Intelligent advertisement content generation and optimization system based on big data

Abstract

The invention relates to the technical field of advertisement marketing, and discloses a big data-based intelligent advertisement content generation and optimization system, which comprises a customer data acquisition module, a competitive product advertisement acquisition module, a competitive product advertisement analysis module, an intelligent generation and prediction module, a content optimization module and an interaction module. The competitive product advertisement acquisition module and the competitive product advertisement analysis module are arranged, so that competitive products of products can be selected, corresponding competitive products can be selected according to different competitive product standards, a foundation is laid for follow-up competitive product advertisement analysis, and accuracy is improved; by analyzing the advertisement data of the competitive product, the relationship between the advertisement and the user is obtained, if the advertisement effect is good, it shows that the user has higher favorite degree and acceptance degree for the advertisement, so that information of the user does not need to be collected, only big data collection and analysis need to be carried out on the advertisement data, and privacy protection of the user is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of advertising marketing, and more particularly to an intelligent advertising content generation and optimization system based on big data. Background Art

[0002] With the rapid development of the Internet, the advertising industry is also constantly changing. The traditional advertising production method requires a large amount of time and manpower, and it is difficult to guarantee the advertising effect. Therefore, how to improve the efficiency and effect of advertising production has become an urgent problem to be solved in the advertising industry. The emergence of big data technology has brought new opportunities to the advertising industry. By analyzing user data, more targeted advertising content can be generated to improve the advertising effect.

[0003] A disclosed document with the publication number CN117541322B discloses an intelligent advertising content generation method and system based on big data analysis, which relates to the technical field of advertising marketing. It includes obtaining advertising target data, obtaining advertising target clustering information based on data processing, obtaining target preference feature information according to the advertising target clustering information, and performing intelligent generation of advertising content according to the advertising content generation information; analyzing the advertising target audience through hierarchical clustering, accurately judging user feature information related to the advertising effect, precisely grasping the needs and preferences of the target audience, accurately generating targeted advertising content by calculating the advertising matching index, better meeting the needs of users, providing more attractive and effective information, enhancing users' recognition and acceptance of the advertisement, evaluating the advertising effect through advertising effect data, obtaining the advertising effect evaluation index, and timely adjusting and optimizing the advertising content to improve the effect and benefit of the advertisement.

[0004] However, when using big data to obtain user feature information, it includes users' personal sensitive information. For example, the user behavior data and user social media data obtained by the disclosed document using big data are likely to infringe on users' personal privacy; at the same time, when performing intelligent generation of advertising content, if all rely on artificial intelligence for generation, the advertising content may contain some harmful content and negative viewpoints, bringing security, data, and ethical risk problems. Therefore, while using artificial intelligence for intelligent generation of advertisements, it is also necessary to identify and optimize the advertising content so that the advertising content is more optimized, rather than only optimizing the advertising effect.

[0005] In view of this, the present invention proposes an intelligent advertising content generation and optimization system based on big data, which avoids the problem of user privacy for intelligent generation of advertising content, and at the same time identifies and optimizes the generated advertising content to make the advertising content more high-quality. Summary of the Invention

[0006] To overcome the above defects of the prior art, the present invention provides an intelligent advertising content generation and optimization system based on big data to solve the problems existing in the above background art.

[0007] The present invention provides the following technical solutions: An intelligent advertising content generation and optimization system based on big data, comprising a customer data collection module, a competitor advertisement acquisition module, a competitor advertisement analysis module, an intelligent generation and prediction module, a content optimization module, and an interaction module; The customer data collection module is used to collect customer demand data and product data, and transmit them to the competitor advertisement acquisition module and the competitor advertisement analysis module; The competitor advertisement acquisition module is used to receive the data of the customer data collection module, obtain competitor advertisement data, and divide it into comparison advertisement data and numerical advertisement data and then transmit them to the competitor advertisement analysis module; The competitor advertisement analysis module is used to receive the data of the competitor advertisement acquisition module and the customer data collection module, analyze the numerical advertisement data and the comparison advertisement data, and obtain the advantageous items and mandatory items; The intelligent generation and prediction module is based on the advantageous items and mandatory items to intelligently generate advertisements, then inputs the advertisement prediction model to predict the advertising placement effect, selects the advertisement with the best placement effect as the alternative advertisement, and then transmits the alternative advertisement and the predicted placement effect to the content optimization module; The content optimization module outputs the alternative advertisement and the predicted placement effect to the interaction module for display. The customer modifies the advertisement content through the content optimization module. After the modification is completed, the modified advertisement is fed back to the intelligent generation and prediction module for placement effect prediction, and then the prediction result is transmitted back to the content optimization module; If the customer does not need to modify the advertisement content, the alternative advertisement is directly used as the finally generated advertisement and transmitted to the interaction module for display; The interaction module is used to display the finally generated advertisement on the human-computer interaction interface.

[0008] Preferably, the customer demand data includes the advertisement content expected to be presented by the customer, copywriting element data, and visual element data; the advertisement content includes product promotion content, service promotion content, brand image content, and promotional activity content; the product promotion content includes product feature introduction, product usage method, comparison of product advantages with competing products, and product price, the service promotion content includes service content, service process and timeliness, service advantages and features, and customer evaluations, the brand image content includes brand story and history, brand values and concepts, brand partners and honors, and brand public welfare activities and responsibilities, the promotional activities include limited-time discounts and offers, holiday promotional activities, points redemption and gift giving, and joint promotions and partners; the copywriting element data includes copywriting content, copywriting font, font format, and font layout; the visual element data includes picture color, picture content, animation content, and video content; the product data is data related to the product for which advertisement is carried out, including product name, product picture, product function, product price, product sales volume, product category, and product historical data; the product historical data includes product historical sales quantity, sales amount, sales duration, and product historical advertisement data, the advertisement data includes advertisement content data, advertisement conversion rate, advertisement click-through rate, and advertisement exposure rate; the product is the product for which advertisement is carried out; the advertisement content data includes the copywriting elements in the advertisement content and the visual elements in the advertisement content.

[0009] Preferably, the specific manner in which the competing product advertisement acquisition module acquires competing product advertisement data is as follows: Extract and fuse the product data features, and the feature fusion adopts linear fusion. Use a linear combination method to fuse different features, and obtain new fused features through weight assignment. Take the new fused features as the fused product data features; Use big data technology to collect N products of the same category as the product for which advertisement is carried out, extract the product data features of the N products and then fuse them to obtain the fused product data features. Calculate the similarity between the fused product data features and the fused product data features, and obtain the products corresponding to the top n fused product data features with the highest similarity as competing products; Take the most recent advertisement data of the n competing products as the competing product advertisement data, that is, there are n competing product advertisement data.

[0010] Preferably, the steps for the competing product advertisement acquisition module to divide the competing product advertisement data into comparison advertisement data and numerical advertisement data are as follows: Step S01: Encode the n competing product advertisement data to obtain chromosomes and construct an initial population; Step S02: Determine the fitness function; Step S03: Conduct natural selection on the chromosomes in the population; Step S04: Conduct crossover recombination on the chromosomes in the population; Step S05: Mutate the chromosomes in the population; Step S06: Obtain a new population. Preset the population generation number as L and the fitness threshold as Q, where L is an integer greater than 0 and Q is a real number greater than 0. Loop through Step S03 - Step S05 until the generation number corresponding to the new population is L or there exists a chromosome in the new population whose corresponding fitness is greater than or equal to the fitness threshold Q. When the loop ends, use the competing product advertisement data corresponding to the chromosome with the maximum fitness in the new population as the optimal advertisement data; Use the competing product advertisement data corresponding to the optimal advertisement data as the comparison advertisement data, and use the remaining n - 1 competing product advertisement data as the numerical advertisement data.

[0011] Preferably, in Step S01, each competing product advertisement data is encoded as A, and A is the chromosome. Randomly generate B chromosomes to form the initial population C, , where C b is the b-th chromosome, and b = 1, 2, 3, …, B; The fitness function is expressed as: ; where f b is the fitness corresponding to the b-th chromosome, GX b is the advertisement effect of the competing product advertisement corresponding to the competing product advertisement data of the b-th chromosome; , where SR b _gg is the advertisement revenue brought by the competing product advertisement corresponding to the competing product advertisement data of the b-th chromosome, and CB b _gg is the advertisement cost of the competing product advertisement corresponding to the competing product advertisement data of the b-th chromosome.

[0012] Preferably, the cosine similarity is used to calculate the similarity between the fused commodity data features and the fused product data features, and the formula is expressed as: ; where cosθ i is the similarity between the i-th fused commodity data feature and the fused product data feature, RH i _sp is the i-th fused commodity data feature vector, RH_cp is the fused product data feature vector, ‖RH i _sp‖ is the norm of the i-th fused commodity data feature vector, ‖RH_cp‖ is the norm of the fused product data feature vector, “∙” is the dot product, and i = 1, 2, 3, …, N.

[0013] Preferably, the method by which the competing product advertisement analysis module obtains the advantageous items and required items is as follows: Calculate the similarity between the copywriting elements and visual elements in the numerical advertisement data and the comparative advertisement data. Both the copywriting elements and visual elements contain copywriting factors and visual factors. All the data items included in the copywriting element data are copywriting factors, and all the data items included in the visual element data are visual factors. Combine the copywriting elements and visual elements into an element set. Both the copywriting factors and visual factors are factors in the element set. Then: The comparative advertisement data is expressed as: , where U D is the element set of the comparative advertisement data, D j is the data feature of the j-th factor in the element set of the comparative advertisement data, and m is the total number of factors in the element set, that is, the sum of the number of copywriting factors and visual factors; The numerical advertisement data is expressed as: , where U s_a is the a-th numerical advertisement data element set, s_a j is the data feature of the j-th factor in the a-th numerical advertisement data element set, and a = 1, 2, 3, …, n - 1; The factors in the element set of the comparative advertisement data and the element set of the numerical advertisement data correspond one by one; Calculate the similarity between all the factor data features in each numerical advertisement data element set and all the corresponding factor data features in the comparative advertisement data element set. Set a screening threshold YU1. If the similarity does not reach the screening threshold, mark this factor in the comparative advertisement data element set as an advantageous item; The required item is a factor that must exist in the generated advertisement content.

[0014] Preferably, the intelligent generation and prediction module generates P initial advertisements, and sequentially inputs the P initial advertisements into the constructed advertisement prediction model to predict the corresponding delivery effects; the input layer of the advertisement prediction model has P nodes for inputting the P initial advertisements, and the output layer also has P nodes for outputting the delivery effects of the P initial advertisements; select the initial advertisement with the best predicted advertisement delivery effect as the alternative advertisement.

[0015] Preferably, the training process of the advertisement prediction model is specifically as follows: Pre-collect R1 groups of advertisement content data of competing products and R2 groups of historical advertisement content data of the product, and use them as analysis data. Then there are a total of (R1 + R2) groups of analysis data. Both R1 and R2 are integers greater than 1. Convert a group of analysis data and its corresponding delivery effect into a corresponding group of feature vectors; Use each set of feature vectors as the input of the advertisement prediction model. The advertisement prediction model outputs a set of delivery effects corresponding to each set of analysis data, aims at the actual delivery effect corresponding to each set of analysis data, where the actual delivery effect is the pre-collected delivery effect corresponding to the analysis data; uses minimizing the sum of prediction errors of all analysis data as the training objective; the formula of the prediction error is expressed as: , where ε q is the prediction error, q is the group number of the feature vectors corresponding to the analysis data, θ q is the delivery effect corresponding to the q-th group of analysis data, μ q is the actual delivery effect corresponding to the q-th group of analysis data. Train the advertisement prediction model until the sum of prediction errors converges and then stop training; The advertisement prediction model is a deep neural network model; it includes an input layer, a hidden layer, and an output layer; each hidden layer includes multiple neurons, and there are connections between each neuron and the neurons in the next layer, and the connections contain weights.

[0016] Preferably, the content optimization module includes an advertisement modification unit and an element bar unit; the advertisement modification unit is used for customers to modify the alternative advertisement content. After the alternative advertisement is displayed on the operation terminal, the customer performs a modification operation on the alternative advertisement. In the advertisement content of the alternative advertisement, each copywriting element and visual element is a separate modification item, and the customer clicks on the modification item to perform a replacement or modification operation; the element bar unit is used to display the copywriting elements and visual elements for replacement and modification; After the customer finishes the modification, save the modified advertisement and send it back to the intelligent generation and prediction module, input it into the advertisement prediction model, predict the delivery effect of the modified advertisement, compare the delivery effect of the modified advertisement with the delivery effect of the alternative advertisement. If the delivery effect of the modified advertisement is better, directly replace the alternative advertisement with the modified advertisement and transmit it to the content optimization module again; if the delivery effect of the alternative advertisement is better, do not replace the alternative advertisement, and at the same time feedback the delivery effect of the modified advertisement to the interaction module for display, and conduct a second inquiry in the interaction module, and the inquiry content is whether to replace the alternative advertisement.

[0017] The technical effects and advantages of the present invention: (1) By providing a competing product advertisement acquisition module and a competing product advertisement analysis module, the present invention facilitates the selection of competing products for a product, selects corresponding competing products according to different competing product criteria, lays a foundation for subsequent analysis of competing product advertisements, and improves accuracy. By analyzing the advertisement data of competing products, the relationship between advertisements and users can be obtained indirectly. If the advertisement effect is good, it indicates that users have a higher degree of preference and acceptance for the advertisement. Therefore, there is no need to collect user information, and only big data collection and analysis of advertisement data are required, which ensures the protection of user privacy. At the same time, corresponding advantageous items are obtained from competing product advertisements, thereby improving the content quality and delivery effect of product advertisements.

[0018] (2) By providing an intelligent generation and prediction module, the present invention facilitates predicting the delivery effect of advertisements by constructing an advertisement prediction model. By integrating a trained neural network model, the non-linear and multi-modal data modeling capabilities of the neural network are fully utilized to effectively capture complex data relationships. According to the parallel computing processing mechanism, the computing efficiency is improved, thereby ensuring the comprehensiveness and accuracy of the analysis. Furthermore, it can ensure the best delivery effect of the intelligently generated advertisements. Brief Description of the Drawings

[0019] Figure 1 It is a structural diagram of an intelligent advertisement content generation and optimization system based on big data according to the present invention. Detailed Embodiments

[0020] Next, the technical solutions in the present invention will be clearly and completely described in conjunction with the accompanying drawings in the present invention. In addition, the forms of each structure described in the following embodiments are merely examples, and an intelligent advertisement content generation and optimization system based on big data involved in the present invention is not limited to the structures described in the following embodiments. All other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0021] As Figure 1 shown, the present invention provides an intelligent advertisement content generation and optimization system based on big data, including a customer data acquisition module, a competing product advertisement acquisition module, a competing product advertisement analysis module, an intelligent generation and prediction module, a content optimization module, and an interaction module; The customer data acquisition module is used to collect customer demand data and product data and transmit them to the competing product advertisement acquisition module and the competing product advertisement analysis module; The competing product advertisement acquisition module is used to receive the data of the customer data acquisition module, obtain competing product advertisement data, and divide it into comparison advertisement data and numerical advertisement data and then transmit it to the competing product advertisement analysis module; The competing product advertisement analysis module is used to receive the data from the competing product advertisement acquisition module and the customer data collection module, analyze the numerical advertisement data and the comparison advertisement data, and obtain the advantageous items and the mandatory items; The intelligent generation and prediction module, based on the advantageous items and the mandatory items, intelligently generates advertisements, then inputs them into the advertisement prediction model to predict the advertisement placement effect, selects the advertisement with the best placement effect as the alternative advertisement, and then transmits the alternative advertisement and the predicted placement effect to the content optimization module; The content optimization module outputs the alternative advertisement and the predicted placement effect to the interaction module for display. The customer modifies the advertisement content through the content optimization module. After the modification is completed, the modified advertisement is sent back to the intelligent generation and prediction module for placement effect prediction, and then the prediction result is transmitted to the content optimization module. If the customer does not need to modify the advertisement content, the alternative advertisement is directly used as the finally generated advertisement and transmitted to the interaction module for display; The interaction module is used to display the finally generated advertisement on the human-computer interaction interface.

[0022] In this embodiment, it should be specifically noted that the customer demand data includes the advertisement content, copywriting element data, and visual element data expected to be presented by the customer; the advertisement content includes product promotion content, service promotion content, brand image content, and promotional activity content, etc.; the product promotion content includes but is not limited to product feature introduction, product usage method, product advantages compared with competing products, and product price, etc., the service promotion content includes but is not limited to service content, service process and timeliness, service advantages and features, and customer evaluations, etc., the brand image content includes but is not limited to brand story and history, brand values and concepts, brand partners and honors, and brand public welfare activities and responsibilities, etc., the promotional activities include but are not limited to limited-time discounts and promotions, holiday promotional activities, points redemption and gift giving, and joint promotions and partners, etc.; the copywriting element data includes but is not limited to copywriting content, copywriting font, font format, and font layout, etc.; the visual element data includes but is not limited to relevant data such as picture color, picture content, animation content, and video content; the product data is data related to the product for which the advertisement is being promoted, including but not limited to product name, product picture, product function, product price, product sales volume, product category, and product historical data; the product historical data includes but is not limited to product historical sales quantity, sales amount, sales duration, and product historical advertisement data, the advertisement data includes but is not limited to advertisement content data, advertisement conversion rate, advertisement click-through rate, and advertisement exposure rate, etc.; the product is the product for which the advertisement is being promoted; the advertisement content data includes but is not limited to the copywriting elements in the advertisement content and the visual elements in the advertisement content, etc.; By asking customers and filling out an electronic questionnaire related to customer demand data, obtain the advertising content, copywriting elements, and visual elements that the customer wants to present; obtain product data through product brochures and product profiles provided by the customer, and use big data technology to obtain the historical data of the product.

[0023] In this embodiment, it should be specifically noted that the specific method for the competing product advertisement acquisition module to obtain competing product advertisement data is as follows: Extract and fuse the product data features. The feature fusion adopts linear fusion, and different features are fused in a linear combination manner. New fused features are obtained through weight assignment, and the new fused features are used as the fused product data features. Use big data technology to collect N products of the same product category as the product being advertised. After extracting the product data features of the N products and fusing them, obtain the fused product data features. Calculate the similarity between the fused product data features and the fused product data features, and obtain the products corresponding to the top n fused product data features with the highest similarity as competing products; the product category includes, but is not limited to, clothing category, food category, beauty category, furniture category, etc.; the method for fusing product data features is the same as the method for fusing product data features, both adopting linear fusion and with the same weight assignment. The method for extracting product data features is the same as the method for extracting product data features. After preprocessing the data, methods such as bag-of-words model, word embedding, and TF-IDF are used to extract text data features, and methods such as color histogram, edge detection, and convolutional neural network model are used to extract image data features. Take the most recent advertisement data of the n competing products as the competing product advertisement data, that is, there are n competing product advertisement data. The weight can be set by those skilled in the art or the customer themselves. If the competing product focuses on selecting products similar to the product function and factors such as product price are not emphasized, then when assigning weights, the weight value corresponding to the data feature of the product function is larger. At this time, more emphasis is placed on considering whether the function of the product is similar to the product, and the impact of price on the selection of competing products is relatively small; if the competing product focuses on selecting products similar to the product price and factors such as product function are not emphasized, then when assigning weights, the weight value corresponding to the data feature of the product price is larger. At this time, more emphasis is placed on considering whether the price of the product is similar to the product price. If the price difference is too large, the possibility of considering this product as a competing product is relatively small.

[0024] In this embodiment, it should be specifically noted that the steps for the competing product advertisement acquisition module to divide the competing product advertisement data into comparison advertisement data and numerical advertisement data are as follows: Step S01: Encode the n competing product advertisement data to obtain chromosomes and construct an initial population. Step S02: Determine the fitness function; Step S03: Conduct natural selection on the chromosomes in the population; Step S04: Conduct crossover recombination on the chromosomes in the population; Step S05: Mutate the chromosomes in the population; Step S06: Obtain a new population. The preset population generation number is L, and the fitness threshold is Q. L is an integer greater than 0, and Q is a real number greater than 0. Loop through steps S03 - S05 until the generation number corresponding to the new population is L or there is a chromosome in the new population whose corresponding fitness is greater than or equal to the fitness threshold Q. When the loop ends, use the competitive product advertisement data corresponding to the chromosome with the maximum fitness in the new population as the optimal advertisement data. Exemplarily, if the preset population generation number is 1, then conduct natural selection, crossover recombination on the chromosomes in the initial population, and obtain a new population. At this time, the generation number corresponding to the new population is 1, so the loop ends; Use the competitive product advertisement data corresponding to the optimal advertisement data as the comparison advertisement data, and use the remaining n - 1 competitive product advertisement data as the numerical advertisement data; Each competitive product advertisement data is encoded as A, and A is the chromosome. Randomly generate B chromosomes to form the initial population C, , where C b is the b-th chromosome, b = 1, 2, 3,..., B; The fitness function is expressed as: ; where f b is the fitness corresponding to the b-th chromosome, and GX b is the advertisement effect of the competitive product advertisement corresponding to the competitive product advertisement data of the b-th chromosome; , where SR b _gg is the advertisement revenue brought by the competitive product advertisement corresponding to the competitive product advertisement data of the b-th chromosome, and CB b _gg is the advertisement cost of the competitive product advertisement corresponding to the competitive product advertisement data of the b-th chromosome. The advertisement effect can also be obtained through advertisement click-through rate, advertisement conversion rate, and customer acquisition cost, etc. The customer acquisition cost is expressed as the total advertisement expenditure divided by the number of new customers obtained, that is, the average cost required to obtain one new customer. The advertisement click-through rate is expressed as the number of advertisement clicks divided by the number of advertisement displays, which is used to measure the ability of the advertisement to attract the target audience to click. The advertisement conversion rate is expressed as the number of users who achieve conversion divided by the number of advertisement clicks, which is used to measure the proportion of users who complete specific actions after clicking the advertisement. The specific actions include but are not limited to purchase, registration, and download, etc.

[0025] In this embodiment, it should be specifically noted that the similarity between the fused product data features and the fused commodity data features can be calculated by appropriate methods such as cosine similarity or Jaccard similarity coefficient that can obtain the data feature similarity. In this embodiment, cosine similarity is used for calculation, and the formula is expressed as: ; where cosθ i is the similarity between the i-th fused commodity data feature and the fused product data feature, RH i _sp is the i-th fused commodity data feature vector, RH_cp is the fused product data feature vector, ‖RH i _sp‖ is the norm of the i-th fused commodity data feature vector, ‖RH_cp‖ is the norm of the fused product data feature vector, "∙" is the dot product, and i = 1, 2, 3,..., N; Sort the similarities between each fused commodity data feature and the fused product data feature. Since the value range of cosine similarity is [-1, 1], the closer to 1, the higher the similarity and the higher the ranking. Take the top n fused commodity data features, and use the corresponding commodities as competing products.

[0026] In this embodiment, it should be specifically noted that the method for the competing product advertisement analysis module to obtain the advantageous items and required items is as follows: Calculate the similarity between the copywriting elements and visual elements in the numerical advertisement data and the comparison advertisement data. Both the copywriting elements and visual elements contain several copywriting factors and visual factors. All the data items included in the copywriting element data are copywriting factors, that is, the copywriting content, copywriting font, font format, etc. are all copywriting factors in the copywriting elements. Similarly, all the data items included in the visual element data are visual factors, that is, the picture color, picture content, animation content, and video content, etc. are all visual factors in the visual elements; Combine the copywriting elements and visual elements into an element set. Both the copywriting factors and visual factors are factors in the element set, then: The comparison advertisement data is expressed as: , where U D is the comparison advertisement data element set, D j is the data feature of the j-th factor in the comparison advertisement data element set, and m is the total number of factors in the element set, that is, the sum of the number of copywriting factors and visual factors; The numerical advertisement data is expressed as: , where U s_a is the a-th numerical advertisement data element set, s_a j is the data feature of the j-th factor in the a-th numerical advertisement data element set, and a = 1, 2, 3,..., n - 1; The factor in the comparative advertisement data element set corresponds one-to-one with the factor in the numerical advertisement data element set; Calculate the similarity between all factor data features in each numerical advertisement data element set and all corresponding factor data features in the comparative advertisement data element set. The similarity calculation can be performed by those skilled in the art using a suitable data feature similarity calculation method, or still use the cosine similarity for calculation; for visual factors, key frames and audio features of animations or videos can be extracted to obtain the corresponding data features as the data features of visual factors; set a screening threshold YU1. If the similarity does not reach the screening threshold, mark the factor in the comparative advertisement data element set as a dominant item. The screening threshold YU1 can be set by those skilled in the art themselves, and the value range satisfies YU1 ∈ (-1, 1); for example, if the factor that does not reach the screening threshold in the first numerical advertisement data element set and the comparative advertisement data element set is the y1 factor, then mark the y1 factor as a dominant item; the dominant item is the factor in the comparative advertisement data that has a large difference from the numerical advertisement data. Therefore, most of the reasons for the better delivery effect of the comparative advertisement data than the numerical advertisement data are due to the existence of the dominant item, rather than the common factors with too high similarity; therefore, selecting the dominant item is beneficial to bringing the factors of the dominant item into the product advertisement content and utilizing the advantages of the existing dominant item to improve the delivery effect of the product advertisement content; The mandatory item is the factor that must exist in the generated advertisement content. It can be the element that the customer requires to exist in the advertisement content in the customer demand data, or it can be obtained by analyzing the product historical advertisement data and taking the factor with the highest occurrence frequency in the historical advertisement data as the mandatory item, or taking the factor whose occurrence frequency exceeds the frequency threshold as the mandatory item. The frequency threshold is a frequency value that can be set by those skilled in the art or the customer themselves.

[0027] In this embodiment, it should be specifically noted that during the process of the intelligent generation and prediction module for intelligent advertisement generation, the mandatory item is the content that must be reflected in the advertisement, and the dominant item is the content selected by itself in the advertisement. Therefore, the process of intelligent advertisement generation can be regarded as a process of permutation and combination, that is, a process of combining the mandatory item and the dominant item, and then generating P initial advertisements; when generating advertisements intelligently, artificial intelligence generation tools such as AdCreative.ai, Sivi.ai, and QuickAds can be used for intelligent advertisement generation; Input the P initial advertisements into the constructed advertisement prediction model in sequence to predict the corresponding delivery effect; the input layer of the advertisement prediction model has P nodes for inputting P initial advertisements, and the output layer also has P nodes for outputting the delivery effects of the P initial advertisements; select the initial advertisement with the best predicted advertisement delivery effect as the alternative advertisement and transmit it to the content optimization module for subsequent operations; The training process of the advertisement prediction model is specifically as follows: Pre-collect the advertisement content data of R1 groups of competing products and the historical advertisement content data of R2 groups of products, and use them as analysis data. Then there are a total of (R1 + R2) groups of analysis data. Both R1 and R2 are integers greater than 1. Convert a group of analysis data and its corresponding delivery effect into a corresponding group of feature vectors. The delivery effect can be the same as the advertisement effect calculation method in the fitness function, or can be calculated through advertisement click-through rate, advertisement conversion rate, customer acquisition cost, etc.; Use each group of feature vectors as the input of the advertisement prediction model. The advertisement prediction model takes a group of delivery effects corresponding to each group of analysis data as the output, and takes the actual delivery effect corresponding to each group of analysis data as the target. The actual delivery effect is the pre-collected delivery effect corresponding to the analysis data; take minimizing the sum of prediction errors of all analysis data as the training objective; the formula of the prediction error is expressed as: , where ε q is the prediction error, q is the group number of the feature vector corresponding to the analysis data, θ q is the delivery effect corresponding to the q-th group of analysis data, μ q is the actual delivery effect corresponding to the q-th group of analysis data. Train the advertisement prediction model until the sum of prediction errors reaches convergence and then stop training; The advertisement prediction model is a deep neural network model; it includes an input layer, a hidden layer, and an output layer; each hidden layer includes multiple neurons, and there are connections between each neuron and the neurons in the next layer. The connections contain weights, which determine the importance and influence of data transmission in the neural network; an activation function is applied to each neuron between the hidden layer and the output layer. The activation function introduces non-linearity and allows the network to learn more complex patterns and features.

[0028] In this embodiment, it should be specifically noted that the content optimization module includes an advertisement modification unit and an element bar unit; the advertisement modification unit is used for customers to modify the alternative advertisement content. After the alternative advertisement is displayed on the operation terminal, the customer can perform modification operations on the alternative advertisement. In the advertisement content of the alternative advertisement, each copywriting element and visual element is a separate modification item, and the customer can click on the modification item to perform replacement or modification operations; the element bar unit is used to display the copywriting elements and visual elements for replacement and modification; After the customer's modification is completed, the modified advertisement is saved and transmitted back to the intelligent generation and prediction module, input into the advertisement prediction model, and the placement effect of the modified advertisement is predicted. The placement effect of the modified advertisement is compared with that of the alternative advertisement. If the placement effect of the modified advertisement is better, the modified advertisement is directly replaced with the alternative advertisement and transmitted to the content optimization module again; if the placement effect of the alternative advertisement is better, the alternative advertisement is not replaced. At the same time, the placement effect of the modified advertisement is fed back to the interaction module for display, and a secondary inquiry is made in the interaction module. The content of the inquiry is whether to replace the alternative advertisement; if the customer still wants to select the modified advertisement when knowing that the placement effect has decreased, the customer's requirement can be met; if the placement effect of the modified advertisement is better than that of the alternative advertisement, it can be directly replaced to obtain a better placement effect while meeting the customer's conditions. The content optimization module can also perform professional inspections on the intelligently generated advertisement content. Professional personnel can modify or delete harmful content and negative views in the advertisement content to ensure the quality of the advertisement content.

[0029] In this embodiment, it should be specifically noted that the elitist method generates F1 offspring chromosomes. For a population with a capacity of B, the fitness values of the B chromosomes are arranged from largest to smallest, and each of the top F1 chromosomes generates an offspring chromosome; the roulette method generates F2 offspring chromosomes, that is, R chromosomes generate F2 offspring chromosomes according to the corresponding roulette probabilities; F1 + F2 = B to keep the offspring population capacity B unchanged and the population generation number increasing. The expression of the roulette probability is: , where ζ b is the roulette probability of the competitive product advertisement data corresponding to the b-th chromosome; The crossover recombination adopts the PMX method. E chromosomes are randomly selected from the population for crossover recombination to obtain E new chromosomes; after the chromosome crossover recombination, the fitness values of the E new chromosomes are calculated. The fitness values of the E new chromosomes and the E chromosomes are sorted from largest to smallest to generate a sorting table, and the E new chromosomes in the sorting table are replaced with the E chromosomes undergoing crossover recombination in the population in ascending order; in this embodiment, it is preferably E = 0.7B. If the calculated E is not an integer, E is rounded up to ensure that the calculated E is an integer. The preset mutation probability is V. According to the mutation probability, the B chromosomes in the population are mutated. The mutation method is to randomly select the positions of two genes in the chromosome and exchange the values of the two genes. In this embodiment, it is preferably V = 0.02, and the mutation probability is preset by those skilled in the art according to the algorithm efficiency and algorithm accuracy.

[0030] Finally, the above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

[0031] As described above, the above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed in the present application can easily think of changes or replacements, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. An intelligent advertising content generation and optimization system based on big data, characterized by: It includes customer data collection module, competitive product advertising acquisition module, competitive product advertising analysis module, intelligent generation and prediction module, content optimization module and interaction module; The customer data collection module is used to collect customer demand data and product data, and transmit them to the competitive product advertisement acquisition module and the competitive product advertisement analysis module; The competitive product advertisement acquisition module is used to receive data from the customer data collection module, obtain competitive product advertisement data, and separate it into comparative advertisement data and numerical advertisement data before transmitting it to the competitive product advertisement analysis module; The competitive product advertisement analysis module is used to receive data from the competitive product advertisement acquisition module and the customer data collection module, analyze the numerical advertisement data and the comparative advertisement data, and obtain the advantageous items and the required items; The intelligent generation and prediction module intelligently generates advertisements based on the advantages and the required items, and then inputs the advertisement prediction model to predict the delivery effect of the advertisements, selects the advertisements with the best delivery effect as candidate advertisements, and then transmits the candidate advertisements and the predicted delivery effect to the content optimization module; The content optimization module outputs the candidate advertisements and the predicted delivery effects to the interactive module for display. The customer modifies the advertisement content through the content optimization module. After the modification is completed, the modified advertisement is fed back to the intelligent generation and prediction module for delivery effect prediction, and then the prediction result is sent back to the content optimization module. If the customer does not need to modify the advertisement content, the alternative advertisement is directly used as the final advertisement and transmitted to the interactive module for display; The interactive module is used to display the finally generated advertisement on the human-computer interactive interface.

2. The intelligent advertising content generation and optimization system based on big data according to claim 1, characterized in that: The customer demand data includes the advertising content, text element data and visual element data that the customer expects to be presented; the advertising content includes product promotion content, service promotion content, brand image content and promotion activity content; the product promotion content includes product feature introduction, product usage method, product advantages and comparison with competing products and product price; the service promotion content includes service content, service process and timeliness, service advantages and characteristics and customer evaluation; the brand image content includes brand story and history, brand values ​​and concepts, brand partners and honors and brand public welfare activities and responsibilities; the promotion activities include limited-time discounts and discounts, holiday promotions, points redemption and gift giving, and joint promotions and partners; The text element data includes text content, text font, font format and font layout; the visual element data includes picture color, picture content, animation content and video content; the product data is data related to the advertised goods, including product name, product picture, product function, product price, product sales, product category and product history data; the product history data includes the product's historical sales quantity, sales amount, sales time and product historical advertising data, and the advertising data includes advertising content data, advertising conversion rate, advertising click-through rate and advertising exposure rate; the product is the advertised commodity; the advertising content data includes the text elements in the advertising content and the visual elements in the advertising content.

3. The intelligent advertising content generation and optimization system based on big data according to claim 2 is characterized by: The specific method for the competitive product advertisement acquisition module to acquire competitive product advertisement data is: Extract and fuse product data features. The feature fusion adopts linear fusion, which fuses different features in a linear combination manner, obtains new fused features through weight distribution, and uses the new fused features as fused product data features; Use big data technology to collect N products of the same category as the product being advertised, extract the product data features of the N products and fuse them to obtain the fused product data features, calculate the similarity between the fused product data features and the fused product data features, and obtain the products corresponding to the fused product data features with the top n similarity rankings as competing products; The most recent piece of advertising data of n competing products is used as competing product advertising data, that is, there are n competing product advertising data.

4. The intelligent advertising content generation and optimization system based on big data according to claim 3 is characterized by: The competitive product advertisement acquisition module divides the competitive product advertisement data into comparative advertisement data and numerical advertisement data, including the following steps: Step S01: Encode n competitive product advertisement data, obtain chromosomes, and construct an initial population; Step S02: Determine the fitness function; Step S03: Perform natural selection on chromosomes in the population; Step S04: performing crossover recombination on chromosomes in the population; Step S05: mutating the chromosomes in the population; Step S06: Obtain a new population, preset the population generation number to be L, the fitness threshold to be Q, where L is an integer greater than 0, and Q is a real number greater than 0; loop through steps S03-S05 until the generation number corresponding to the new population is L or the fitness corresponding to a chromosome in the new population is greater than or equal to the fitness threshold Q, then the loop ends, and the competitive product advertising data corresponding to the chromosome with the maximum fitness in the new population is taken as the optimal advertising data; The competitor product advertising data corresponding to the optimal advertising data is used as comparative advertising data, and the remaining n-1 competitor product advertising data is used as numerical advertising data.

5. The intelligent advertising content generation and optimization system based on big data according to claim 4 is characterized by: In step S01, each competitive product advertisement data is encoded as A, where A is a chromosome, and B chromosomes are randomly generated to form an initial population C. , where C b is the b-th chromosome, b=1, 2, 3, …, B; The fitness function is expressed as: ; Among them, f b is the fitness corresponding to the bth chromosome, GX b is the advertising effect of the competing product advertisement corresponding to the competing product advertisement data of the b-th chromosome; , where SR b _gg is the advertising revenue brought by the competitive product advertising corresponding to the competitive product advertising data of the bth chromosome, CB b _gg is the advertising cost of the competing product advertisement corresponding to the competing product advertisement data of the b-th chromosome.

6. The intelligent advertising content generation and optimization system based on big data according to claim 5 is characterized by: The similarity between the fused commodity data features and the fused product data features is calculated using cosine similarity, and the formula is: ； Among them, cosθ i is the similarity between the fused commodity data feature and the fused product data feature, RH i _sp is the feature vector of the ith fused commodity data, RH_cp is the feature vector of the fused product data, ‖RH i _sp‖ is the modulus of the feature vector of the i-th fused commodity data, ‖RH_cp‖ is the modulus of the feature vector of the fused product data, "∙" is the dot product, i=1, 2, 3, …, N.

7. The intelligent advertising content generation and optimization system based on big data according to claim 6 is characterized by: The competitive product advertising analysis module obtains the advantages and required items in the following manner: The similarity calculation is performed on the text elements and visual elements in the numerical advertising data and the comparative advertising data. Both the text elements and the visual elements contain text factors and visual factors. The data items contained in the text element data are all text factors, and the data items contained in the visual element data are all visual factors. The text elements and the visual elements are combined into an element set. The text factors and the visual factors are both factors in the element set. Then: The comparative advertising data is expressed as: , where U D To compare the advertising data element set, D j To compare the data characteristics of the jth factor in the advertising data element set, m is the total number of factors in the element set, that is, the sum of the number of copywriting factors and visual factors; Numerical advertising data is represented as: , where U s_a is the ath numerical advertisement data element set, s_a j is the data feature of the jth factor in the ath numerical advertising data element set, a=1, 2, 3, …, n-1; The comparison advertisement data element set corresponds one to one with the factors in the numerical advertisement data element set; Calculate the similarity between all factor data features in each numerical advertisement data element set and all corresponding factor data features in the comparison advertisement data element set, set a screening threshold YU1, and if the similarity does not reach the screening threshold, mark the factor in the comparison advertisement data element set as a dominant item; The required items are factors that must exist in the generated advertisement content.

8. The intelligent advertising content generation and optimization system based on big data according to claim 7 is characterized by: The intelligent generation and prediction module generates P initial advertisements, and inputs the P initial advertisements into the constructed advertisement prediction model in sequence to predict the corresponding delivery effects; the input layer of the advertisement prediction model has P nodes for inputting the P initial advertisements, and the output layer also has P nodes for outputting the delivery effects of the P initial advertisements; the initial advertisement with the best predicted delivery effect is selected as the alternative advertisement.

9. The intelligent advertising content generation and optimization system based on big data according to claim 8, characterized in that: The training process of the advertisement prediction model is specifically as follows: Collect the advertising content data of the R1 group of competitors and the historical advertising content data of the R2 group of products in advance and use them as analysis data. Then there are (R1+R2) groups of analysis data in total. Both R1 and R2 are integers greater than 1. Convert a group of analysis data and its corresponding delivery effect into a corresponding group of feature vectors. Each set of feature vectors is used as the input of the advertising prediction model. The advertising prediction model takes a set of delivery effects corresponding to each set of analysis data as output, takes the actual delivery effect corresponding to each set of analysis data as the target, and the actual delivery effect is the delivery effect corresponding to the analysis data collected in advance; the training goal is to minimize the sum of the prediction errors of all analysis data; the formula of the prediction error is expressed as: , where ε q is the prediction error, q is the group number of the eigenvector corresponding to the analyzed data, θ q is the delivery effect corresponding to the qth group of analysis data, μ q For the actual delivery effect corresponding to the qth group of analysis data, the advertising prediction model is trained until the sum of the prediction errors reaches convergence and the training is stopped; The advertisement prediction model is a deep neural network model, which includes an input layer, a hidden layer and an output layer. Each hidden layer includes multiple neurons, each neuron is connected to the neurons in the next layer, and the connection contains weights.

10. The intelligent advertising content generation and optimization system based on big data according to claim 9, characterized in that: The content optimization module includes an advertisement modification unit and an element bar unit; the advertisement modification unit is used for the customer to modify the content of the candidate advertisement. After the candidate advertisement is displayed on the operation terminal, the customer performs a modification operation on the candidate advertisement. In the advertisement content of the candidate advertisement, each text element and visual element is a separate modification item. The customer clicks on the modification item to perform a replacement or modification operation; the element bar unit is used to display the text elements and visual elements for replacement and modification; After the customer completes the modification, the modified advertisement is saved and sent back to the intelligent generation and prediction module, input into the advertisement prediction model, and the delivery effect of the modified advertisement is predicted. The delivery effect of the modified advertisement is compared with that of the alternative advertisement. If the delivery effect of the modified advertisement is better, the modified advertisement is directly replaced with the alternative advertisement and transmitted to the content optimization module again; if the delivery effect of the alternative advertisement is better, the alternative advertisement is not replaced, and the delivery effect of the modified advertisement is fed back to the interactive module for display, and a second inquiry is made in the interactive module to inquire whether to replace the alternative advertisement.

Images