Marketing strategy generation method and system based on artificial intelligence

Abstract

The embodiment of the invention relates to the technical field of artificial intelligence, and discloses a marketing strategy generation method and system based on artificial intelligence, and the method comprises the steps: firstly obtaining current and same historical return information of a target user; determining an expected demand index of the target user for the pre-purchased commodity according to the return information; correcting the expected demand index in combination with a purchase impulse model corresponding to the target user portrait to obtain actual demand intensity; when the actual demand intensity exceeds a threshold value, obtaining candidate commodity information associated with the pre-purchased commodity; and then determining a plurality of target commodities according to the return reason and the purchase impulse model, and forming marketing strategy information containing a marketing display sequence. Therefore, according to the technical scheme, the user demand can be accurately analyzed, the demand evaluation is corrected in combination with the purchase impulse factor, the accuracy of judging the actual demand of the user is improved, a marketing strategy is formulated based on the actual demand of the user, commodities are accurately recommended, and the marketing effect is improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and particularly to a method and system for generating marketing strategies based on artificial intelligence. Background Art

[0002] With the rapid development of e-commerce, precise marketing strategies have become the key to improving user experience and enterprise benefits. Traditional marketing methods mainly rely on user behavior data or historical purchase records for product recommendations, but ignore the demand signals contained in users' return behavior. The existing technologies have the following problems: return data is only used for quality monitoring and not converted into demand insights, and the formulation of marketing strategies depends on empirical rules, unable to achieve personalized recommendations, resulting in low marketing efficiency. Summary of the Invention

[0003] The main objective of the present invention is to provide a method and system for generating marketing strategies based on artificial intelligence, aiming to solve the technical problem of low marketing efficiency in the existing technologies.

[0004] To achieve the above objective, in a first aspect, an embodiment of the present application provides a method for generating a marketing strategy based on artificial intelligence, which is applied to a marketing management system. The method includes: Obtain the current return information of a target user and historical return information associated with the current return information, where the historical return information is the same as the product information in the current return information, and both the current return information and the historical return information carry return reason information; Determine an expected demand index of the target user according to the current return information and the historical return information of the target user, where the expected demand index is used to represent the expected demand intensity of the target user for pre-purchased products; Obtain a purchase impulse model corresponding to the user profile of the target user, and correct the expected demand index based on the purchase impulse model to obtain a target demand index, where the target demand index is used to represent the actual demand intensity of the target user for pre-purchased products; When the actual demand intensity is greater than a intensity threshold, obtain the product information of candidate products associated with the pre-purchased products; Determine multiple target products from the candidate products according to the return reason information and the purchase impulse model, and form marketing strategy information; Market and display the multiple target products to the target user according to the marketing strategy information, where the marketing strategy information at least includes the marketing display order of the multiple target products.

[0005] In a possible implementation manner, the determining an expected demand index of the target user according to the current return information and the historical return information of the target user includes: Determine the order frequency and return reason similarity of the target user for the pre-purchased product based on the current return information and historical return information of the target user; Input the order frequency and return reason similarity of the target user for the pre-purchased product into the demand index prediction model to obtain the expected demand index of the target user, where the order frequency is positively correlated with the expected demand index, and the return reason similarity is positively correlated with the expected demand index.

[0006] In a possible implementation manner, the determining the return reason similarity of the target user for the pre-purchased product according to the current return information and historical return information of the target user includes: Perform text vectorization processing on the return reason information to obtain a return reason virtualization vector; Calculate the cosine value between every two return reason virtualization vectors and sum them to obtain a cosine similarity representation value; Input the cosine similarity representation value into the similarity regression model to obtain the return reason similarity, where the similarity regression model satisfies the following expression: ; Wherein, is the cosine similarity representation value, is the return reason virtualization vector, and a and b are control quantities for the size of the return reason similarity.

[0007] In a possible implementation manner, before obtaining the purchase impulse model corresponding to the user portrait of the target user, it further includes: Input the user identity information and historical purchased product information of the target user into the portrait generation model to obtain the user portrait of the target user; Construct the purchase impulse model of the target user according to the user portrait of the target user, where the purchase impulse model includes purchase ability information and purchase impulse information, and the purchase ability information is used to represent the payment ability of the target user, and the purchase impulse information is used to represent the immediate purchase willingness of the target user.

[0008] In a possible implementation manner, the correcting the expected demand index based on the purchase impulse model to obtain the target demand index includes: Obtain a first representation value corresponding to the purchase ability information and a second representation value corresponding to the purchase impulse information. The first representation value is used to represent the size of the payment ability of the target user, and the second representation value is used to represent the intensity of the immediate purchase willingness of the target user; Determine a correction index for the expected demand index according to the first representation value and the second representation value; Obtain the target demand index according to the expected demand index and the correction index.

[0009] In a possible implementation, determining a correction index for the expected demand index according to the first characterization value and the second characterization value includes: Obtaining the influence weights of the first characterization value and the second characterization value on the expected demand index respectively; Performing weighted summation on the first characterization value and the second characterization value to obtain a correction index for the expected demand index; Obtaining a target demand index according to the expected demand index and the correction index includes: When the correction index is greater than or equal to an index threshold, performing positive correction on the expected demand index; When the correction index is less than the index threshold, performing negative correction on the expected demand index.

[0010] In a possible implementation, the candidate products include substitute products of the pre-ordered product of the same category, and / or the candidate products include substitute products of the pre-ordered product of a different category. Obtaining product information of candidate products associated with the pre-ordered product includes: Obtaining substitute products of the pre-ordered product of the same category to obtain first candidate products, and obtaining substitute products of the pre-ordered product of a different category to obtain second candidate products, where the number of the first candidate products is greater than or equal to the number of the second candidate products; Extracting product information of the first candidate products and the second candidate products to obtain product information of candidate products associated with the pre-ordered product.

[0011] In a possible implementation, determining multiple target products from candidate products according to the return reason information and the purchase impulse model and forming marketing strategy information includes: Selecting products that match the return reason from candidate products according to the return reason information to obtain candidate target products; Selecting products that match the purchase impulse model from the candidate target products according to the purchase impulse model to obtain target products; Determining the marketing display order of the target products according to the return reason matching degree and the purchase impulse model matching degree; Forming marketing strategy information according to the marketing display order of the target products, and the marketing strategy information is used to market and display the multiple target products to the target users.

[0012] In a possible implementation, selecting products that match the return reason from candidate products according to the return reason information to obtain candidate target products includes: Classify the historical return reasons of candidate products into structured tags; For each candidate product, count the proportion of its historical returns that are the same as the current return reason label of the target user to obtain a similarity proportion value; Select candidate products with similarity proportion values less than a preset proportion threshold as candidate target products; Selecting a product that matches the purchase impulse model from the candidate target products according to the purchase impulse model to obtain a target product includes: Extract the purchase ability information of the target user from the purchasing power model and determine the price range of products that the user can afford; Traverse the candidate target products, obtain the price of each product, and include the products that fall within the price range of products that the user can afford in the target product set.

[0013] In a second aspect, an embodiment of the present application further provides a marketing management system, including: a memory and a processor, where the memory is used to store program code; the processor is used to call the program code to execute the method described in the first aspect.

[0014] Different from the prior art, a method for generating an artificial intelligence-based marketing strategy provided by an embodiment of the present application first obtains the current return information of the target user and the historical return information with the same product information, and both have return reasons; then determines the expected demand index of the target user for the pre-purchased product based on this return information; then corrects the expected demand index in combination with the purchase impulse model corresponding to the target user portrait to obtain a target demand index reflecting the actual demand intensity; when the actual demand intensity exceeds the threshold, obtain the candidate product information associated with the pre-purchased product; then determine multiple target products according to the return reason and the purchase impulse model and form marketing strategy information including the marketing display order; finally, display the target products to the target user according to this strategy. In this way, this technical solution can accurately analyze user needs, correct demand evaluation in combination with purchase impulse factors, improve the accuracy of judging the actual needs of users, and then formulate a marketing strategy based on the real needs of users, accurately recommend products, improve marketing effects, and enhance user satisfaction and purchase conversion rates. Description of the Drawings

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on the structures shown in these drawings without creative efforts.

[0016] Figure 1Schematic flowchart of the artificial intelligence-based marketing strategy generation method in some embodiments of the present application; Figure 2 Schematic flowchart of step S200 of the artificial intelligence-based marketing strategy generation method in some other embodiments of the present application; Figure 3 Schematic flowchart of step S500 of the artificial intelligence-based marketing strategy generation method in some other embodiments of the present application; Figure 4 Schematic diagram of the hardware structure of the marketing management system in some embodiments of the present application.

[0017] The realization of the object of the present invention, functional features and advantages will be further described in conjunction with the embodiments with reference to the accompanying drawings. Detailed implementation manners

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0019] It should be noted that all directional indications (such as up, down, left, right, front, back,...) in the embodiments of the present invention are only used to explain the relative positional relationship and movement conditions between components in a specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indications will also change accordingly.

[0020] In addition, the descriptions involving "first", "second", etc. in the present invention are only for descriptive purposes, and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of such features. In addition, "and / or" throughout the text includes three scenarios. Taking A and / or B as an example, it includes the technical solution of A, the technical solution of B, and the technical solution that both A and B are satisfied. In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the fact that those of ordinary skill in the art can implement it. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the protection scope required by the present invention.

[0021] With the rapid development of e-commerce, precision marketing strategies have become the key to enhancing user experience and corporate efficiency. Traditional marketing methods mainly rely on user behavior data or historical purchase records for product recommendations, but ignore the demand signals contained in users' return behavior. The existing technologies have the following problems: return data is only used for quality monitoring and not converted into demand insights, and the formulation of marketing strategies relies on empirical rules, unable to achieve personalized recommendations, resulting in low marketing efficiency.

[0022] It can be understood that when a user returns a product during a current purchase due to dissatisfaction with the purchase, in order to attract the user to shop on the platform again, the platform system needs to accurately grasp the user's real needs to accurately recommend products, improve marketing effectiveness, and enhance user satisfaction and purchase conversion rate. Based on this, the present application provides an artificial intelligence-based marketing strategy generation method. As Figures 1 - 3 shown, the following takes the marketing management system executing this artificial intelligence-based marketing strategy generation method as an example for illustration. It should be noted that although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here. Please refer to the appendix Figure 1 , this method includes the following steps S100 - step S600: Step S100, obtain the current return information of the target user and the historical return information associated with the current return information, where the historical return information is the same as the product information in the current return information, and both the current return information and the historical return information carry return reason information; Specifically, after the target user completes the current shopping return, the system obtains the current return information of the target user and the historical return information associated with the current return information. Association can refer to the association of return orders with the same product name or the same product use, and both the current return information and the historical return information carry return reason information.

[0023] For example, when a user returns a "water cup" purchased, the system traverses the return information of all products related to the water cup in the historical return information, especially the return information of all products related to the water cup in the historical return information within a recent period of time.

[0024] Exemplarily, if a certain user's currently returned product is a "water cup" and the return reason is "the color of the water cup does not match", and the system can find relevant information on 3 water cup returns within the last 1 month based on the current return information, and the relevant return reasons are "poor quality", "inconsistent size", etc.

[0025] Step S200, determine the expected demand index of the target user according to the current return information and the historical return information of the target user, where the expected demand index is used to represent the expected demand intensity of the target user for pre-purchased products; It can be understood that the intensity of the user's willingness to purchase a certain commodity or the intensity of the expected demand can be seen from the return information. For example, when the number of purchases and the frequency of returns of a certain commodity by the user are large, it indicates that the user really wants to buy the commodity, but due to various reasons, the user fails to select a commodity that meets the needs.

[0026] Therefore, in one embodiment, the step S200: determining the expected demand index of the target user according to the current return information and the historical return information of the target user includes: S210. Determine the order frequency and the similarity of return reasons of the target user for the pre-purchased commodity according to the current return information and the historical return information of the target user; S220. Input the order frequency and the similarity of return reasons of the target user for the pre-purchased commodity into the demand index prediction model to obtain the expected demand index of the target user, where the order frequency is positively correlated with the expected demand index, and the similarity of return reasons is positively correlated with the expected demand index.

[0027] Specifically, the system first counts the number of orders of the target user for the pre-purchased commodity (or similar commodities). The pre-purchased commodity refers to the commodity that the user has tried to purchase currently or historically but may return due to dissatisfaction. The order frequency reflects the degree of interest or the intensity of demand of the user for such commodities. If the user frequently places orders for the same type of commodity, it indicates that the user has a strong willingness to purchase such commodities. The system then analyzes the reasons for the target user's previous returns and calculates the similarity between these reasons. For example, it can be achieved through text analysis techniques, such as keyword matching, topic models (such as LDA) or semantic similarity calculation. The similarity of return reasons implies the continuous dissatisfaction of the user with certain aspects of the commodity, which also reflects from the side that the user has a continuous demand for such commodities but has not found a completely satisfactory choice.

[0028] The demand index prediction model is a pre-trained machine learning model used to predict the expected demand index according to the order frequency and the similarity of return reasons of the user. The model can be trained based on historical data, where the historical data includes the purchase behaviors, return records and corresponding demand intensity labels of a large number of users, etc. In the embodiment of the present application, the demand index prediction model outputs the expected demand index of the target user, and this index is a quantitative value used to characterize the expected demand intensity of the user for the pre-purchased commodity. In the model design of the embodiment of the present application, both the order frequency and the similarity of return reasons are positively correlated with the expected demand index. It shows that if the user frequently places orders and the return reasons are similar, the system will consider that the user has a high and continuous demand for such commodities.

[0029] In one embodiment, determining the similarity of the return reasons of the target user for the pre-purchased goods according to the current return information and historical return information of the target user includes: performing text vectorization processing on the return reason information to obtain a return reason virtualization vector; calculating the cosine value between every two return reason virtualization vectors and summing them to obtain a cosine similarity representation value; inputting the cosine similarity representation value into a similarity regression model to obtain the return reason similarity, where the similarity regression model satisfies the following expression: ; where is the cosine similarity representation value, is the return reason virtualization vector, and a and b are control quantities for the size of the return reason similarity.

[0030] Specifically, first, the word embedding technology in natural language processing (NLP), such as models like Word2Vec, GloVe, or BERT based on Transformer, can be used to convert the return reason information into vector form to obtain the return reason virtualization vector. After obtaining the return reason virtualization vector, for every two vectors and , the cosine similarity formula is used to calculate the cosine value between them, where is the vector dot product, and are the norms of the vectors respectively. The cosine values between all pairs of vectors are added to obtain the cosine similarity representation value . Finally, the cosine similarity representation value is input into the similarity regression model to obtain the return reason similarity. Among them, a and b are control quantities for the size of the return reason similarity and can be set according to actual needs.

[0031] Exemplarily, if all the return reason information is vectorized to obtain vectors , and , at this time, the cosine value between and , the cosine value between and , and the cosine value between and are summed to obtain the cosine similarity representation value, and finally, this cosine similarity representation value is substituted into the similarity regression model to obtain the return reason similarity.

[0032] It can be understood that the return reason information is vectorized to obtain a virtualized return reason vector, and the cosine values between every two virtualized return reason vectors are calculated and summed to obtain a cosine similarity representation value. If the cosine similarity representation value is closer to 0, it indicates that the return reasons are more similar; if the cosine similarity representation value is closer to 1, it indicates that the return reasons are less similar, that is, the difference in return reasons is greater.

[0033] Step S300: Obtain a purchase impulse model corresponding to the user profile of the target user, and correct the expected demand index based on the purchase impulse model to obtain a target demand index, where the target demand index is used to represent the actual demand intensity of the target user for the pre-purchased product. It can be understood that the obtained expected demand index only considers the return information of the user. Although the expected demand index can reflect the user's purchase intention for the pre-purchased product to a certain extent, it cannot truly reflect the user's active needs and passive demands.

[0034] Therefore, after obtaining the expected demand index of the target user, it is necessary to obtain a purchase impulse model corresponding to the user profile of the target user, and correct the expected demand index based on the purchase impulse model to obtain a target demand index. In this way, the target demand index combines the return information of the target user and the information of the purchase impulse model, and can more accurately reflect the actual demand intensity of the target user for the pre-purchased product. This is very important for merchants because it can help merchants more accurately predict user needs, thereby formulating more effective sales strategies and inventory plans.

[0035] In one embodiment, before obtaining the purchase impulse model corresponding to the user profile of the target user, it further includes: inputting the user identity information and historical purchased product information of the target user into a profile generation model to obtain the user profile of the target user; constructing the purchase impulse model of the target user according to the user profile of the target user, where the purchase impulse model includes purchase ability information and purchase impulse information, the purchase ability information is used to represent the payment ability of the target user, and the purchase impulse information is used to represent the immediate purchase intention of the target user.

[0036] Specifically, the user identity information of the target user may include the user's age, gender, occupation, geographical location, etc. The historical purchase information may include the user's past purchase records, including the types, quantities, prices, purchase times of the purchased goods, and the time spent on purchasing the goods, etc. A pre-trained portrait generation model is used to construct the user portrait of the target user. This model can generate the user portrait based on the user's identity information and historical purchase information. The portrait generation model can adopt machine learning or deep learning techniques to learn the characteristics and preferences of users by analyzing a large amount of user data. After obtaining the user portrait of the target user, machine learning algorithms such as logistic regression, decision tree, random forest, neural network, etc. can be used to construct the purchase impulse model. During the model training process, a large amount of data containing user portraits and corresponding purchase behavior labels is required for supervised learning.

[0037] Among them, the purchase impulse model includes purchase ability information and purchase impulse information. The purchase ability information is used to characterize the payment ability of the target user, and the purchase impulse information is used to characterize the immediate purchase willingness of the target user. For the convenience of calculation, the purchase ability information can be quantified by a numerical value, and the purchase impulse information can also be quantified by a numerical value. For example, the purchase ability information is quantified by a first characterization value, and the purchase impulse information is quantified by a second characterization value.

[0038] In one embodiment, the step of correcting the expected demand index based on the purchase impulse model to obtain the target demand index includes: obtaining the first characterization value corresponding to the purchase ability information and the second characterization value corresponding to the purchase impulse information. The first characterization value is used to characterize the magnitude of the payment ability of the target user, and the second characterization value is used to characterize the intensity of the immediate purchase willingness of the target user; determining a correction index for the expected demand index according to the first characterization value and the second characterization value; and obtaining the target demand index according to the expected demand index and the correction index.

[0039] Specifically, the first characterization value, the second characterization value, and the influence weights of the first characterization value and the second characterization value on the expected demand index can be obtained first. Then, the first characterization value and the second characterization value are weighted and summed to obtain a correction index for the expected demand index. The correction index synthesizes the payment ability and immediate purchase willingness of the user and is used to adjust the expected demand index. If the correction index is greater than or equal to the index threshold, it indicates that both the payment ability and immediate purchase willingness of the user are strong. Therefore, a positive correction of the expected demand index is required, that is, the value of the expected demand index is increased to reflect a stronger actual demand intensity. If the correction index is less than the index threshold, it indicates that the payment ability or immediate purchase willingness of the user is weak. Therefore, a negative correction of the expected demand index is required, that is, the value of the expected demand index is decreased to reflect a weaker actual demand intensity.

[0040] Step S400: When the actual demand intensity is greater than the intensity threshold, obtain the product information of candidate products associated with the pre-ordered product; After obtaining the target demand index, that is, after obtaining the actual demand intensity of the target user for the pre-ordered product, if the actual demand intensity is greater than the intensity threshold, it indicates that the user has a strong willingness to purchase the pre-ordered product. Based on this information, the system can take a series of subsequent actions to optimize the user experience, improve the sales conversion rate, or recommend relevant products. For example, first obtain the product information of candidate products associated with the pre-ordered product, and select the products that are more suitable for the target user from the candidate products as the target products, so as to accurately recommend the target products to the user.

[0041] In one embodiment, the candidate products may include similar alternative products of the pre-ordered product. For example, when the pre-ordered product is a single bed, the candidate products may be single beds, double beds, etc. The candidate products may also include non-similar alternative products of the pre-ordered product. For example, when the pre-ordered product is a single bed, the candidate products may be a folding sofa, etc., because in some cases the sofa can be used as a sleeping space.

[0042] The obtaining of the product information of candidate products associated with the pre-ordered product includes: obtaining similar alternative products of the pre-ordered product to get the first candidate products, and obtaining non-similar alternative products of the pre-ordered product to get the second candidate products, where the number of the first candidate products is greater than or equal to the number of the second candidate products; extracting the product information of the first candidate products and the second candidate products to obtain the product information of candidate products associated with the pre-ordered product.

[0043] Specifically, the system finds multiple alternative products similar to the pre-ordered product through a search algorithm or database query. The number of these products is usually set to be greater than or equal to the number of non-similar alternative products because similar products are more likely to meet the direct needs of users. The system also finds products that are not similar to the pre-ordered product but may have substitutability through a search algorithm or database query. The number of these products may be relatively small because their substitutability may be weaker or more dependent on the user's personal preferences and specific needs. For each type of candidate product (whether similar or non-similar), the system extracts its product information, which may include product name, price, picture, specification parameters, user evaluation, etc. These information will be integrated to form a set of product information of candidate products associated with the pre-ordered product, so as to provide sufficient data support for the selection of target products.

[0044] Step S500: Determine multiple target products from the candidate products according to the return reason information and the purchase impulse model, and form marketing strategy information; In one embodiment, step S500: determining multiple target products from the candidate products according to the return reason information and the purchase impulse model, and forming marketing strategy information, including: S510. Selecting products that match the return reason from the candidate products to obtain candidate target products according to the return reason information; S520. Selecting products that match the purchase impulse model from the candidate target products to obtain target products according to the purchase impulse model; S530. Determining the marketing display order of the target products according to the return reason matching degree and the purchase impulse model matching degree; S540. Forming marketing strategy information according to the marketing display order of the target products, where the marketing strategy information is used to market and display the multiple target products to the target user.

[0045] Specifically, those products that do not conflict with the user's return reason or can solve the user's return problem can be first screened out from the candidate products to obtain candidate target products. Then, those products that meet the user's purchase impulse model can be further screened out from the candidate target products to obtain target products. After obtaining the target products, the system can calculate the return reason matching degree and the purchase impulse model matching degree of each target product, and sort the target products according to these matching degrees. Other factors, such as the sales history, inventory situation, and price competitiveness of the products, can also be considered in the sorting. In this way, the system can more accurately identify the products that meet the user's needs and purchase impulse, and formulate corresponding marketing strategies, thereby improving the marketing effect and user satisfaction.

[0046] For example, when the return reason information of the target user is a return due to poor quality, then those products with a relatively low return rate due to quality can be screened out from the candidate products as candidate target products. Then, those products with a comparable price (within the acceptable range of the target user) can be further screened out from the candidate target products as the final target products.

[0047] In one embodiment, the selecting products that match the return reason from the candidate products to obtain candidate target products according to the return reason information includes: classifying the historical return reasons of the candidate products into structured tags; for each candidate product, counting the proportion of its historical returns with the same tags as the current return reason tag of the target user to obtain a similarity proportion value; and selecting the candidate products with a similarity proportion value less than a preset proportion threshold as candidate target products.

[0048] Specifically, the system can first collect the historical return reasons of candidate products, which may come from multiple users and multiple time periods. Then, the system performs text analysis on these return reasons to identify common return causes and classify them into structured tags. These tags may include "inappropriate size", "quality problem", "color mismatch", "dislike the style", etc. For each candidate product, the system traverses its historical return records and counts the number of records whose return reason tags are the same as the current return reason tags of the target user. Then, the system calculates the proportion of this number to the total number of returns of this product to obtain a similarity proportion value. This value reflects the matching degree of this product with the current return reason of the target user. Finally, select those products with similarity proportion values less than the preset proportion threshold as candidate target products. Since these products are not relevant or have less conflict with the return reasons of the target user, they are more likely to be accepted by the user. In this way, by recommending products that do not conflict with the target return reasons to the user, the system can increase the user's interest and willingness to purchase these products, thereby improving the sales conversion rate.

[0049] In one embodiment, the step of selecting a product that matches the purchase impulse model from the candidate target products according to the purchase impulse model to obtain a target product includes: extracting the purchase ability information of the target user from the purchasing power model to determine the price range of products that the user can afford; traversing the candidate target products, obtaining the price of each product, and including the products whose prices fall within the price range of products that the user can afford in the target product set.

[0050] Specifically, the system first accesses the purchasing power model and extracts the purchase ability information of the target user, which is usually manifested as the price range of products that the user can afford. This price range reflects the economic constraints and budget limitations of the user when purchasing products. Then the system traverses the list of candidate target products screened previously. For each product, the system obtains its current selling price, etc. Then the system compares the price of each candidate target product with the price range of products that the user can afford. If the price of the product falls within the user's price range, the system includes this product in the target product set. In this way, it is ensured that the selected target products not only do not conflict with the user's return reasons, but also meet the user's purchase ability and budget limitations.

[0051] In this way, by comprehensively considering the return reason information and purchasing power information of the target user to determine the target products, this application can provide more personalized product recommendations for users, improve the accuracy of recommendations and user satisfaction, thereby improving the sales conversion rate.

[0052] The return reason matching degree and the purchase impulse model matching degree can be the matching degrees in terms of return reasons and purchasing power. For example, the smaller the conflict between the historical returns of the target product and the user's return reasons, the higher the matching degree. The closer the product price of the target product is to the user's purchasing power, the higher the matching degree. After obtaining the target product and the corresponding matching degree, the marketing display ranking of the target product can be carried out according to the corresponding matching degree to obtain the marketing strategy information.

[0053] Step S600: Market and display the multiple target products to the target user according to the marketing strategy information, where the marketing strategy information at least includes the marketing display order of the multiple target products.

[0054] Specifically, after obtaining the marketing strategy information, the target products can be displayed according to the marketing display ranking information of the target products in the marketing strategy information. That is, the target product with the highest return reason matching degree and purchase impulse model matching degree is displayed at the front, so as to improve the sales conversion rate. In addition, technologies such as the payment system, short video fission, and NFC interaction can also be implemented in a three-terminal collaborative manner and implanted on various payment hardware devices to reduce the customer acquisition cost and improve the sales conversion rate.

[0055] Based on this, a method for generating an artificial intelligence-based marketing strategy provided by an embodiment of the present application first obtains the current return information of the target user and the historical return information with the same product information, and both carry return reasons; then determines the expected demand index of the target user for the pre-purchased product based on these return information; then corrects the expected demand index in combination with the purchase impulse model corresponding to the target user portrait to obtain the target demand index reflecting the actual demand intensity; when the actual demand intensity exceeds the threshold, obtains the candidate product information associated with the pre-purchased product; then determines multiple target products according to the return reasons and the purchase impulse model and forms the marketing strategy information including the marketing display order; finally, displays the target products to the target user according to this strategy. In this way, this technical solution can accurately analyze the user's needs, correct the demand assessment in combination with the purchase impulse factor, improve the accuracy of judging the user's actual needs, and then formulate a marketing strategy based on the user's true needs, accurately recommend products, improve the marketing effect, and enhance user satisfaction and purchase conversion rate.

[0056] As Figure 4 shown, Figure 4 is a schematic hardware structure diagram of a marketing management system in some embodiments of the present application. The marketing management system provided by the embodiment of the present application includes a memory 1000 and a processor 2000. Among them, the memory 1000 is used to store computer-readable instructions, and the processor 2000 is used to call the computer-readable instructions to execute the artificial intelligence-based marketing strategy generation method as described above.

[0057] Among them, the processor 2000 is used to provide computing and control capabilities to control the marketing management system to perform corresponding tasks. For example, it controls the marketing management system to execute the artificial intelligence-based marketing strategy generation method in any of the above method embodiments. The method includes: obtaining the current return information of the target user and the historical return information associated with the current return information, where the historical return information is the same as the commodity information in the current return information, and both the current return information and the historical return information carry return reason information; determining the expected demand index of the target user according to the current return information and the historical return information of the target user, where the expected demand index is used to characterize the expected demand intensity of the target user for the pre-purchased commodity; obtaining the purchase impulse model corresponding to the user profile of the target user, and correcting the expected demand index based on the purchase impulse model to obtain the target demand index, where the target demand index is used to characterize the actual demand intensity of the target user for the pre-purchased commodity; in the case where the actual demand intensity is greater than the intensity threshold, obtaining the commodity information of the candidate commodities associated with the pre-purchased commodity; determining a plurality of target commodities from the candidate commodities according to the return reason information and the purchase impulse model and forming marketing strategy information; marketing and presenting the plurality of target commodities to the target user according to the marketing strategy information, where the marketing strategy information at least includes the marketing presentation order of the plurality of target commodities.

[0058] The processor 2000 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), a hardware chip, or any combination thereof; it may also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The above PLD may be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof.

[0059] The memory 1000, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs, non-transitory computer-executable programs, and modules, such as the program instructions / modules corresponding to the artificial intelligence-based marketing strategy generation method in the embodiments of the present application. By running the non-transitory software programs, instructions, and modules stored in the memory 1000, the processor 2000 can implement the artificial intelligence-based marketing strategy generation method in any of the above method embodiments.

[0060] Specifically, the memory 1000 may include volatile memory (VM), such as random access memory (RAM); the memory 1000 may also include non-volatile memory (NVM), such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid-state drive (SSD), or other non-transitory solid-state storage devices; the memory 1000 may further include a combination of the above types of memories.

[0061] In summary, the marketing management system of the present application adopts the technical solution of any one of the above embodiments of the artificial intelligence-based marketing strategy generation method. Therefore, it at least has the beneficial effects brought by the technical solutions of the above embodiments, which will not be elaborated here one by one.

[0062] The embodiments of the present application also provide a computer-readable storage medium, such as a memory including program codes, and the above program codes can be executed by a processor to complete the artificial intelligence-based marketing strategy generation method in the above embodiments. For example, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CDROM), a magnetic tape, a floppy disk, and an optical data storage device, etc.

[0063] The embodiments of the present application also provide a computer program product, which includes one or more program codes, and the program codes are stored in a computer-readable storage medium. The processor of the warning system reads the program codes from the computer-readable storage medium, and the processor executes the program codes to complete the steps of the artificial intelligence-based marketing strategy generation method provided in the above embodiments.

[0064] Those of ordinary skill in the art can understand that all or part of the steps to implement the above embodiments can be completed by hardware, or can be completed by hardware related to program code. The program can be stored in a computer-readable storage medium, and the storage medium mentioned above can be a read-only memory, a magnetic disk, an optical disk, etc.

[0065] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0066] Through the description of the above embodiments, those of ordinary skill in the art can clearly understand that each embodiment can be implemented by means of software plus a general hardware platform, and of course, it can also be implemented by hardware. Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0067] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention. Any equivalent structural transformation made by using the description and drawings of the present invention under the inventive concept of the present invention, or direct / indirect application in other related technical fields, is included in the patent protection scope of the present invention.

Claims

1. A marketing strategy generation method based on artificial intelligence, applied to a marketing management system, characterized in that: The method comprises: Acquire the current return information of the target user and the historical return information associated with the current return information, wherein the historical return information is the same as the commodity information in the current return information, and both the current return information and the historical return information carry the return reason information; Determining an expected demand index of the target user according to the current return information and historical return information of the target user, wherein the expected demand index is used to characterize the expected demand intensity of the target user for the pre-ordered goods; Acquire a purchase impulse model corresponding to the user profile of the target user, and correct the expected demand index based on the purchase impulse model to obtain a target demand index, where the target demand index is used to characterize the actual demand intensity of the target user for the pre-purchased goods; When the actual demand intensity is greater than an intensity threshold, obtaining product information of a candidate product associated with the pre-ordered product; Determine multiple target commodities from the candidate commodities according to the return reason information and the purchase impulse model and form marketing strategy information; The plurality of target commodities are marketed and presented to the target user according to the marketing strategy information, wherein the marketing strategy information at least includes a marketing presentation order of the plurality of target commodities.

2. The method for generating marketing strategies based on artificial intelligence according to claim 1, characterized in that: The determining the expected demand index of the target user according to the current return information and the historical return information of the target user includes: Determine the target user's order frequency for the pre-ordered product and the similarity of the return reasons based on the target user's current return information and historical return information; The target user's order frequency for pre-ordered goods and the similarity of return reasons are input into the demand index estimation model to obtain the target user's expected demand index, wherein the order frequency is positively correlated with the expected demand index, and the similarity of return reasons is positively correlated with the expected demand index.

3. The method for generating marketing strategies based on artificial intelligence according to claim 2, characterized in that: The determining the similarity of the target user's return reasons for the pre-purchased goods according to the target user's current return information and historical return information includes: Performing text vectorization processing on the return reason information to obtain a return reason virtualized vector; Calculate the cosine value between each two return reason virtual vectors and sum them up to obtain the cosine similarity representation value; The cosine similarity representation value is input into the similarity regression model to obtain the return reason similarity, wherein the similarity regression model satisfies the following expression: ； in, is the cosine similarity representation value, is the virtualized vector of the return reason, and a and b are the control values ​​of the similarity of the return reason.

4. The method for generating marketing strategies based on artificial intelligence according to claim 1, characterized in that: Before acquiring the purchase impulse model corresponding to the user portrait of the target user, the method further includes: Inputting the user identity information and historical purchase information of the target user into a portrait generation model to obtain a user portrait of the target user; A purchase impulse model of the target user is constructed according to the user portrait of the target user, wherein the purchase impulse model includes purchasing power information and purchasing impulse information, the purchasing power information is used to characterize the payment ability of the target user, and the purchasing impulse information is used to characterize the immediate purchasing intention of the target user.

5. The method for generating marketing strategies based on artificial intelligence according to claim 4, characterized in that: The step of correcting the expected demand index based on the purchase impulse model to obtain a target demand index includes: Obtaining a first characterization value corresponding to the purchasing power information and a second characterization value corresponding to the purchasing impulse information, wherein the first characterization value is used to characterize the payment power of the target user, and the second characterization value is used to characterize the intensity of the target user's immediate purchasing intention; Determining a correction index for the expected demand index according to the first characterization value and the second characterization value; A target demand index is obtained according to the expected demand index and the correction index.

6. The method for generating marketing strategies based on artificial intelligence according to claim 5, characterized in that: The determining a correction index for the expected demand index according to the first characterization value and the second characterization value includes: Obtaining the influence weights of the first characterization value and the second characterization value on the expected demand index respectively; Performing a weighted summation on the first characterization value and the second characterization value to obtain a correction index for the expected demand index; The obtaining of a target demand index according to the expected demand index and the correction index comprises: When the correction index is greater than or equal to the index threshold, performing a positive correction on the expected demand index; When the correction index is less than the index threshold, the expected demand index is reversely corrected.

7. The method for generating marketing strategies based on artificial intelligence according to claim 1, characterized in that: The candidate commodities include similar substitute commodities of the pre-ordered commodities, and / or the candidate commodities include non-similar substitute commodities of the pre-ordered commodities, and the obtaining of commodity information of the candidate commodities associated with the pre-ordered commodities includes: Acquire similar alternative commodities of the pre-ordered commodity to obtain first candidate commodities, and acquire non-similar alternative commodities of the pre-ordered commodity to obtain second candidate commodities, wherein the quantity of the first candidate commodities is greater than or equal to the quantity of the second candidate commodities; The product information of the first candidate product and the second candidate product is extracted to obtain the product information of the candidate product associated with the pre-order product.

8. The method for generating marketing strategies based on artificial intelligence according to claim 7, characterized in that: The step of determining a plurality of target commodities from the candidate commodities according to the return reason information and the purchase impulse model and forming marketing strategy information includes: According to the return reason information, select a product that matches the return reason from the candidate products to obtain a candidate target product; Selecting a product that matches the purchase impulse model from the candidate target products according to the purchase impulse model to obtain a target product; Determining the marketing display order of the target product according to the matching degree of the return reason and the matching degree of the purchase impulse model; The marketing strategy information is formed according to the marketing display sequence of the target commodities, and the marketing strategy information is used for marketing and displaying the multiple target commodities to the target users.

9. The method for generating marketing strategies based on artificial intelligence according to claim 8, characterized in that: The selecting a commodity matching the reason for return from candidate commodities according to the reason for return to obtain a candidate target commodity includes: Divide the historical return reasons of candidate products into structured labels; For each candidate product, the proportion of its historical returns with the same return reason label as the target user's current return reason label is counted to obtain a similarity ratio value; Select candidate products whose similarity ratio values ​​are less than a preset ratio threshold as candidate target products; The step of selecting a commodity matching the purchase impulse model from the candidate target commodities according to the purchase impulse model to obtain a target commodity comprises: Extract the purchasing power information of target users from the purchasing power model and determine the price range of goods that users can afford; Traverse the candidate target products, obtain the price of each product, and add the products that fall into the price range that the user can afford to the target product set.

10. A marketing management system, characterized in that: include: A memory and a processor, wherein the memory is used to store program codes; The processor is used to call the program code to execute the method according to any one of claims 1 to 9.

Images