Catering member intelligent marketing method and system based on large model

Abstract

The invention discloses a catering member intelligent marketing method and system based on a large model, and relates to the technical field of large model and business data cross analysis, and the method comprises the steps: firstly interpreting a commercial tenant demand, if the commercial tenant demand is a query demand, recognizing the query demand through the large model, organizing a language, and returning a query result; if yes, whether the commercial tenants need to be grouped and selected is determined, and if yes, designated grouping information is read, and then designated group selection operation is performed through interaction with the commercial tenants; if not, all information of a default grouping method is read for intelligent group selection; and marketing suggestions are returned according to the merchant marketing demands and the user grouping and selecting information. According to the method, multi-source data are integrated, semantic understanding, behavior prediction and dynamic optimization capabilities of a large model are utilized, and after marketing activity feasibility auditing and improvement are performed in combination with manual work, a system guides a merchant to execute a corresponding marketing activity, so that a more personalized and accurate marketing strategy can be realized, and the user experience and the merchant operation efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of cross - analysis of large models and commercial data, and particularly relates to an intelligent marketing method and system for catering members based on large models. Background Art

[0002] With the development of Internet and mobile payment technologies, the market competition in the catering industry has become increasingly fierce. Merchants have gradually realized the importance of membership marketing in enhancing customer loyalty and increasing consumption frequency. Traditional membership marketing systems mainly rely on simple incentive mechanisms such as point rules and discount offers to attract users. However, these methods are difficult to meet the personalized needs of users and cannot effectively improve user stickiness and consumption conversion rates. At the same time, the catering industry has accumulated a large amount of user behavior data, including multi - source data such as order information, consumption preferences, and geographical locations. Existing systems are difficult to deeply integrate and mine these data, resulting in the lack of precision in formulating and implementing marketing activities, and the pushed content is often ignored or rejected by users. In addition, the data isolation problem of online order platforms also limits the possibility of cross - scenario analysis, further affecting the effect of membership marketing.

[0003] Specifically, traditional membership marketing systems exhibit the following characteristics in practice: 1. Fixed point rules and discount offers: Existing membership marketing systems mostly adopt preset point reward mechanisms and general discount plans to attract users. Although this method is simple and direct, it lacks personalized incentive strategies and is difficult to meet the diverse needs of different users.

[0004] 2. Dependence on preset rules and simple algorithm models: Marketing decisions in traditional systems are mainly based on fixed business rules or simple statistical analysis methods (such as linear regression, classification trees, etc.). These methods are difficult to accurately capture the dynamic behavior changes and potential needs of users, resulting in the lack of pertinence and precision of the pushed content.

[0005] 3. Limited data integration ability: In actual operation, traditional membership systems often cannot effectively integrate multi - source heterogeneous data (such as order data, payment data, social interaction data, etc.). This limits the in - depth understanding and analysis of user behavior. A common phenomenon is that the user portraits of the system are not fine enough, resulting in insignificant marketing activity effects.

[0006] 4. Low efficiency and poor user experience: Since traditional marketing activities mostly rely on manual formulation and manual adjustment, it is difficult to achieve automation and real - time. This not only reduces operational efficiency but also may miss the best market opportunities. In addition, the selection of push channels and the timing are often not accurate enough, which is likely to cause user disgust or neglect.

[0007] With the development of large models and deep learning technologies, the intelligent data analysis ability has been significantly improved, which also provides new technical possibilities for member marketing in the catering industry. Summary of the Invention

[0008] The technical problem to be solved by the present invention is that the traditional member marketing system is difficult to meet the personalized needs of users and cannot deeply integrate and mine multi-source data, resulting in a lack of accuracy in the formulation and execution of marketing activities.

[0009] To solve the above technical problems, the technical solution of the present invention is as follows: A method for intelligent marketing of catering members based on a large model, comprising the following steps: Step 1: Identify the merchant's needs. If the merchant inputs by voice, use an automatic speech recognition model to convert the voice into text; Step 2: Interpret the merchant's needs and distinguish whether the merchant's needs are query needs or marketing suggestion needs; Step 3: If the merchant's needs are query needs, identify the query needs through the large model, search the database, and then organize the language to return the query results; if the merchant's needs are marketing suggestion needs, execute Step 4; Step 4: Confirm whether the merchant needs to perform group segmentation and group selection. If so, read the specified group segmentation information and then interact with the merchant to specify the group selection operation; if not, read all the information of the default group segmentation method and perform intelligent group selection based on the knowledge base and large model information; Step 5: Return marketing suggestions according to the merchant's marketing needs and user group segmentation and group selection information; Step 6: After the feasibility of the activity is reviewed by the large model and manually, the system will guide the merchant to execute the corresponding platform marketing activity.

[0010] Preferably, the user group segmentation and group selection method in Step 4 includes: group segmentation based on contribution value, that is, dividing users according to the consumption ratio of different user groups with different contribution values; group segmentation based on customer unit price, that is, dividing users according to the consumption ratio of different user groups with different customer unit prices; group segmentation based on age; group segmentation based on gender; group segmentation based on member "age" (i.e., registration time); group segmentation based on consumption frequency, and the consumption frequency is divided from three perspectives: the past 1 year, the past 1 quarter, and the past 1 month.

[0011] Preferably, for the method of clustering based on contribution value (the contribution value is the total consumption amount of the user): Obtain the order data for one year, calculate the contribution values of all users, and then divide all users into different contribution value intervals with 500 yuan as the division interval. For each interval, the total contribution value of all users within the interval can be calculated, and correspondingly, the contribution value proportion of each contribution value interval can be calculated (the total contribution value of the interval users divided by the total contribution value of all users). Based on the contribution value proportions of different interval user groups, interact with the merchant and then perform group selection or clustering.

[0012] Preferably, for the method of clustering based on average order value: Take 100 yuan as the division interval, divide all users into different average order value intervals. For each interval, the consumption proportion of all users within the interval can be calculated. Based on the consumption proportions of different interval user groups, interact with the merchant and then perform group selection or clustering.

[0013] A catering membership intelligent marketing system based on a large model, including: Merchant demand recognition module: Recognize merchant demands. If the customer inputs through voice, call the automatic speech recognition model to convert the voice into text. Merchant demand interpretation module: The merchant demand interpretation module is used to interpret the text converted by the speech recognition module or the directly input text, and specifically identify whether the merchant's demand is a query demand or a marketing suggestion demand. Database query module: The database query module searches the database according to the merchant's query demand, organizes the language and returns the query result. Database: The database is used to store the registration information and order information of all members, etc. User clustering and group selection module: Match the merchant's clustering and group selection demands and execute clustering and group selection, and return the clustering and group selection results. Marketing suggestion generation module: Based on the merchant's marketing demands and user clustering and group selection information, call the large model based on the knowledge base to return marketing suggestions. Knowledge base: The knowledge base is used to store the merchant's past marketing activities and the merchant's detailed information.

[0014] Preferably, it further includes an artificial review module. The artificial review module rationalizes the marketing suggestions manually and feeds them back to the marketing suggestion generation module. The marketing suggestion generation module uses the feedback as supplementary prompt words to interact with the large model to regenerate marketing suggestions. The artificial review module manually decides whether to adopt the marketing suggestions and supplements the detailed information related to the marketing activities.

[0015] Preferably, it further includes an automatic push module. The automatic push module uses the large language model to identify the push channels and content in the marketing suggestions, and jumps to the corresponding channels to guide the merchant to formulate marketing activities.

[0016] Preferably, it further includes a feedback collection mechanism module, which identifies the effectiveness of marketing activities based on sales volume and customer flow, and feeds back the feedback information to the knowledge base.

[0017] Preferably, it further includes an information supplement and modification module, which updates marketing resource restrictions and merchant-specific marketing requirements by interacting with merchants, and saves the updated content to the knowledge base.

[0018] Preferably, it further includes a pre-calculation module, which is used to count the information of each user group in each segmentation.

[0019] The present invention has the following beneficial effects compared with the prior art: By integrating multi-source data and utilizing the semantic understanding, behavior prediction, and dynamic optimization capabilities of large models, this application can achieve more personalized and precise marketing strategies, thereby enhancing the user experience and merchant operation efficiency.

[0020] The intelligent catering membership marketing system based on large models provided by this application significantly improves the accuracy, efficiency, and user experience of marketing activities through in-depth analysis of user behavior data and the use of advanced technical means, providing more effective operation support for merchants.

[0021] This application uses large language models and knowledge bases, combines the resource restrictions of merchants (such as budgets, event times, etc.), generates personalized marketing activity suggestions, and can provide differentiated coupon or recommended dish information for members of different segments, effectively enhancing user participation and conversion rates.

[0022] This application relies on advanced AI technology and intelligent systems to achieve full-life-cycle automated management of marketing activities. From strategy formulation, content generation to effect evaluation, it can be completed without manual intervention, significantly improving operation efficiency. In addition, the system can respond in real time to changes in user behavior and market dynamics to ensure the timeliness of marketing activities. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a logic diagram of an intelligent marketing system for catering membership based on large models of this application; Figure 2 It is an engineering implementation diagram of an intelligent marketing system for catering membership based on large models of this application; Figure 3 It is a screenshot of the usage status of an intelligent engineering system for catering membership based on large models of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] The following further describes the specific embodiments of the present invention in conjunction with the accompanying drawings. It should be noted here that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation on the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0025] As Figure 1 , 2 shown in and Figure 3, the present application provides a method for intelligent marketing of catering members based on a large model, and the method specifically includes the following steps: Step 1: Identify the merchant's needs. If the merchant inputs through voice, use an automatic speech recognition model to convert the voice into text; Step 2: Interpret the merchant's needs and distinguish whether the merchant's needs are query needs or marketing suggestion needs; Step 3: If the merchant's needs are query needs, after identifying the query needs through the large model and searching the database, organize the language to return the query results; if the merchant's needs are marketing suggestion needs, then execute Step 4. Specifically, the query statement is generated using the qwen-max model, and the merchant's needs are input to the large model. The large model will generate an SQL query statement, and the system will execute the SQL statement generated by the large model for database query operations; Step 4: Confirm whether the merchant needs to perform group segmentation and group selection. If so, read the specified group segmentation information, and then interact with the merchant to specify the group selection operation; if not, read all the information of the default group segmentation method, and perform intelligent group selection based on the knowledge base and large model information. For example: if it is a marketing activity for a new store, it is a general activity without a targeted group; if it is a Women's Day activity, only select the female group, which is group selection; if the merchant wants to conduct marketing activities for different user groups with different contribution value levels respectively, then segment the users based on the contribution value, where

[0026] The specific solution of the clustering method based on contribution value (the contribution value is the total consumption amount of the user) is as follows: Take the order data of one year, calculate the contribution value of all users, and then divide all users into different contribution value intervals with 500 yuan as the division interval. The total contribution value of all users in each interval can be calculated, and the contribution value ratio of each contribution value interval can be calculated accordingly (the total contribution value of the interval users divided by the total contribution value of all users). According to the contribution value ratio of different interval user groups, interact with the merchant to further select groups (for example: the merchant can select the group with the highest contribution value ratio to formulate marketing activities) or cluster (specify marketing activities respectively) to assist in the generation of subsequent marketing suggestions. The data volume and the endpoint values of the division interval of this clustering method can be optimized according to the actual needs of the merchant; The specific technical solution of the clustering method based on unit price is as follows: Take 100 yuan as the division interval, divide all users into different unit price intervals, and the consumption ratio of all users in each interval can be calculated. According to the consumption ratio of different interval user groups, interact with the merchant to further select groups (for example: the merchant can select the group with the highest consumption ratio to formulate marketing activities) or cluster to assist in the generation of subsequent marketing suggestions. The data volume and the endpoint values of the division interval of this clustering method can be optimized according to the actual needs of the merchant.

[0027] Step 5: Return marketing suggestions according to the merchant's marketing needs and the user group selection information; For example: A Cantonese restaurant wants to hold a marketing activity for International Women's Day, and the returned marketing suggestions are as follows: (1) Low-cost and high-perceived gifts: Customize small items such as bookmarks and sachets and integrate Cantonese cultural elements. The cost is low but it is easy to arouse the desire for collection.

[0028] (2) Package combination strategy: Promote high-margin beauty dishes (such as stews and desserts), and match with seasonal vegetables to balance costs.

[0029] (3) Time period diversion: Promote workplace packages at lunchtime and focus on family / girlfriend gatherings at dinner to increase the table turnover rate.

[0030] Step 6: After the large model and manual review the feasibility of the activity, the system will guide the merchant to execute the corresponding platform marketing activity.

[0031] As Figure 1 and 2 shown, this application also provides a catering membership intelligent marketing system based on a large model, including Merchant demand recognition module: Recognize merchant demands. If the merchant inputs by voice, call the automatic speech recognition model to convert the voice into text. If it is text, directly recognize it; Merchant Requirement Interpretation Module: The merchant requirement interpretation module is used to interpret the text converted by the speech recognition module or directly input text, and specifically identify whether the merchant's requirement is a query requirement or a marketing suggestion requirement; Query Requirement: Merchants can query member gender, age, registration date, consumption records, etc. For example, merchants can ask the intelligent marketing system for information such as the gender ratio of member users and the average customer unit price of member users.

[0032] Marketing Requirement: Merchants can put forward specific marketing requirements to the intelligent marketing system. For example: how to increase the consumption frequency of members, how to increase the customer unit price of member consumption, how to increase the passenger flow, how to convert low-value members into high-value members, and how to do marketing activities for a certain holiday, etc.

[0033] Database Query Module: The database query module searches the database according to the merchant's query requirement, and then organizes the language to return the query result; Database: The database is used to store the registration information and order information of all members, specifically such as member detailed order information, member tags. Member tags include: member gender, birthday date, registration date, etc.

[0034] Query Example: When a merchant asks about the number of male users among member users, the large model based on the query workflow guidance generates an SQL statement. After the system recognizes the SQL statement, it executes it and returns the statistical result to the large model. The large model further performs language organization processing and then outputs it to the merchant. The large model is Figure 2 The open-source large language model in the auxiliary tool, which belongs to the auxiliary tool. Here, qwen is used to generate SQL, and the model used for language organization of the query result is optional (optional qwen, deepseek, etc.).

[0035] User Grouping and Selection Module: Match the merchant's grouping and selection requirements and perform grouping and selection, and return the grouping and selection results. Statistics of user default grouping information; match the merchant's grouping and selection requirements and perform grouping and selection; return the grouping and selection results; rationality evaluation. User grouping and selection methods include: grouping based on contribution value, dividing users according to the consumption ratio of different user groups with different contribution values; grouping based on customer unit price, dividing users according to the consumption ratio of different customer unit price user groups; grouping based on age; grouping based on gender; grouping based on member "age" i.e., registration time; grouping based on consumption frequency, which is divided from three perspectives: the past 1 year, the past 1 quarter, and the past 1 month. The large model identifies whether the merchant's requirement needs grouping and selection. If so, it reads the specified grouping information, and then performs the specified selection operation through interaction with the merchant; if not, it will read all the information of the default grouping method, and based on knowledge base information such as marketing scenarios and marketing resource limitations, the large model performs intelligent selection.

[0036] The marketing suggestion generation module uses a knowledge base to call upon a large model to return marketing suggestions based on merchant marketing needs and user group selection information. The decisions made by the large model must be based on the knowledge base, and the execution of the decisions must be manually reviewed and self-checked.

[0037] The knowledge base includes: past marketing campaign information; detailed merchant information; known merchant marketing resource limitations; and tutorials for issuing coupons through different marketing channels. Merchant details include: business model; customer base; and merchant scale. Marketing resource limitations include: marketing funding limitations; marketing channel limitations, etc.

[0038] Preferably, it also includes a manual review module, which manually rationalizes the marketing suggestions and feeds back to the marketing suggestion generation module. The marketing suggestion generation module uses the feedback as a supplementary prompt word to interact with the large model to regenerate the marketing suggestions. The manual review module manually decides whether to adopt the marketing suggestions and supplements the detailed information related to the marketing activities.

[0039] Preferably, the system also includes an automatic push module, which uses a large language model to identify the push channel and content in the marketing proposal, redirects to the corresponding channel, and guides merchants in developing marketing campaigns. The system uses a large language model to identify information such as the push channel and content in the marketing proposal; identifies the channel and redirects to the corresponding channel; if the channel supports automatic URL parameter filling, the marketing campaign information is passed via URL parameters; if the channel does not support automatic URL parameter filling, the system redirects to a page to guide merchants in creating coupons and developing marketing campaigns. Automated push functionality: After manual review and confirmation, the system automatically pushes marketing campaign information to target users via SMS or other means, reducing manual intervention and improving efficiency. Specifically, some platforms support automatic URL parameter filling (adding URL parameters after calling the platform API URL, for example, https: / / example.com / product?source=facebook&campaign=spring_sale, where source=facebook and campaign=spring_sale are URL parameters). The platform's coupon issuance API is called and the coupon details (such as expiration date, coupon amount, whether there is a minimum purchase requirement or no purchase requirement) are sent to the platform, which then automatically fills in the information based on the received information. Some platforms do not support automatic filling of URL parameters and can only guide merchants to create coupons and complete coupon information on their own.

[0040] Preferably, it also includes a feedback collection mechanism module, which identifies the effectiveness of marketing activities based on sales volume and customer flow, and feeds feedback information back to the knowledge base.

[0041] Preferably, it further includes an information supplement and modification module. The information supplement and modification module updates the marketing resource restrictions and merchant-specific marketing requirements by interacting with the merchant, and saves the updated content to the knowledge base.

[0042] Preferably, it further includes a pre-calculation module. The pre-calculation module is used to count the information of each user group in each segmentation. The pre-calculation module updates the user groups every day at midnight, reads data from the database for calculation, and then updates the segmentation table in the database.

[0043] The embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings, but the present invention is not limited to the described embodiments. For those skilled in the art, without departing from the principle and spirit of the present invention, various changes, modifications, substitutions, and variations to these embodiments still fall within the protection scope of the present invention.

Claims

1. An intelligent marketing method for catering members based on large models, characterized in that: It includes the following steps: Step 1: Identify the merchant's needs. If the merchant inputs by voice, use an automatic speech recognition model to convert the voice into text; Step 2: Interpret the merchant's needs and distinguish whether the merchant's needs are query needs or marketing suggestion needs; Step 3: If the merchant has a query need, identify the query need through a large model, search the database, and then organize the language to return the query result; if the merchant has a marketing suggestion need, execute Step 4; Step 4: Confirm whether the merchant needs to perform group segmentation and group selection. If so, read the specified group segmentation information and then interact with the merchant to specify the group selection operation; if not, read all the information of the default group segmentation method and perform intelligent group selection based on the knowledge base and large model information; Step 5: Return marketing suggestions according to the merchant's marketing needs and the user's group segmentation and group selection information; Step 6: After the large model and manual review the feasibility of the activity, the system will guide the merchant to execute the corresponding platform marketing activity.

2. The intelligent marketing method for catering members based on a large model according to claim 1, characterized in that: The user group segmentation and group selection methods in Step 4 include: group segmentation based on contribution value, that is, dividing users according to the consumption ratio of different user groups with different contribution values; group segmentation based on customer unit price, that is, dividing users according to the consumption ratio of different customer unit price user groups; group segmentation based on age; group segmentation based on gender; group segmentation based on member "age" (i.e., registration time); group segmentation based on consumption frequency, and the consumption frequency is divided from three perspectives: the past 1 year, the past 1 quarter, and the past 1 month.

3. The intelligent marketing method for catering members based on a large model according to claim 2, wherein: The group segmentation method based on contribution value (the contribution value is the total consumption amount of the user): Take the order data of 1 year, calculate the contribution value of all users, and then divide all users into different contribution value intervals with 500 yuan as the division interval. The total contribution value of all users in each interval can be calculated, and the contribution value ratio of each contribution value interval can be calculated accordingly (the total contribution value of the interval users divided by the total contribution value of all users). According to the contribution value ratio of different interval user groups, interact with the merchant to perform group selection or group segmentation.

4. The intelligent marketing method for catering members based on a large model according to claim 3, characterized in that: The group segmentation method based on customer unit price: Divide all users into different customer unit price intervals with 100 yuan as the division interval. The consumption ratio of all users in each interval can be calculated. According to the consumption ratio of different interval user groups, interact with the merchant and then perform group selection or group segmentation.

5. A catering membership intelligent marketing system based on a large model, characterized in that: including Merchant demand identification module: Identify the merchant's needs. If the customer inputs by voice, call the automatic speech recognition model to convert the voice into text; Merchant demand interpretation module: The merchant demand interpretation module is used to interpret the text converted by the speech recognition module or directly input text, and specifically identify whether the merchant's needs are query needs or marketing suggestion needs; Database query module: The database query module searches the database according to the merchant's query needs and then organizes the language to return the query result; Database: The database is used to store the registration information and order information of all members, etc.; User group segmentation and group selection module: Match the merchant's group segmentation and group selection needs and perform group segmentation and group selection, and return the group segmentation and group selection results; Marketing suggestion generation module: According to the merchant's marketing needs and the user's group segmentation and group selection information, based on the knowledge base, call the large model to return marketing suggestions; Knowledge base: The knowledge base is used to store the merchant's past marketing activities and the merchant's detailed information.

6. The intelligent marketing system for catering members based on a large model according to claim 5, characterized in that: It also includes a manual review module. The manual review module rationalizes marketing suggestions manually and feeds them back to the marketing suggestion generation module. The marketing suggestion generation module uses the feedback as supplementary prompt words to interact with the large model to regenerate marketing suggestions. The manual review module manually decides whether to adopt the marketing suggestions and supplements the detailed information related to the marketing activities.

7. The intelligent marketing system for catering members based on a large model according to claim 6, characterized in that: It also includes an automatic push module. The automatic push module uses a large language model to identify the push channels and content in the marketing suggestions, and after jumping to the corresponding channels, guides the merchant to formulate marketing activities.

8. The intelligent marketing system for catering members based on a large model according to claim 7, wherein: It also includes a feedback collection mechanism module. The feedback collection mechanism module identifies the effectiveness of the marketing activities based on sales volume and customer flow, and feeds back the feedback information to the knowledge base.

9. The intelligent marketing system for catering members based on a large model according to claim 8, wherein: It also includes an information supplement and modification module. The information supplement and modification module interacts with the merchant to update the marketing resource limitations and the merchant's specific marketing requirements, and saves the updated content to the knowledge base.

10. The intelligent marketing system for catering members based on a large model according to claim 9, wherein: It also includes a pre-computation module. The pre-computation module is used to count the information of each user group in each segmentation.

Images