A Marketing Plan Generation Method and System Based on Multi-Model Integration

Abstract

This invention provides a marketing plan generation method and system based on a multi-model combination. The marketing plan generation method includes: acquiring customer tag data and product data; using a pre-trained first model to perform business classification on the customer tag data and product data respectively, generating tag business label data and product business label data; using a pre-trained second model to generate the product-customer tag contribution between the customer tag data and product data; acquiring target product information, and combining the tag business label data, product business label data, and product-customer tag contribution to generate a tag logical expression; and using a pre-trained third model to generate a marketing plan based on the target product information and the tag logical expression. This invention achieves deep matching between customer tags and product features through a multi-model collaborative mechanism, improving the accuracy and automation level of marketing plans.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and more specifically to a method for generating marketing plans based on the combination of multiple models. Background Technology

[0002] In the telecommunications market, marketing plans are a core element of business development. They not only help companies accurately target customers and deliver suitable products, but also effectively enhance brand awareness and product conversion rates, creating sustainable economic benefits. The core objective of developing a marketing plan is to use scientific data mining techniques, combined with the needs and characteristics of different customers, to customize personalized marketing strategies, thereby improving marketing effectiveness and customer satisfaction.

[0003] Current marketing plans rely heavily on human experience, a complex and inefficient process. Marketers must first conduct in-depth statistical and correlation analysis of customer data, then segment customers based on factors such as age, gender, DOU, ARPU, and interests, with product improvement as the goal. Finally, they select target customer groups and develop marketing recommendation strategies based on their personal experience.

[0004] Because the marketing plan generation process described above relies entirely on human experience, it is highly subjective and lacks standardized data analysis and decision-making basis. Furthermore, the limited capacity of humans to process massive amounts of complex data results in insufficient accuracy in identifying target customers, failing to accurately capture their true needs. This, in turn, affects the targeting and effectiveness of the marketing plan, making it difficult to meet the modern market's demands for efficient and precise marketing. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a marketing plan generation method based on multi-model integration. The aim is to achieve deep matching between customer tags and product characteristics through a multi-model collaborative mechanism, thereby improving the accuracy and automation level of marketing plans.

[0006] This invention discloses a marketing plan generation method based on multi-model combination, comprising: Acquire customer tag data and product data; The first pre-trained model is used to classify customer tag data and product data into business categories, generating tag business label data and product business label data. The tag business label data is used to represent the relationship between customer tag data and business categories, and the product business label data is used to represent the relationship between product data and business categories. The pre-trained second model is used to generate the product-customer tag contribution between customer tag data and product data. The product-customer tag contribution is used to characterize the relationship between customer tag data, product data and marketing influence degree. Marketing influence degree is used to characterize the influence of customer tag data on the corresponding product of product data. Marketing influence degree includes positive influence and negative influence. Obtain target product information, and combine it with tag business marking data, product business marking data and product-customer tag contribution to generate tag logical expressions. The tag logical expressions are used to describe the target customer characteristics corresponding to the target product information. A pre-trained third model is used to generate a marketing plan corresponding to the target product information based on the target product information and the tag logic expression.

[0007] Preferably, the customer tag data includes the English name of the tag, the Chinese name of the tag, the tag description, the tag data example, and the tag code value. The tag description is used to describe the customer's basic attributes, behavioral characteristics, business preferences, and status information. Product data includes product name, product category, product attributes, and product characteristics.

[0008] Preferably, the marketing plan generation method based on multi-model combination further includes: Construct a first prompt word, which is used to instruct the first model to match at least one business category for the customer tag data; Construct a second cue word, which is used to instruct the first model to match at least one business category for the product data.

[0009] Preferably, the pre-trained first model is used to perform business classification on customer tag data and product data respectively, generating tag business label data and product business label data, including: The customer tag data and the first prompt word are input into the first model. The first model generates the business category corresponding to the customer tag data based on the Chinese name and description of the tag, and outputs the tag business label data. The product data and the second prompt word are input into the first model. The first model generates the business category corresponding to the product data based on the product name, product category and product characteristics, and outputs the product business tag data.

[0010] Preferably, by combining tag business tagging data, product business tagging data, and product-customer tag contributions, a tag logical expression is generated, including: Based on the product business tagging data, the target business scenario corresponding to the target product information is generated, and based on the tag business tagging data, the customer tag data corresponding to the target business scenario is generated as the first customer profile. Based on the product-customer tag contribution, obtain the positive contribution customer tag data corresponding to the target product information to serve as the second customer profile. The positive contribution customer tag data is the customer tag data corresponding to the degree of marketing influence being positive. The first customer profile and the second customer profile are merged to obtain merged customer tags, and tag logical expressions are generated based on the merged customer tags. The tag logical expressions are used to define the target customer group.

[0011] Preferably, the positive contribution customer tag data corresponding to the target product information is obtained based on the product-customer tag contribution, including: Valid contribution data with a positive marketing impact were selected from the product-customer tag contributions. Based on the effective contribution data, obtain at least one target customer tag data corresponding to the target product information; The target customer tag data is sorted in descending order of its contribution value in terms of marketing impact, and the top-ranked target customer tag data is extracted to obtain the positive contribution customer tag data.

[0012] Preferably, after obtaining the integrated customer tags, the method further includes: Logical conflict verification is performed on the integrated customer tags. Logical conflict verification is used to verify whether there are logical contradictions in the standard rooms of integrated customers. Perform tag deduplication on the merged customer tags that pass the logical conflict check; Add regular logical tags to the merged customer tags after tag deduplication.

[0013] Preferably, generating a tag logic expression based on the integrated customer tags includes: Based on a predefined expression template, logical operations are performed on the merged customer tags to obtain tag logical expressions. The logical operations include AND, OR, and NOT operations.

[0014] Preferably, a pre-trained third model is used to generate a marketing plan corresponding to the target product information based on the target product information and the tag logical expression, including: Construct a third prompt word, which is used to instruct the third model to generate a marketing plan according to preset format and preset content requirements; Input the third prompt word, target product information, and tag logic expression into the third model so that the third model can output a marketing plan. The marketing plan includes one or more of the following: plan name, plan description, product introduction, customer group information, estimated number of customers, and marketing script.

[0015] This invention discloses a marketing plan generation system based on multi-model integration, used to execute the aforementioned marketing plan generation method based on multi-model integration. The marketing plan generation system based on multi-model integration includes: The data acquisition module is configured to acquire customer tag data and product data. The business classification module is configured to: use a pre-trained first model to classify customer tag data and product data into business categories, generating tag business label data and product business label data. The tag business label data is used to represent the relationship between customer tag data and business classification, and the product business label data is used to represent the relationship between product data and business classification. The contribution calculation module is configured to: use a pre-trained second model to generate product-customer tag contribution between customer tag data and product data. Product-customer tag contribution is used to characterize the relationship between customer tag data, product data and marketing influence degree. Marketing influence degree is used to characterize the influence of customer tag data on the corresponding product of product data. Marketing influence degree includes positive influence and negative influence. The tag generation module is configured to: obtain target product information, and combine tag business tag data, product business tag data and product-customer tag contribution to generate tag logical expressions. The tag logical expressions are used to describe the target customer characteristics corresponding to the target product information. The solution output module is configured to generate a marketing solution corresponding to the target product information based on the target product information and the tag logical expression using a pre-trained third model.

[0016] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention, through a multi-model collaborative approach, deeply correlates and analyzes customer tag data and product data, effectively solving the problems of strong subjectivity, low efficiency, and insufficient accuracy in traditional manually generated marketing plans. Specifically, the first model achieves accurate matching of customer tags and product data within business scenarios, providing a business dimension foundation for subsequent customer profiling; the second model, by quantifying the marketing impact of customer tags on products, filters out key tags that positively contribute to the target product, further narrowing the target customer range; finally, the third model, based on standardized target product information and tag logical expressions, automatically generates a structurally complete and highly targeted marketing plan, realizing a shift in marketing decision-making from relying on manual experience to data-driven, model-collaborative approaches. This multi-model combined method improves the efficiency and standardization of marketing plan generation, provides a scientific basis for customized marketing strategies, and helps improve product conversion rates and customer satisfaction. Attached Figure Description

[0017] Figure 1A flowchart illustrating the marketing plan generation method based on multi-model combination provided by the present invention; Figure 2 This is a schematic diagram of the marketing solution generation system based on multi-model combination provided by the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] The present invention will now be described in further detail with reference to the accompanying drawings.

[0020] This invention provides a marketing plan generation method based on a combination of multiple models, such as... Figure 1 As shown, the marketing plan generation method based on multi-model combination includes the following steps.

[0021] Step S1: Obtain customer tag data and product data.

[0022] In this embodiment of the invention, customer tag data includes the English name of the tag, the Chinese name of the tag, the tag description, a tag data example, and the tag code value. The tag description describes the customer's basic attributes, behavioral characteristics, business preferences, and status information. Product data includes the product name, product category, product attributes, and product features. Obtaining customer tag data and product data provides fundamental information support for subsequent model processing, ensuring that the association between customer tags and product features is calculable and business-reasonable.

[0023] S2. Using the pre-trained first model, perform business classification on customer tag data and product data respectively, and generate tag business label data and product business label data.

[0024] In this embodiment of the invention, the first model is based on a large-scale model intent classification method to achieve accurate classification of customer tags and product data. Tag business label data is used to characterize the association between customer tag data and business classification, while product business label data is used to characterize the association between product data and business classification.

[0025] When using the first model for business classification, it is first necessary to construct a first prompt word and a second prompt word. The first prompt word is used to instruct the first model to match at least one business category for customer tag data. The second prompt word is used to instruct the first model to match at least one business category for product data.

[0026] Next, customer tag data and the first prompt word are input into the first model. The first model generates the business category corresponding to the customer tag data based on the Chinese name and description of the tag, and outputs the tag business label data. Product data and the second prompt word are input into the first model. The first model generates the business category corresponding to the product data based on the product name, product category, and product characteristics, and outputs the product business label data.

[0027] In this way, the first model can be used to achieve a unified mapping between customer tags and product data business categories, ensuring that data from different sources are comparable and correlated under the same business dimension. This provides a unified business semantic foundation for subsequent tag and product matching, improving the accuracy and interpretability of cross-domain association.

[0028] S3. Utilize the pre-trained second model to generate the product-customer tag contribution between customer tag data and product data.

[0029] In this embodiment of the invention, the product-customer tag contribution is used to characterize the correlation between customer tag data, product data and marketing influence degree, and the marketing influence degree is used to characterize the degree of influence of customer tag data on the product corresponding to the product data, including positive influence and negative influence.

[0030] The second model is based on large-scale model correlation analysis. By learning the correlation patterns between customer tags and product conversions in historical marketing data, it quantifies the marketing impact of each customer tag on different products. For example, the input of the second model is the Chinese name and description of the customer tag data, and the product name and product characteristics of the product data. The output is a product-customer tag contribution matrix containing a triple of "customer tag-product-marketing impact".

[0031] In some embodiments, the degree of marketing impact is represented numerically, with positive values ​​representing positive impact and negative values ​​representing negative impact. The larger the absolute value, the stronger the impact. Through the calculation of the second model, the limitations of traditional human experience judgment can be overcome, and potential, non-intuitive correlations between customer tags and products can be mined from massive amounts of data, providing data support for the subsequent accurate positioning of target customer groups.

[0032] S4. Obtain target product information and, in conjunction with tag business marking data, product business marking data, and product-customer tag contributions, generate tag logic expressions.

[0033] In this embodiment of the invention, the tag logical expression is used to describe the target customer characteristics corresponding to the target product information. During the generation of the tag logical expression, a target business scenario corresponding to the target product information is generated based on the product business tag data, and customer tag data corresponding to the target business scenario is generated based on the tag business tag data, serving as a first customer profile. The first customer profile is used to describe the characteristics of the ideal customer group corresponding to the target product in a specific business scenario.

[0034] Furthermore, based on the product-customer tag contribution, positive contribution customer tag data corresponding to the target product information is obtained as the second customer profile. This positive contribution customer tag data represents the customer tag data corresponding to a positive marketing impact. Specifically, in obtaining positive contribution customer tag data, firstly, valid contribution data with a positive marketing impact is filtered from the product-customer tag contributions. Next, at least one target customer tag data corresponding to the target product information is obtained based on the valid contribution data. Finally, the target customer tag data is sorted in descending order of marketing impact contribution value, and a predetermined number of target customer tag data points are extracted from the top rankings to obtain the positive contribution customer tag data. The second customer profile focuses on the customer characteristics that actually generate positive conversions under data-driven principles, reflecting the true preference group for the target product in the market.

[0035] Furthermore, the first and second customer profiles are merged to obtain merged customer tags. Tag logical expressions are then generated based on these merged customer tags, which are used to define the target customer group. Specifically, logical operations are performed on the merged customer tags according to a predefined expression template to obtain tag logical expressions. These logical operations include AND, OR, and NOT operations.

[0036] In some embodiments, after obtaining the merged customer tags, a logical conflict check is performed on the merged customer tags. The logical conflict check is used to verify whether there are logical contradictions between merged customer tags. Then, the merged customer tags that pass the logical conflict check are deduplicated; subsequently, regular logical tags are added to the deduplicated merged customer tags.

[0037] This approach further enhances the rationality and completeness of integrated customer tags, preventing deviations in target customer group definitions due to tag conflicts. For example, if the integrated customer tags contain contradictory labels such as "age < 25 years old" and "age > 40 years old," the logical conflict check will automatically identify and prompt corrections. Through this series of processes, the integrated customer tags retain the ideal characteristics preset for the business scenario while incorporating real conversion factors verified by data, laying a solid foundation for generating accurate target customer group definitions in the future.

[0038] The following example illustrates the process of generating tag logic expressions. Assume the target product is a "smart home air purifier," and the target business scenario corresponding to its product business tag data is "home health." The first customer profile consists of related tags within this scenario, such as "family with elderly / children," "concerned about indoor air quality," and "living in an area prone to smog." Simultaneously, positively impactful tags are selected from the product-customer tag contributions, sorted by contribution value, and the top 5 are extracted to obtain the second customer profile, such as "searched for air purification products in the past 3 months," "purchased smart home devices," "monthly spending over 2000 yuan," "living area ≥ 80㎡," and "interested in health and wellness information." After merging the two profiles, a logic conflict check is performed, such as checking for contradictory tags like "living in an area prone to smog" and "living in a coastal city with good air quality." If a conflict is not detected, it is automatically removed. Next, duplicate tags such as "concerned about indoor air quality" and "interested in health and wellness information" are deduplicated. Finally, regular logic tags such as "age 25-55 years old" and "family monthly income ≥ 15000 yuan" are added. The final generated tag logic expression is: (Family with elderly / children OR Concerned about indoor air quality OR Living in an area with high smog incidence) AND (Searched for air purification products in the past 3 months OR Purchased smart home devices) AND Monthly spending over 2000 yuan AND Living area ≥ 80㎡ AND Age between 25-55 years old AND Family monthly income ≥ 15000 yuan. This expression, through a combination of AND and OR operations, covers potential demand groups in the business scenario and incorporates high conversion characteristics at the data level, achieving precise definition of the target customer group.

[0039] S5. Utilize the pre-trained third model to generate a marketing plan corresponding to the target product information based on the target product information and the tag logic expression.

[0040] In the process of generating a marketing plan, a third prompt word is first constructed. This third prompt word instructs the third model to generate a marketing plan according to preset format and content requirements. Next, the third prompt word, target product information, and tag logic expressions are input into the third model, causing it to output a marketing plan. The marketing plan includes one or more of the following: plan name, plan description, product introduction, customer group information, estimated customer group size, and marketing script.

[0041] This ensures that marketing plans are both accurate and feasible. By semantically parsing the tag logic expressions through the model, it automatically matches the core needs of the target customer group with the product's value points, avoiding the problem of product selling points being out of touch with customer needs in traditional plans. Through the generation capabilities of the third model, marketing plans are no longer fragmented ideas relying on experience, but rather systematic plans with standardized structures, clear logic, and a high degree of alignment with the needs of the target customers, effectively improving the efficiency and effectiveness of marketing execution.

[0042] like Figure 2 As shown, the present invention also provides a marketing plan generation system based on multi-model integration, used to execute a marketing plan generation method based on multi-model integration. The marketing plan generation system based on multi-model integration includes: a data acquisition module 201, a business classification module 202, a contribution calculation module 203, a tag generation module 204, and a plan output module 205.

[0043] Specifically, the data acquisition module 201 is configured to acquire customer tag data and product data. The business classification module 202 is configured to use a pre-trained first model to classify customer tag data and product data into business categories, generating tag business label data and product business label data. The tag business label data is used to characterize the relationship between customer tag data and business categories, and the product business label data is used to characterize the relationship between product data and business categories. The contribution calculation module 203 is configured to use a pre-trained second model to generate product-customer tag contribution between customer tag data and product data. The product-customer tag contribution is used to characterize the relationship between customer tag data, product data, and marketing influence degree. The marketing influence degree is used to characterize the degree of influence of customer tag data on the corresponding product in the product data. The marketing influence degree includes positive and negative influence. The tag generation module 204 is configured to acquire target product information and, in combination with tag business label data, product business label data, and product-customer tag contribution, generate tag logical expressions. The tag logical expressions are used to describe the target customer characteristics corresponding to the target product information. The solution output module 205 is configured to generate a marketing plan corresponding to the target product information based on the target product information and the tag logical expression using a pre-trained third model.

[0044] As can be seen from the above technical solution, this invention provides a marketing plan generation method and system based on multi-model combination. The marketing plan generation method includes: acquiring customer tag data and product data; using a pre-trained first model to perform business classification on the customer tag data and product data respectively, generating tag business label data and product business label data; using a pre-trained second model to generate the product-customer tag contribution between the customer tag data and product data; acquiring target product information, and combining the tag business label data, product business label data, and product-customer tag contribution to generate a tag logical expression; and using a pre-trained third model to generate a marketing plan based on the target product information and the tag logical expression. This invention achieves deep matching between customer tags and product features through a multi-model collaborative mechanism, improving the accuracy and automation level of marketing plans.

[0045] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A marketing plan generation method based on multi-model combination, characterized in that, include: Acquire customer tag data and product data; The customer tag data and the product data are classified into business categories using a pre-trained first model, generating tag business label data and product business label data. The tag business label data is used to characterize the relationship between the customer tag data and the business category, and the product business label data is used to characterize the relationship between the product data and the business category. The product-customer tag contribution between the customer tag data and the product data is generated using a pre-trained second model. The product-customer tag contribution is used to characterize the correlation between the customer tag data, the product data, and the degree of marketing influence. The degree of marketing influence is used to characterize the degree of influence of the customer tag data on the product corresponding to the product data. The degree of marketing influence includes positive influence and negative influence. Obtain target product information, and combine the tag business marking data, the product business marking data, and the product-customer tag contribution to generate a tag logical expression. The tag logical expression is used to describe the target customer characteristics corresponding to the target product information. A pre-trained third model is used to generate a marketing plan corresponding to the target product information based on the target product information and the tag logical expression.

2. The marketing plan generation method according to claim 1, characterized in that, The customer tag data includes the English name of the tag, the Chinese name of the tag, the tag description, the tag data example, and the tag code value. The tag description is used to describe the customer's basic attributes, behavioral characteristics, business preferences, and status information. The product data includes product name, product category, product attributes, and product characteristics.

3. The marketing plan generation method according to claim 2, characterized in that, Also includes: Construct a first prompt word, which is used to instruct the first model to match at least one business category for the customer tag data; Construct a second prompt word, which is used to instruct the first model to match at least one business category for the product data.

4. The marketing plan generation method according to claim 3, characterized in that, The step of using a pre-trained first model to perform business classification on the customer tag data and the product data, respectively, to generate tag business label data and product business label data, includes: The customer tag data and the first prompt word are input into the first model. The first model generates the business category corresponding to the customer tag data based on the Chinese name of the tag and the tag description, and outputs the tag business mark data. The product data and the second prompt word are input into the first model. The first model generates a business category corresponding to the product data based on the product name, the product category and the product characteristics, and outputs the product business tag data.

5. The marketing plan generation method according to claim 1, characterized in that, The step of combining the tag business tagging data, the product business tagging data, and the product-customer tag contribution to generate a tag logical expression includes: Based on the product business tagging data, a target business scenario corresponding to the target product information is generated, and based on the tag business tagging data, customer tag data corresponding to the target business scenario is generated as a first customer profile. Based on the product-customer tag contribution, obtain the positive contribution customer tag data corresponding to the target product information to serve as the second customer profile. The positive contribution customer tag data is the customer tag data corresponding to the marketing influence degree being positive. The first customer profile and the second customer profile are merged to obtain merged customer tags, and the tag logical expression is generated based on the merged customer tags. The tag logical expression is used to define the target customer group.

6. The marketing plan generation method according to claim 5, characterized in that, The step of obtaining the positive contribution customer tag data corresponding to the target product information based on the product-customer tag contribution includes: Valid contribution data with a positive impact level are selected from the product-customer tag contributions; Based on the effective contribution data, obtain at least one target customer tag data corresponding to the target product information; The target customer tag data is sorted in descending order of contribution value according to the degree of marketing influence, and the top preset number of target customer tag data is extracted to obtain the positive contribution customer tag data.

7. The marketing plan generation method according to claim 5, characterized in that, After obtaining the integrated customer tags, it also includes: The integrated customer tags are subjected to logical conflict verification, which is used to verify whether there is a logical contradiction between the integrated customer standard rooms. Perform tag deduplication on the fused customer tags that pass the logical conflict check; After deduplication of the tags, the merged customer tags are supplemented with regular logical tags.

8. The marketing plan generation method according to claim 5, characterized in that, The step of generating the tag logical expression based on the fused customer tag includes: Based on a predefined expression template, logical operations are performed on the fused customer tags to obtain the tag logical expression. The logical operations include AND, OR, and NOT operations.

9. The marketing plan generation method according to claim 1, characterized in that, The step of generating a marketing plan corresponding to the target product information using a pre-trained third model based on the target product information and the tag logical expression includes: A third prompt word is constructed, which is used to instruct the third model to generate a marketing plan according to a preset format and preset content requirements; The third prompt word, the target product information, and the tag logic expression are input into the third model so that the third model outputs the marketing plan. The marketing plan includes one or more of the following: plan name, plan description, product introduction, customer group information, estimated number of customers, and marketing script.

10. A marketing plan generation system based on multi-model combination, used to execute any one of the marketing plan generation methods based on multi-model combination as described in claims 1-9, characterized in that, include: The data acquisition module is configured to acquire customer tag data and product data; The business classification module is configured to: use a pre-trained first model to classify the customer tag data and the product data into business categories, respectively, and generate tag business label data and product business label data. The tag business label data is used to characterize the association between the customer tag data and the business classification, and the product business label data is used to characterize the association between the product data and the business classification. The contribution calculation module is configured to: generate product-customer tag contribution between the customer tag data and the product data using a pre-trained second model, wherein the product-customer tag contribution is used to characterize the correlation between the customer tag data, the product data and the degree of marketing influence, wherein the degree of marketing influence is used to characterize the degree of influence of the customer tag data on the product corresponding to the product data, and wherein the degree of marketing influence includes positive influence and negative influence; The tag generation module is configured to: acquire target product information, and combine the tag business marking data, the product business marking data and the product-customer tag contribution to generate a tag logical expression, wherein the tag logical expression is used to describe the target customer characteristics corresponding to the target product information; The solution output module is configured to generate a marketing solution corresponding to the target product information based on the target product information and the tag logical expression using a pre-trained third model.

Images