AI tea recommendation intelligent system based on large language model

Abstract

The present invention relates to the technical field of intelligent product recommendation, and more specifically, to an AI tea recommendation intelligent system based on a large language model. It includes a big data intelligent platform, a product information interaction platform, a multi-product recommendation module, and a selection probability calculation module. The present invention obtains tea data information in the industry through a big data intelligent platform, formulates an allocation mechanism, classifies and processes the collected tea products according to the allocation mechanism, automatically classifies and processes the newly entered tea data, and re-plans tea recommendations through a multi-product recommendation module combined with a recommended tea product recommendation rate. It also re-weights recommendations based on the recommended tea product recommendation rate fed back by the selection probability calculation module, thereby improving recommendation adaptability and user purchase rate. At the same time, it can increase user trust in recommended tea products, provide sales channels for new types of tea in the future, match and adapt to user needs, and avoid limited sales channels due to low popularity of new types of tea.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent product recommendation, and more specifically, to an AI tea recommendation intelligent system based on a large language model. Background Art

[0002] Tea is the main source of daily beverages, and its types are distributed in various regions. Different regions have their own unique insights into the quality of tea. Buying tea is not an easy task. To get good tea, you need to master a lot of knowledge, such as the grade standards, prices and market conditions of various types of tea, as well as tea evaluation and inspection methods. The quality of tea is mainly identified from four aspects: color, aroma, taste, and shape. However, for ordinary tea drinkers, when buying tea, they can generally only look at the appearance and color of the dry tea and smell the dry aroma, making it even more difficult to judge the quality of the tea.

[0003] Generally, tea quality can be identified through sensory evaluation. This involves using the human senses of sight, smell, taste, and taste to grasp the inherent characteristics of tea. Finally, the quality of tea is comprehensively judged through observation, smelling, touching, and tasting. People in some regions prefer mellow tea, while others prefer light fragrance. Therefore, different groups have different opinions on the quality of tea. With the rise of e-commerce, online tea sales have become increasingly popular. However, due to the wide variety of teas available online, people in different regions have different standards for judging tea quality and different factors in choosing tea. As a result, when selecting tea online, most users prefer to buy familiar tea varieties. If the advantages of each tea and the factors driving user purchases cannot be broken down and analyzed, sellers will be unable to determine the specific tea habits of current users. Consequently, most recommended teas fail to meet user needs, preventing new tea varieties from adapting to the market and matching them with users with corresponding needs.

[0004] In order to address the above problems, there is an urgent need for an AI tea recommendation intelligent system based on a large language model. Summary of the Invention

[0005] The purpose of the present invention is to provide an AI tea recommendation intelligent system based on a large language model to solve the problems raised in the above background technology.

[0006] To achieve the above objectives, an AI tea recommendation intelligent system based on a large language model is provided, which includes a big data intelligent platform, a product information interaction platform, a multi-style product recommendation module, and a selection probability calculation module.

[0007] The big data intelligent platform is used to obtain tea product data information within the industry, formulate an allocation mechanism, classify the collected tea products according to the allocation mechanism, formulate allocation mechanism weights, build a neural learning network, and automatically classify the newly entered tea product data;

[0008] The product information interaction platform is used to build a tea product data information interaction platform, through which users' tea product data information is shared in real time, and the users' tea-using habits are obtained by sharing the users' tea product data information;

[0009] The diversified product recommendation module formulates diversified tea recommendation patterns based on the user's tea-drinking habits and recommends teas of corresponding patterns to the user;

[0010] The selection probability calculation module collects user purchase data of recommended tea products, calculates the recommendation rate of recommended tea products, and feeds the recommendation rate of recommended tea products back to the diversified product recommendation module. The diversified product recommendation module re-plans tea product recommendations based on the recommendation rate of recommended tea products.

[0011] As a further improvement of this technical solution, the big data intelligent platform includes an underlying logic data acquisition module, an allocation mechanism definition module, and a neural network construction module;

[0012] The underlying logic data acquisition module is used to obtain tea product data information within the industry;

[0013] The output end of the bottom logic data acquisition module is connected to the input end of the allocation mechanism definition module, and the allocation mechanism definition module formulates tea classification rules based on the tea product data information in the industry;

[0014] The output end of the allocation mechanism definition module is connected to the input end of the neural network construction module. The neural network construction module is used to compare the various evaluation criteria of different types of tea, divide the weights of different factors of different types of tea, establish a tea classification standard database, compare the various factors of new tea, and provide a reference standard for the price of new tea.

[0015] As a further improvement of the present technical solution, the method for providing a new tea price reference standard by the neural network construction module includes the following steps:

[0016] S1. Compare various factors of new tea with the tea classification standard database;

[0017] S2. Combine the comparison results to obtain the matching price levels;

[0018] S3. Recommending new teas that are suitable for each price level and obtaining the recommendation rate for the corresponding price level;

[0019] S4. Arrange the tea leaves in order of their recommendation rates, and select the price grade with the highest recommendation rate as the recommended price for the new tea leaves.

[0020] As a further improvement of the present technical solution, the various factors for comparing the new tea leaves in S1 include the size, color and shape of the tea leaves.

[0021] As a further improvement of the present technical solution, the product information interaction platform includes a tea keyword extraction module and a chat content acquisition module. The tea keyword extraction module is used to extract tea keywords that users interact with in chats on the tea data information interaction platform. The output end of the tea keyword extraction module is connected to the input end of the chat content acquisition module. The chat content acquisition module combines the tea keywords in the chat interactions to determine the type of tea that the current user is talking about.

[0022] As a further improvement of this technical solution, the chat content acquisition module adopts a keyword comparison algorithm, and its algorithm formula is as follows:

[0023] ;

[0024] ;

[0025] ;

[0026] in The tea keyword set obtained from the current user's chat interaction. to Each tea keyword obtained from the current user's chat interaction, is the tea keyword set to be compared with it, to For each tea keyword to be compared, is the tea keyword comparison function, Keyword comparison overlap rate, It is the keyword matching overlap rate threshold.

[0027] As a further improvement of the present technical solution, the output end of the chat content acquisition module is connected to a tea data sharing module, and the tea data sharing module is used to share the tea type information mentioned by the user.

[0028] As a further improvement of the present technical solution, the method for the diversified product recommendation module to formulate a diversified tea recommendation model includes the following steps:

[0029] S10, obtaining the type of tea mentioned by the current user;

[0030] S20, combining with the tea classification standard database, determining various factors of the currently acquired tea;

[0031] S30. Based on the weights of the factors corresponding to the factors, various teas that match the weights of the factors are recommended as recommended teas, and a diversified tea recommendation model is established.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] In this AI tea recommendation intelligent system based on a large language model, tea data information in the industry is obtained through a big data intelligent platform, an allocation mechanism is formulated, the collected teas are classified and processed according to the allocation mechanism, and the newly entered tea data is automatically classified and processed. The tea recommendations are re-planned through a multi-style product recommendation module combined with the recommended tea recommendation rate, and the recommendation weight is re-drawn according to the recommended tea recommendation rate fed back by the selection probability calculation module, thereby improving the recommendation adaptability and user purchase rate. At the same time, it can increase users' trust in recommended teas, provide sales channels for new types of tea in the future, match and adapt to user needs, and avoid limited sales channels due to low popularity of new types of tea. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a flowchart of the overall structure of the present invention.

[0035] The meaning of each number in the figure is:

[0036] 10. Bottom-level logic data acquisition module;

[0037] 20. Allocation mechanism definition module;

[0038] 30. Neural network building blocks;

[0039] 40. Tea keyword extraction module;

[0040] 50. Chat content acquisition module;

[0041] 60. Tea data sharing module;

[0042] 70. Diverse product recommendation module;

[0043] 80. Select the probability calculation module. DETAILED DESCRIPTION

[0044] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0045] See also Figure 1As shown, an AI tea recommendation intelligent system based on a large language model is provided, including a big data intelligent platform, a product information interaction platform, a multi-style product recommendation module 70, and a selection probability calculation module 80;

[0046] The big data intelligent platform is used to obtain tea product data information within the industry, formulate an allocation mechanism, classify the collected tea products according to the allocation mechanism, formulate the allocation mechanism weights, build a neural learning network, and automatically classify the newly entered tea product data;

[0047] The product information interaction platform is used to build a tea product data information interaction platform, through which users' tea product data information is shared in real time, and users' tea-using habits are obtained through the shared tea product data information;

[0048] The multi-product recommendation module 70 formulates a multi-product recommendation model based on the user's tea-drinking habits and recommends tea products corresponding to the model to the user;

[0049] The selection probability calculation module 80 collects user purchase recommended tea data, calculates the recommended tea recommendation rate, and feeds the recommended tea recommendation rate back to the multi-style product recommendation module 70. The multi-style product recommendation module 70 re-plans the tea recommendation based on the recommended tea recommendation rate.

[0050] In specific use, during the tea recommendation process, first, the industry's tea data information is obtained through the big data intelligent platform, and an allocation mechanism is formulated. The collected teas are classified according to the allocation mechanism. For example, tea is classified according to the price gradient of tea. Different teas in the same price level may have different points. For example, the taste of tea may change due to different production processes. Therefore, tea types in the same price batch need to be divided according to the taste, color, and shape of the tea. A weight for the allocation mechanism is formulated, that is, a weight is assigned to each tea grade. A neural learning network is constructed to match the newly entered tea data to the grade and automatically classify them.

[0051] The user shares the user's tea data information in real time through the tea data information interaction platform, and obtains the user's tea-using habits by sharing the user's tea data information, such as obtaining the user's chat related keywords, tea pictures and tea knowledge popular science, to determine the type of tea currently being evaluated, and transmit the obtained tea type information to the multi-style product recommendation module 70. The multi-style product recommendation module 70 retrieves the classification data of the corresponding tea from the big data intelligent platform, obtains its tea grade distribution and corresponding weight, and recommends multi-style products according to the grade distribution and corresponding weight. For example, the tea priced at 1000 / jin includes four types of tea: A, B, C and D, while the light-fragrant tea includes tea A and Tea C, belonging to the mellow-flavored type includes Tea B and Tea D, belonging to the large-leaf type includes Tea A, Tea B and Tea C, and belonging to the small-leaf type includes Tea D. The user's tea is a large-leaf light-fragrant tea, and its price is similar to the above-mentioned tea types. It is impossible to determine whether the user's preference is due to the large-leaf type, the light-fragrant type, or both. In this case, Tea A and Tea C can be recommended to the user, and the order of recommendation is based on the weight of the types of Tea A and Tea C. For example, the light-fragrant quality of Tea A is higher than that of Tea C. At this time, in the light-fragrant type tea category, the light-fragrant quality of Tea A is higher than that of Tea C. Therefore, the recommendation order is to recommend Tea A first and then Tea C.

[0052] After completing the tea recommendation, the probability calculation module 80 is selected to collect the user's purchase data of the recommended tea, and the recommended tea recommendation rate is calculated, that is, the ratio of the number of recommended teas to the number of user purchases is calculated, and the recommended tea recommendation rate is fed back to the multi-style product recommendation module 70. The multi-style product recommendation module 70 re-plans the tea recommendation based on the recommended tea recommendation rate. For example, tea A and tea C are both light fragrance type, and the weight of tea A is higher than that of tea C in this type. At the same time, tea A and tea C are both large leaf type, and the weight of tea C is higher than that of tea A in this type. When the user is recommending tea products, under the same recommendation quantity, the purchase rate of tea A is much higher than that of tea C, indicating that the user prefers the fragrance of tea at the same price. At this time, the recommendation weight is re-planned when making the second recommendation, further improving the recommendation adaptability and the user purchase rate. At the same time, it can increase the user's trust in the recommended tea, provide sales channels for new types of tea in the future, match the needs of users, and avoid the limited sales channels due to the low popularity of new types of tea.

[0053] In addition, the big data intelligent platform includes an underlying logic data acquisition module 10, an allocation mechanism definition module 20, and a neural network construction module 30;

[0054] Among them, the underlying logic data acquisition module 10 is used to obtain tea product data information in the industry;

[0055] The output end of the underlying logic data acquisition module 10 is connected to the input end of the allocation mechanism definition module 20. The allocation mechanism definition module 20 combines the tea product data information in the industry to formulate tea classification rules;

[0056] The output end of the allocation mechanism definition module 20 is connected to the input end of the neural network construction module 30. The neural network construction module 30 is used to compare the various evaluation criteria of different types of tea, divide the weights of different factors of different types of tea, establish a tea classification standard database, compare the various factors of new tea, and provide a reference standard for the price of new tea.

[0057] First, the underlying logic data acquisition module 10 acquires industry-wide tea product data, namely, the production process, shape, size, color, and flavor of tea. The acquired tea product data is then transmitted to the allocation mechanism definition module 20. The allocation mechanism definition module 20, in combination with the industry-wide tea product data, formulates tea classification rules, namely, divides price ranges and classifies teas within a unified price range into the same grade. Since teas within the same grade have different advantages, this is because different regions have different standards for judging tea quality. Therefore, the neural network construction module 30 compares the various evaluation criteria for different types of tea, divides the weights of different tea factors for different types, establishes a tea classification standard database, compares the various factors of new tea, and provides a price reference standard for new tea. For example, for two teas N and M at the same price grade, the color of tea N is the highest in the current price grade, so its color factor weight is the highest grade, while the leaf size of tea M is the highest in the current price grade, so its leaf size factor weight is the highest grade, which serves as a reference standard for the subsequent classification of new teas.

[0058] It is worth noting that the tea product data information in the industry obtained by the underlying logic data acquisition module 10 also includes the following data:

[0059] User behavior data: including user clicks, browsing, searches, purchases, and other behavioral data in applications (search engines, apps, and web pages, etc.); transaction data: including user purchase records, purchase amounts, purchase times, etc.; user feedback data: including user evaluations, ratings, and feedback on products; product data: including product images, descriptions, prices, inventory, and other information; user basic information: including basic information such as user age, gender, and geographic location; environmental data: including environmental factors such as time, location, and weather that may affect user behavior. After completing data collection, the collected data needs to be deeply processed and analyzed to reveal hidden patterns and trends in the data. This process includes three steps: data cleaning, data conversion, and data normalization. Steps are taken to ensure the quality and consistency of the data, and redundant data is removed. Statistical methods and machine learning algorithms are then used to analyze the data to determine the associations and patterns between the data, such as the relationship between purchasing trends and weather, and the relationship between product ratings and user purchases. A prediction model is constructed to predict future market trends, product sales, user behavior, etc. The prediction process is displayed through visualization technology to intuitively show the prediction results. The prediction results are then fed back to users to help them understand the meaning and possible impact of the prediction results. Decision support is provided to users based on the prediction results to help users make better business decisions and provide valuable insights and suggestions to decision makers in the tea industry, thereby optimizing business operations and improving efficiency and effectiveness.

[0060] Furthermore, the method for the neural network building module 30 to provide a new tea price reference standard includes the following steps:

[0061] S1. Compare various factors of new tea with the tea classification standard database;

[0062] S2. Combine the comparison results to obtain the matching price levels;

[0063] S3. Recommending new teas that are suitable for each price level and obtaining the recommendation rate for the corresponding price level;

[0064] S4. Arrange the tea leaves in order of their recommendation rates, and select the price grade with the highest recommendation rate as the recommended price for the new tea leaves.

[0065] In the process of comparing prices of new tea, since there are teas with the same factors in different price grades, there will often be multiple matching prices in the process of matching new tea prices. For example, the tea leaves are the same size and color, but because the taste evaluation standards in different regions are different, the prices of the two types of tea are different. At this time, if the new tea is successfully matched with it, it will be matched with two different prices. At this time, it is necessary to compare the various factors of the new tea through the tea classification standard database, combine the comparison results, eliminate irrelevant price influences, and obtain the various price grades that match it. At this time, it is necessary to make adaptive new tea recommendations according to each price grade, and obtain the recommendation rate of the corresponding price grade, that is, recommend the new tea to users who match the same batch in two different price forms, and obtain the recommendation rate of new tea in different price states, and then obtain the matching price that best suits the new tea according to the size of the recommendation rate.

[0066] Furthermore, the various factors used to compare the new tea leaves in S1 include the size, color, and shape of the tea leaves. During the comparison process, new tea samples are randomly selected and compared with the tea classification standard database, and a comparison difference range is divided. For example, during the tea size comparison process, the average size of each sample is measured and compared with the size of tea leaves in each price grade. Tea price grades that exceed the comparison difference range are eliminated, and the price grade that successfully matches is selected as the preliminary price of the new tea leaves.

[0067] Specifically, the product information interaction platform includes a tea keyword extraction module 40 and a chat content acquisition module 50. The tea keyword extraction module 40 is used to extract tea keywords that users interact with in chats on the tea data information interaction platform. The output end of the tea keyword extraction module 40 is connected to the input end of the chat content acquisition module 50. The chat content acquisition module 50 combines the tea keywords interacted with in chats to determine the type of tea the current user is talking about. During the process of users interacting with tea information on the tea data information interaction platform, the tea keyword extraction module 40 extracts the tea keywords that users interacted with in chats on the tea data information interaction platform, such as the shape, size and color of the tea, as a basis for later identifying the current type of tea. Subsequently, the chat content acquisition module 50 combines the tea keywords interacted with in chats to determine the type of tea the current user is talking about.

[0068] In addition, the chat content acquisition module 50 adopts a keyword comparison algorithm, and its algorithm formula is as follows:

[0069] ;

[0070] ;

[0071] ;

[0072] in The tea keyword set obtained from the current user's chat interaction. to Each tea keyword obtained from the current user's chat interaction, is the tea keyword set to be compared with it, to For each tea keyword to be compared, is the tea keyword comparison function, Keyword comparison overlap rate, is the keyword matching overlap rate threshold. Less than the keyword matching overlap rate threshold When the tea keyword comparison function The output is 0, indicating that the tea obtained by the current user chat interaction does not belong to the tea compared with it. Not less than the keyword matching overlap rate threshold When the tea keyword comparison function The output is 1, indicating that the tea leaves obtained by the current user chat interaction belong to the tea leaves being compared.

[0073] Since new users do not have enough knowledge about tea, they cannot obtain effective tea-related information and cannot determine the tea needs of this type of users. Furthermore, the output end of the chat content acquisition module 50 is connected to the tea data sharing module 60. The tea data sharing module 60 is used to share the tea type information mentioned by users. By sharing the tea type information mentioned by users through the tea data sharing module 60, new users can obtain data information about tea, such as keywords about tea, to provide data reference for obtaining the tea needs of new users in the later stage.

[0074] Furthermore, the method for the diversified product recommendation module 70 to formulate a diversified tea product recommendation model includes the following steps:

[0075] S10, obtaining the type of tea mentioned by the current user;

[0076] S20, combining with the tea classification standard database, determining various factors of the currently acquired tea;

[0077] S30. Based on the weights of the factors corresponding to the factors, various teas that match the weights of the factors are recommended as recommended teas, and a diversified tea recommendation model is established.

[0078] In the process of recommending multiple types of tea, first, it is necessary to obtain the type of tea mentioned by the current user, and then combine it with the tea classification standard database to determine the various factors of the currently obtained tea. For example, the tea belongs to the light-fragrant large-leaf type, and the type of tea the user prefers is not sure whether it is the light-fragrant type or the large-leaf type or both. Therefore, combined with the factor weights corresponding to each factor, various teas with suitable factor weights are recommended as recommended teas, and a multi-type tea recommendation model is established. When making recommendations, light-fragrant tea types with the same factor weight as the tea and large-leaf teas with the same factor weight are recommended. Finally, the recommendation rate fed back is used to determine which factor the user is more inclined to for tea, thereby further improving the recommendation adaptability.

[0079] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely preferred examples of the present invention and are not intended to limit the present invention. Various changes and improvements may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and improvements fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. An AI tea recommendation system that automatically analyzes tea data, characterized by: It includes a user behavior data acquisition module (10), a user transaction data acquisition module (20), a tea product purchase recommendation module (30), a user feedback data acquisition module (40), and a tea product recommendation model construction module (50); Wherein, the user behavior data acquisition module (10) is used to acquire data information of tea products that users pay attention to on the Internet; The user transaction data acquisition module (20) is used to acquire transaction data of users purchasing tea products; The tea purchase recommendation module (30) combines the transaction data of the user's tea purchase to obtain the amount of tea purchased by the user at different time states, and recommends tea of the same amount and different types to the user based on the current tea amount; The user feedback data acquisition module (40) is used to acquire the type of recommended tea purchased by the user and feedback data, and transmit the type of recommended tea purchased by the user and the feedback data to the tea purchase recommendation module (30); The tea recommendation model building module (50) combines the recommended tea types, the user's purchased tea types and the corresponding feedback data information to obtain the influence weight of each tea type on the user's purchase of the recommended tea in different stages of tea amount, and establishes a tea recommendation model based on the influence weight ratio, predicts the tea purchase trend of different tea purchase amount strata, and plans the corresponding tea type recommendation according to the tea purchase trend; The method for establishing a tea recommendation model in the tea recommendation model building module (50) comprises the following steps: S1. Planning unit statistics time T, determine the recommended tea varieties for the same purchase amount, and count the number of recommended tea varieties ; S2. Recommending recommended tea varieties with the same purchase amount to users with the same tea purchase amount tier; S3. Count the purchase volume of different recommended tea varieties within the unit statistical time T = Purchased users Purchase quantity ; S4. Combine user feedback data from the same tea purchase amount stratum to obtain the praise rate of different recommended tea types , To purchase the recommended tea varieties that have received positive reviews, The total purchase amount of recommended tea types by users in the same tea purchase amount stratum within the unit statistical time T; S5. Calculate the weights of different recommended tea types ,in The purchase quantity of the recommended tea type currently calculated, is the time period constant; S6. According to the weight of different recommended tea types Size: sort the recommended tea varieties with the same purchase amount by weight, and use this as the priority recommendation order for the same purchase amount; S7. Repeat the above steps to count the changes in weights of tea products with the same purchase amount in different time periods. Combined with environmental changes in different time periods, predict the environmental factors and influence trends that affect the purchase of recommended tea products.

2. The AI tea recommendation system for automatic analysis of tea data according to claim 1, characterized in that: The tea product data information acquired by the user behavior data acquisition module (10) includes application browsing, keyword search and tea product click volume.

3. The AI tea recommendation system for automatic analysis of tea data according to claim 1 is characterized by: The transaction data in the user transaction data acquisition module (20) includes purchase records, purchase time and purchase amount.

4. The AI tea recommendation system for automatic analysis of tea data according to claim 1 is characterized by: The tea purchase recommendation module (30) includes a tea price level positioning unit (310), a tea data information storage unit (320), and a tea type positioning recommendation unit (330); The tea product price tier positioning unit (310) locates the current user's tea product purchase price tier based on the user's tea purchase transaction data; The tea product data information repository (320) collects tea product purchase information within the industry, establishes a tea product repository, and stores purchase amount information of various types of tea products; The tea type positioning recommendation unit (330) combines the tea purchase amount level of the positioned user and retrieves different types of teas with the same tea amount from the tea storage library as recommended teas.

5. The AI tea recommendation system for automatic analysis of tea data according to claim 4 is characterized by: The tea types in the tea data information repository (320) include tea taste, tea craftsmanship and tea year.

6. The AI tea recommendation system for automatic analysis of tea data according to claim 1, characterized in that: The user feedback data acquisition module (40) includes a recommended tea quantity acquisition unit (410), a tea evaluation data acquisition unit (420), and an evaluation difference analysis unit (430). The recommended tea quantity acquisition unit (410) is used to obtain the recommended tea types purchased by users and the corresponding tea purchase quantities. The tea evaluation data acquisition unit (420) is used to obtain evaluation data after users purchase tea. The evaluation difference analysis unit (430) combines the evaluation data to distinguish between negative reviews and positive reviews.

7. The AI tea recommendation system for automatic analysis of tea data according to claim 1, characterized in that: The output end of the tea recommendation model construction module (50) is connected to an environmental factor data acquisition module (60), and the environmental factor data acquisition module (60) is used to obtain tea production environmental factors, including tea production time, production place and production place weather.

Images