarette smoke, and the characteristics of any single monomer cannot be felt.

[0058] 4) Miscellaneous smell: refers to the slight or obvious bad smell that is not the smell of tobacco itself, such as grass smell, burnt smell, earthy smell, etc.

[0059] 5) Irritation: refers to the discomfort caused by smoke to the senses, such as the rush of smoke to the nasal cavity, the impact on the oral cavity, and the tingling and burning feeling in the throat.

[0060] 6) Aftertaste: refers to the taste sensation left after the smoke is exhaled from the mouth and nose. High-quality cigarettes usually have a clean, comfortable and long-lasting aftertaste, which can leave a deep impression on consumers.

[0061] 7) Strength: The strength of a cigarette is related to the amount of tar, nicotine in the smoke and other chemical components of the cigarette. Different consumers have different preferences for strength.

[0062] The preprocessing of the collected evaluation data may include data cleaning, removal of outliers and missing values, and standardization to ensure data consistency and comparability. This may also include coding the evaluation data, such as coding cigarette type as Type, cigarette brand as Brand, assessor as Assessor, and evaluation index as Index.

[0063] Step S2: using the evaluation data of the preset benchmarking expert group as a benchmark to perform quantitative identification of the tendency;

[0064] In order to avoid subjective interference to the smokers, several experienced and sensitive smokers can be screened and determined from all smokers in advance to form a benchmark expert group, and their evaluation results can be used as the closest to the true score of the cigarette samples. By comparing the evaluation results of each smoker and the standard expert group, the tendency of the sample data of the smokers on various evaluation indicators can be quantified. Among them, a high tendency can be marked as "H", a low tendency can be marked as "L", and no obvious tendency can be marked as "M".

[0065] Step S3: establishing a tendency association rule base of the assessors based on the identification results;

[0066] Specifically, the association rule base of the smokers can be constructed by using, but not limited to, the Apriori algorithm or the FP-Growth algorithm: extract all frequent item sets in the preset smokers' tendency transaction database, and form the association rules of the smokers on different types and brands of cigarettes and evaluation indicators. To expand:

[0067] (1) Extract all frequent itemsets in the transaction database: Assume It is the tendency transaction database of the assessors. The elements of are called items; the set of items is called an itemset. The support of an itemset A is defined as the proportion of transactions containing A in the transaction database D. The support of itemset A is not less than the preset minimum support threshold, that is, , then the item set A is called a frequent item set. The calculation formula of support is as follows:

[0068] (2) Form the association rules of the reviewers based on the frequent item sets. The association rules can be expressed as a logical implication A=>B, where A and B are two different non-empty sub-item sets of D. For example: {type= , Brand = , indicator = aroma, evaluator = }=>{tendency=H}.

[0069] Step S4, using the association rule library, mining the strong correlation between the inclinations of the reviewers on different types and brands of cigarettes and evaluation indicators;

[0070] During implementation, the lift of each association rule data can be calculated based on the association rule library of the smokers, and the association rules with a lift greater than 1 can be selected to explore the strong correlation between the smokers' preferences for different types and brands of cigarettes and evaluation indicators. The lift is defined as the ratio of the proportion of transactions in the transaction database D that contain item set A and also contain item set B to the proportion of transactions that contain item set B. The lift calculation formula can be referred to as follows:

[0071] Step S5, constructing a smoking reviewer tendency representation model based on the tendency strong correlation result, wherein the model is used to predict the individual evaluation tendency of the smoking reviewers in the process of evaluating unknown cigarette samples;

[0072] Step S6: using a preset cigarette smoking evaluation sample set to verify and evaluate the smoking evaluation personnel's tendency characterization model.

[0073] After verification, the accurate and reliable model is deployed in sensory evaluation practice. By adjusting the evaluation process and scoring criteria, the impact of the personal tendencies of evaluators on the evaluation results can be significantly reduced.

[0074] In summary, the main design concept of the present invention is to collect sensory evaluation data of smokers and select experienced and sensitive smokers as the benchmark expert group, use their evaluation results as the comparison benchmark, quantify and identify the sample tendencies of all smokers; then, use the association rule method to build the association rule library and tendency characterization model of smokers, explore the internal relationship between smokers in different types and brands of cigarettes and evaluation indicators, and characterize the personal tendencies in the evaluation process, so as to improve the objectivity and accuracy of the sensory evaluation of cigarettes. The present invention can significantly improve the objectivity and accuracy of the evaluation results, and can guide the direction of product research and development, and optimize product formulas and processes.

[0075] If the expressions expressing orientation are mentioned in the embodiments of the present invention, they are relative concepts based on the embodiments. In addition, "at least one" means one or more, and "plurality" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Among them, A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, a and b, a and c, b and c, or a and b and c, wherein a, b, c can be single or multiple.

[0076] The above describes in detail the structure, features and effects of the present invention based on the embodiments shown in the drawings, but the above is only a preferred embodiment of the present invention. It should be noted that the technical features involved in the above embodiments and their preferred methods can be reasonably combined and matched into a variety of equivalent schemes by those skilled in the art without departing from or changing the design ideas and technical effects of the present invention; therefore, the present invention is not limited to the scope of implementation shown in the drawings, and all changes made in accordance with the concept of the present invention, or modifications to equivalent embodiments with equivalent changes, which still do not exceed the spirit covered by the specification and drawings, should be within the protection scope of the present invention.

Claims

1. A method for mining and characterizing the tendency of cigarette sensory evaluators based on association rules, characterized in that: include: Collect and preprocess cigarette sensory evaluation data; Use the preset evaluation benchmark to quantitatively identify the tendency; Establish the tendency association rule base of the assessors based on the identification results; By using the association rule library, the tendency association relationship of the smokers on different types and brands of cigarettes and evaluation indicators is mined; According to the tendency association relationship, a smoking reviewer tendency representation model is constructed, wherein the model is used to predict the individual evaluation tendency of the smoking reviewers in the process of evaluating unknown cigarette samples; The cigarette smoking tester tendency characterization model is verified and evaluated using a preset cigarette smoking test sample set.

2. The method for mining and characterizing the tendency of cigarette sensory evaluators based on association rules according to claim 1 is characterized in that: The collecting and preprocessing of cigarette sensory evaluation data includes: Cigarette samples of different types and brands are used as evaluation objects. Multiple smokers conduct sensory evaluation on the cigarette samples in advance and record the evaluation data of each smoker on each indicator. Clean and standardize the collected evaluation data; In addition, the cigarette type, cigarette brand, evaluator, and evaluation index of the evaluation data are coded.

3. The method for mining and characterizing the tendency of cigarette sensory evaluators based on association rules according to claim 1, characterized in that: The method of using a preset evaluation benchmark to quantitatively identify the tendency includes: The evaluation results of pre-formed benchmarking experts are used as the evaluation benchmark for cigarette samples; The evaluation data of each evaluator is compared with the evaluation benchmark to obtain a quantitative expression of the evaluator's tendency towards each evaluation indicator.

4. The method for mining and characterizing the tendency of cigarette sensory evaluators based on association rules according to claim 1, characterized in that: The method of constructing the tendency association rule library includes: extracting all frequent item sets in the preset tendency transaction database of the smoking reviewers, and forming association rules of the smoking reviewers on different types and brands of cigarettes and evaluation indicators.

5. The method for mining and characterizing the tendency of cigarette sensory evaluators based on association rules according to claim 1, characterized in that: Ways to mine tendency associations include: Based on the association rule library of the smokers, the lift of each association rule data is calculated, and the association rules with lift greater than the given threshold are screened out to obtain the strong correlation between the smokers' preferences for different types and brands of cigarettes and evaluation indicators.

6. The method for mining and characterizing the tendency of cigarette sensory evaluators based on association rules according to any one of claims 1 to 5, characterized in that: The cigarette sensory evaluation data includes at least one or more of the following combinations: gloss, aroma, harmony, miscellaneous smell, irritation, aftertaste, and strength.

Images