Mobile phone advertising delivery method and system

Abstract

The present invention relates to a method and system for placing mobile phone advertisements, comprising: matching multiple basic landing pages from preset landing pages according to multiple key data information displayed by a target mobile phone advertisement to be placed; determining the probability of each key data information appearing in a display area corresponding to each style feature of the basic landing page based on prior page conversion data on the target mobile phone, and selecting a target style feature from the style features according to the probability in each basic landing page to generate a spare landing page; associating each key data information as landing page-associated key data information to a display area of ​​a target style feature corresponding to the spare landing page to generate a target landing page, and placing the key data information of a target mobile phone advertisement corresponding to the target landing page on the target mobile phone; and updating the probability of each key data information appearing in a display area of ​​each style feature of the basic landing page according to the acquired regional page conversion a posteriori data of the display area of ​​each style feature on the target landing page.

Description

Technical Field

[0001] The present disclosure relates to the field of advertising technology, and in particular to a method and system for delivering mobile phone advertisements. Background Art

[0002] With the rapid development of mobile Internet, mobile advertising has become an important channel for enterprises to promote their products and services. However, traditional mobile advertising often adopts a "one-size-fits-all" strategy, that is, showing the same advertising content and landing page to all users, ignoring the personalized needs of users and the diversity of mobile phone interfaces, resulting in low advertising click-through rate and conversion rate. In order to improve the accuracy and effectiveness of advertising, intelligent advertising technology has emerged, among which the optimization design of landing pages is particularly critical.

[0003] Existing smart advertising delivery technologies mainly rely on user behavior data and interest preferences in terms of landing page optimization, and recommend relevant content through algorithms to improve user experience and advertising effectiveness. However, most of these technologies focus on personalized recommendations of advertising content, while ignoring the impact of landing page style features on advertising conversion rates. Different user groups have different preferences for landing page style features (such as layout, color, font, etc.), and these preferences directly affect ad clicks and conversion behaviors. Summary of the invention

[0004] The purpose of the present invention is to provide a method and system for placing mobile phone advertisements, which aims to rationally analyze prior page conversion data, intelligently select the landing page style features that best suit the target user, and generate a highly personalized target landing page, thereby improving the click-through rate and conversion rate of the advertisement.

[0005] In order to achieve the above-mentioned purpose, a first aspect of the embodiments of the present disclosure provides a method for placing mobile phone advertisements, comprising:

[0006] According to a plurality of key data information displayed by the target mobile phone advertisement to be delivered, a plurality of basic landing pages when the target mobile phone advertisement is delivered on the mobile phone are matched from preset landing pages, wherein each of the key data information corresponds to a basic landing page, and the style features in each of the basic landing pages are different;

[0007] Based on the prior page conversion data on the target mobile phone, determine the probability that each of the key data information appears in the display area corresponding to each of the style features of the basic landing page in each of the basic landing pages, select the target style feature from the style features according to the probability in each of the basic landing pages, and generate a backup landing page;

[0008] Associating each of the key data information as landing page-associated key data information to the display area of ​​the target style feature corresponding to the backup landing page, generating a target landing page, and delivering the key data information of the target mobile phone advertisement corresponding to the target landing page to the target mobile phone;

[0009] According to the acquired regional page conversion a posteriori data of the display area corresponding to each style feature on the target landing page, the probability of each key data information appearing in the display area corresponding to each style feature of the basic landing page is updated. The regional page conversion a posteriori data is used to characterize the probability of page conversion occurring when visiting the display area on the target mobile phone.

[0010] In one embodiment, the determining, based on the prior page conversion data on the target mobile phone, the probability that each of the key data information appears in the display area corresponding to each of the style features of the basic landing page in each of the basic landing pages includes:

[0011] Converting the a priori style features corresponding to the a priori page conversion data on the target mobile phone into a first landing page feature vector, and converting each of the style features of the basic landing page into a second landing page feature vector;

[0012] Calculate the characteristic Euclidean distance between the first landing page characteristic vector and the second landing page characteristic vector corresponding to each of the style characteristics of each of the basic landing pages;

[0013] The priori page conversion data corresponding to the first landing page feature vector whose feature Euclidean distance corresponding to each style feature is less than the first preset distance threshold is used as the priori element in the first set object corresponding to the style feature;

[0014] Inputting the landing page associated key data information corresponding to the priori element in the first set object corresponding to each of the style features and each of the key data information into a preset neural network model, and obtaining the correlation between each of the landing page associated key data information output by the preset neural network model and the priori element in the first set object corresponding to each of the style features;

[0015] Based on the correlation between each of the key data information output by the preset neural network model and the landing page associated key data information in the first set object corresponding to each of the style features, the probability of each of the key data information appearing in the display area corresponding to each of the style features of the basic landing page is determined.

[0016] In one embodiment, the determining, based on the correlation between each of the key data information output by the preset neural network model and the landing page associated key data information in the first set object corresponding to each of the style features, the probability that each of the key data information appears in the display area corresponding to each of the style features of the basic landing page in each of the basic landing pages comprises:

[0017] According to the correlation between each of the key data information output by the preset neural network model and the landing page-related key data information in the first set object corresponding to each of the style features, determining, from the key data information, a priori elements in the second set object corresponding to each of the style features whose correlation exceeds a preset correlation threshold;

[0018] Determine the data feature vector of the key data information corresponding to the a priori element in the second set object corresponding to each of the style features, and determine the central feature vector corresponding to the second set object according to the data feature vector of the key data information corresponding to the a priori element in the second set object corresponding to each of the style features;

[0019] Calculating a first Euclidean distance between data feature vectors corresponding to the key data information corresponding to each of the prior elements in the second set object, and calculating a second Euclidean distance between the data feature vector corresponding to the key data information corresponding to each of the prior elements in the second set object and the corresponding central feature vector;

[0020] According to the first Euclidean distance and the second Euclidean distance of the key data information corresponding to each prior element in the second set object, the probability of each key data information appearing in the display area corresponding to each style feature of the basic landing page in each basic landing page is determined.

[0021] In one embodiment, determining the probability that each of the key data information appears in the display area corresponding to each of the style features of the basic landing page in each of the basic landing pages according to the first Euclidean distance and the second Euclidean distance of the key data information corresponding to each of the prior elements in the second set object includes:

[0022] Calculate the average of the first Euclidean distance and the second Euclidean distance of the key data information corresponding to each of the prior elements in the second set object;

[0023] Calculate the standard deviation of the first Euclidean distance and the second Euclidean distance of the key data information corresponding to each of the prior elements in the second set object;

[0024] Substituting the mean value and the standard deviation corresponding to the key data information corresponding to each of the prior elements in the second set object into a Gaussian distribution formula to obtain a probability density function corresponding to each of the prior elements in the second set object;

[0025] Substitute the mean of the first Euclidean distance and the second Euclidean distance corresponding to each key data information corresponding to each prior element in the second set object into the corresponding probability density function to obtain the probability that each key data information appears in the display area corresponding to each style feature of the basic landing page in each basic landing page.

[0026] In one implementation, determining the central feature vector corresponding to the second set object according to the data feature vector of the key data information corresponding to the priori element in the second set object corresponding to each of the style features includes:

[0027] Mapping the starting point of the data feature vector of the key data information corresponding to the a priori element in the second set object corresponding to each of the style features to the same origin coordinate, to obtain an origin coordinate system constructed by the data feature vector of the key data information corresponding to the a priori element in the second set object corresponding to each of the style features;

[0028] Determining the centroid coordinates according to the end point of the data feature vector of the key data information corresponding to the priori elements in the second set object corresponding to each of the style features in the origin coordinate system;

[0029] According to the center of gravity coordinates, a central feature vector corresponding to the second set of objects is determined.

[0030] In one implementation, determining the central feature vector corresponding to the second set of objects according to the barycentric coordinates includes:

[0031] The data feature vector whose end point is closest to the center of gravity coordinates is determined as the central feature vector corresponding to the second set object.

[0032] In one embodiment, the determining, based on the prior page conversion data on the target mobile phone, the probability that each of the key data information appears in the display area corresponding to each of the style features of the basic landing page in each of the basic landing pages includes:

[0033] Acquire multiple prior page conversion data on the target mobile phone, wherein the prior page conversion data includes click behavior records, browsing behavior records, and conversion behavior records in each of the basic landing pages;

[0034] For each of the basic landing pages, analyzing the display area of ​​each of the style features in the basic landing pages on the display page during the display process;

[0035] Based on the click behavior records, browsing behavior records and conversion behavior records in the prior page conversion data in each of the basic landing pages, counting the frequency of each of the key data information appearing in the display area corresponding to each of the style features;

[0036] Calculating the occurrence probability of each of the key data information in the display area corresponding to each of the style features of each of the basic landing pages, wherein the occurrence probability is obtained based on the ratio of the frequency of the key data information in the display area to the total frequency;

[0037] Each of the occurrence probabilities is determined as the probability that each of the key data information appears in the display area corresponding to each of the style features of the basic landing page in each of the basic landing pages.

[0038] In one implementation, the step of selecting a target style feature from the style features according to the probability in each of the basic landing pages to generate a backup landing page includes:

[0039] The style feature with the highest probability in each of the basic landing pages is selected as the target style feature to generate a backup landing page.

[0040] In one embodiment, the key data information includes at least one of the product name, discount information, brand logo, display background and font color information, font size and font information corresponding to the target mobile phone advertisement.

[0041] A second aspect of the embodiments of the present disclosure provides a mobile phone advertising delivery system, including:

[0042] a memory having a computer program stored thereon;

[0043] A processor is used to execute the computer program in the memory to implement the steps of any one of the methods in the first aspect.

[0044] The present invention provides a method and system for placing mobile phone advertisements. Compared with the prior art, it has the following beneficial effects:

[0045] According to the key data information of the target mobile phone advertisement to be delivered, multiple basic landing pages are matched from the preset landing page library. Each key data information corresponds to a basic landing page, and each basic landing page has different style features. The diversity and pertinence of the landing page content are ensured. Based on the prior page conversion data on the target mobile phone, the probability of each key data information appearing in the display area corresponding to each style feature of different basic landing pages is calculated. Through big data analysis, the user's preference and conversion rate difference for different style features are identified. The style feature with the highest probability in each basic landing page is selected as the target style feature to generate a backup landing page. This step ensures a high degree of match between the landing page style feature and the user's preference. The key data information is associated with the display area of ​​the target style feature corresponding to the backup landing page, the target landing page is generated, and it is delivered. The advertising content and the landing page style feature are accurately matched, which improves the attractiveness of the advertisement and the user's participation. According to the regional page conversion a posteriori data of the display area corresponding to each style feature on the target landing page, the display probability of the key data information in each basic landing page is updated in real time. The advertising delivery system can continuously learn and optimize, and improve the accuracy and effect of advertising delivery.

[0046] The present disclosure makes the advertising content more in line with the preferences of target users through intelligent matching and optimization of landing page style features, thereby significantly improving the click-through rate of advertisements. The personalized target landing page design can more effectively guide users to perform expected actions, such as purchasing products, filling out forms, etc., thereby greatly improving the conversion rate of advertisements. Through precise delivery and efficient conversion, advertisers can obtain higher returns at a lower cost and reduce the overall cost of advertising investment. Personalized and highly relevant landing page design can improve user experience, enhance user acceptance and favorability of advertising content, and is conducive to the shaping and dissemination of brand image. Furthermore, through intelligent matching and optimization of landing page style features, the precision and efficiency of mobile phone advertising delivery are achieved, which significantly improves advertising effects and user experience.

[0047] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The accompanying drawings are used to provide a further understanding of the present disclosure and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present disclosure but do not constitute a limitation of the present disclosure. In the accompanying drawings:

[0049] Figure 1 It is a flow chart of a method for placing mobile phone advertisements according to an embodiment of the specification.

[0050] Figure 2 It is a block diagram of a mobile phone advertising device according to an embodiment of the specification. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] The specific implementation of the present disclosure is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the present disclosure, and is not used to limit the present disclosure.

[0053] In order to achieve the above-mentioned purpose, the present invention provides a method for placing mobile phone advertisements. Figure 1 This is a flow chart of a method for placing mobile phone advertisements according to an embodiment. The method comprises:

[0054] In step S11, according to the multiple key data information displayed by the target mobile phone advertisement to be delivered, multiple basic landing pages when the target mobile phone advertisement is delivered on the mobile phone are matched from the preset landing pages, wherein each of the key data information corresponds to one of the basic landing pages, and the style features in each of the basic landing pages are different;

[0055] Among them, key data information is the core information related to the target mobile phone advertisement, such as product name, discount information, brand logo, etc., which is used to attract user clicks and conversions. Basic landing page is a pre-designed web page template that contains different style features and is used to carry the content display after the advertisement is clicked. Style features can include, for example: page layout, color scheme, font style, picture and video elements, interactive functions, etc.

[0056] In the disclosed embodiment, based on the key data information in the advertisement, a basic landing page that best matches the advertisement is selected from a plurality of preset basic landing pages. The matching criteria may include: the relevance of the key data information to the landing page content, the fit between the landing page style features and the advertisement style, etc.

[0057] For example, suppose the key data information of a mobile phone advertisement includes "new smartphone", "HD camera", "large battery capacity", etc. Select the basic landing pages that highlight the features of new smartphones, HD camera effects, and large battery capacity advantages from the preset landing pages. In this way, when users click on the ad, they will be directed to a landing page that is highly relevant and attractive to the ad content, thereby improving conversion rate and user experience.

[0058] In step S12, based on the prior page conversion data on the target mobile phone, the probability of each of the key data information appearing in the display area corresponding to each of the style features of the basic landing page is determined, and according to the probability in each of the basic landing pages, the target style feature is selected from the style features to generate a backup landing page;

[0059] Among them, the prior page conversion data is based on historical data, reflecting the statistical data of the performance of different style features in the landing page (such as click-through rate, conversion rate). It may include user behavior data, page visit data, conversion event data, etc. The target style feature is the style feature that is most likely to promote user conversion in each basic landing page based on the prior data.

[0060] In the disclosed embodiment, the prior page conversion data on the target mobile phone is used to analyze the conversion effect of each key data information under different style features, and the style feature with the highest conversion probability is selected as the target style feature to construct an alternative landing page. For example, based on historical data, it is found that placing the "limited time offer" information in a banner advertisement at the top of the page has the best conversion effect, so this style feature is selected as the target style feature to generate an alternative landing page.

[0061] In the disclosed embodiment, the prior page conversion data is analyzed to determine the probability of each key data information appearing in the display area corresponding to different style features in each basic landing page. The calculation of the probability may be based on a statistical model or a machine learning algorithm, taking into account a variety of factors, such as the attractiveness of the style feature, user preferences, historical conversion effects, etc.

[0062] For each basic landing page, select the style features that are likely to meet the requirements as the target style features. For specific key data information, for example, select the style features that perform best in historical data and are most likely to promote user conversions to generate alternative landing pages. The alternative landing page will retain the core content and structure of the basic landing page, but will be optimized in terms of style features to ensure that the advertising content can be displayed to users in the most effective way.

[0063] In the disclosed embodiment, it is assumed that the prior page conversion data shows that for the key data information "new smartphone", placing it in a banner ad at the top of the page has the best conversion effect. Then, when generating the alternate landing page, the system will choose to place the information "new smartphone" in a banner ad at the top of the page as the target style feature. In this way, when the user visits the alternate landing page, the probability of being attracted by the "new smartphone" information in the most eye-catching position increases, thereby increasing the possibility of clicks and conversions.

[0064] In step S13, each of the key data information is associated as landing page associated key data information in the display area of ​​the target style feature corresponding to the backup landing page, a target landing page is generated, and the key data information of the target mobile phone advertisement corresponding to the target landing page is delivered to the target mobile phone;

[0065] The landing page associated key data information is obtained by associating the key data information of the advertisement with the specific display location of the landing page.

[0066] In the disclosed embodiment, each key data information in the advertisement is associated with the display area corresponding to the target style feature in the backup landing page. This association ensures that the core information of the advertisement can be prominently displayed on the landing page, thereby attracting the user's attention and guiding them to convert. By associating the key data information with the display area of ​​the target style feature, the system generates the final target landing page. The target landing page not only retains the core content and structure of the backup landing page, but also displays the core information of the advertisement in a key position, thereby improving the attractiveness and conversion effect of the page.

[0067] Furthermore, after the target landing page is generated, it is associated with the corresponding target mobile phone advertisement and delivered to the target mobile phone. During the delivery process, the user's geographical location, device type, network status, etc. can be referenced to ensure that the advertisement can accurately reach the target audience.

[0068] In the disclosed embodiment, each key data information in the advertisement is accurately associated with the corresponding display area of ​​the target style feature in the backup landing page. This association is based on the previously analyzed prior page conversion data and the determined probability to ensure that the core information of the advertisement can be optimally displayed on the landing page. The key data information is automatically inserted into the display area of ​​the target style feature in the backup landing page. During the insertion process, the system will consider the layout, font, color and other details of the information to ensure that the information is consistent with the overall style of the landing page. At the same time, the landing page is optimized as a whole, such as adjusting the page layout, optimizing the loading speed, etc., to improve the user experience.

[0069] The target landing page not only retains the core content and structure of the backup landing page, but also highlights the core information of the advertisement in key positions. The page design is concise and clear, which can quickly attract users' attention and guide them to conversion. At the same time, the target landing page also takes into account users' browsing habits and device characteristics to ensure a good user experience on different devices.

[0070] For example, suppose the key data information of a mobile phone advertisement includes "new smartphone", "high-definition camera", and "extra-large battery capacity". In the alternative landing page, the system has determined to place the "new smartphone" in the banner ad at the top of the page as the target style feature. When generating the target landing page, the system will accurately associate the key data information of "new smartphone" with the banner ad position at the top of the page, and adjust the font, color and other details to be consistent with the overall style of the landing page. At the same time, the system will also consider the user's browsing habits and device characteristics, and optimize the page layout and loading speed. The final generated target landing page will highlight the "new smartphone" information while providing a good user experience and advertising effect.

[0071] In step S14, based on the acquired regional page conversion a posteriori data of the display areas corresponding to the style features on the target landing page, the probability of each key data information appearing in the display areas corresponding to the style features of the basic landing page in each basic landing page is updated. The regional page conversion a posteriori data is used to characterize the probability of page conversion occurring when accessing the display area on the target mobile phone.

[0072] The regional page conversion a posteriori data is data actually collected about page conversions of users in a specific display area after the advertisement is released. The regional page conversion a posteriori data can be used as a priori page conversion data again to re-execute step S12, for example.

[0073] Among them, the regional page conversion posterior data is the data collected during the actual delivery process, reflecting the probability of page conversion in the display area corresponding to different style features when the target mobile phone user visits the landing page. Conversion may include user behaviors such as clicks, purchases, registrations, and downloads.

[0074] In the disclosed embodiment, the probability of each key data information appearing in the display area corresponding to different style features in each basic landing page is updated based on the regional page conversion posterior data. This update is based on the actual delivery effect, so it can more accurately reflect the actual impact of different style features on user conversion. By updating the probability, the system can optimize the matching relationship between the advertising content and the landing page. For style features with better performance, the system will increase the probability of displaying the key data information of the advertisement; for style features with poor performance, the system will reduce the probability of displaying the key data information of the advertisement.

[0075] For example, suppose that during the actual delivery process, the system finds that placing the key data information "Limited Time Offer" in the sidebar in the middle of the page has a better conversion effect. Then, when updating the probability, the system will increase the probability of displaying the key data information "Limited Time Offer" in the sidebar in the middle of the page. At the same time, if the system finds that placing a certain key data information in the footer at the bottom of the page has a poor conversion effect, then the system will reduce the probability of displaying the key data information in the footer at the bottom of the page. Through this continuous optimization process, the system can ensure that the matching relationship between the advertising content and the landing page is always maintained in the best state, thereby improving the user's conversion rate and advertising effect.

[0076] The method of the above technical solution matches multiple basic landing pages from a preset landing page library according to the key data information of the target mobile phone advertisement to be delivered. Each key data information corresponds to a basic landing page, and each basic landing page has different style features. The diversity and pertinence of the landing page content are ensured. Based on the prior page conversion data on the target mobile phone, the probability of occurrence of each key data information in the display area corresponding to each style feature of different basic landing pages is calculated. Through big data analysis, the user's preference and conversion rate difference for different style features are identified. The style feature with the highest probability in each basic landing page is selected as the target style feature to generate a spare landing page. This step ensures a high degree of matching between the landing page style feature and the user's preference. The key data information is associated with the display area of ​​the target style feature corresponding to the spare landing page, the target landing page is generated, and it is delivered. The advertising content and the landing page style feature are accurately matched, which improves the attractiveness of the advertisement and the user's participation. According to the regional page conversion a posteriori data of the display area corresponding to each style feature on the target landing page, the display probability of the key data information in each basic landing page is updated in real time. The advertising delivery system can continuously learn and optimize to improve the accuracy and effect of advertising delivery.

[0077] The present disclosure makes the advertising content more in line with the preferences of target users through intelligent matching and optimization of landing page style features, thereby significantly improving the click-through rate of advertisements. The personalized target landing page design can more effectively guide users to perform expected actions, such as purchasing products, filling out forms, etc., thereby greatly improving the conversion rate of advertisements. Through precise delivery and efficient conversion, advertisers can obtain higher returns at a lower cost and reduce the overall cost of advertising investment. Personalized and highly relevant landing page design can improve user experience, enhance user acceptance and favorability of advertising content, and is conducive to the shaping and dissemination of brand image. Furthermore, through intelligent matching and optimization of landing page style features, the precision and efficiency of mobile phone advertising delivery are achieved, which significantly improves advertising effects and user experience.

[0078] In a possible implementation, in step S12, based on the prior page conversion data on the target mobile phone, determining the probability that each of the key data information appears in the display area corresponding to each of the style features of the basic landing page in each of the basic landing pages includes:

[0079] In step S121, the a priori style features corresponding to the a priori page conversion data on the target mobile phone are converted into a first landing page feature vector, and each style feature of the basic landing page is converted into a second landing page feature vector;

[0080] The prior style feature may be a page style feature that performs well in historical data. The first landing page feature vector is the prior style feature converted into a vector form, and the second landing page feature vector is the style feature of the basic landing page converted into a vector form.

[0081] In the disclosed embodiment, style features (such as color, layout, font, etc.) can be converted into numerical vectors. Through vectorization, the similarity of style features can be converted into the distance between vectors, so as to perform quantitative comparison.

[0082] In step S122, the characteristic Euclidean distance between the first landing page characteristic vector and the second landing page characteristic vector corresponding to each of the style features of each of the basic landing pages is calculated;

[0083] Among them, the characteristic Euclidean distance is used to measure the straight-line distance between two vectors in multidimensional space. It can quantify the similarity between two vectors and use the Euclidean distance formula to calculate the distance between two characteristic vectors. The smaller the distance, the higher the similarity between the two vectors. Calculate the Euclidean distance between the characteristic vector of the first landing page and the characteristic vector of the second landing page. If the distance is small, it means that the style features of the basic landing page are similar to the style features in the prior page conversion data.

[0084] In step S123, the priori page conversion data corresponding to the first landing page feature vector whose feature Euclidean distance corresponding to each style feature is less than the first preset distance threshold is used as a priori element in the first set object corresponding to the style feature;

[0085] The first preset distance threshold is used to screen the threshold of the prior page conversion data similar to the basic landing page style features. The prior element is an element in the prior page conversion data that meets the distance threshold condition.

[0086] In the disclosed embodiment, the calculated characteristic Euclidean distance is compared with a preset distance threshold. If the distance is less than the threshold, it is considered that the prior page conversion data is similar to the style features of the basic landing page, and it is used as a priori element. For example, assume that the first preset distance threshold is 2. If the calculated characteristic Euclidean distance is 1.5, which is less than the threshold 2, it is considered that the prior page conversion data is similar to the style features of the basic landing page, and it is used as a priori element.

[0087] In step S124, the landing page associated key data information corresponding to the priori element in the first set object corresponding to each of the style features and each of the key data information are input into a preset neural network model to obtain the correlation between each of the landing page associated key data information output by the preset neural network model and the priori element in the first set object corresponding to each of the style features;

[0088] The preset neural network model is a machine learning model for learning the correlation between key data information and style features. The correlation is the degree of correlation between the key data information and the style features.

[0089] In the disclosed embodiment, the selected prior elements and their corresponding landing page-related key data information are input into the neural network model. The neural network model outputs the correlation between the key data information and the style features by learning the patterns in the data. The selected prior elements and their corresponding landing page-related key data information (such as "new smartphone") are input into the neural network model. Through learning, the neural network model outputs the correlation between "new smartphone" and style features such as red background and central layout.

[0090] In step S125, based on the correlation between each of the key data information output by the preset neural network model and the landing page associated key data information in the first set object corresponding to each of the style features, the probability of each of the key data information appearing in the display area corresponding to each of the style features of the basic landing page is determined.

[0091] In the disclosed embodiment, the probability of the key data information in the display area corresponding to the specific style feature of the basic landing page is determined based on the correlation of the neural network model output. The higher the probability, the more suitable the key data information is to appear in the display area of ​​the style feature. Assume that the correlation between the neural network model output "new smartphone" and the central layout is high. It can be determined that the probability of "new smartphone" appearing in the display area corresponding to the central layout in the basic landing page is high.

[0092] The above technical solution converts the prior style features corresponding to the prior page conversion data on the target mobile phone and the style features of the basic landing page into numerical vector form, which is convenient for mathematical calculation and comparison. By calculating the Euclidean distance between the first landing page feature vector and the second landing page feature vector, the similarity between the two vectors is quantified, providing a basis for subsequent screening and correlation analysis. By setting a first preset distance threshold, the prior page conversion data similar to the style features of the basic landing page is screened out as the prior element. It ensures that the prior elements of the subsequent analysis have a certain similarity in style features with the basic landing page, thereby improving the accuracy and effectiveness of the analysis.

[0093] Furthermore, the landing page-associated key data information corresponding to the screened prior elements and each key data information are input into the preset neural network model, and the learning ability of the neural network is used to analyze the correlation between the key data information and the style features. The neural network model can capture the complex patterns and nonlinear relationships in the data, thereby more accurately revealing the intrinsic connection between the key data information and the style features. According to the correlation output by the neural network model, the probability of each key data information appearing in the display area corresponding to each style feature in each basic landing page is determined. It provides a quantitative basis for the optimization of the landing page, and the layout and display method of the key data information on the landing page can be adjusted according to the probability to improve the user experience and advertising effect. The accurate calculation of the probability of the key data information appearing in the display area corresponding to each style feature on the basic landing page is realized, which provides strong support for the optimization of the landing page. It helps to improve the accuracy and effectiveness of advertising, and improve user experience and conversion rate.

[0094] In one embodiment, in step S125, determining the probability that each of the key data information output by the preset neural network model appears in the display area corresponding to each of the style features of the basic landing page according to the correlation between each of the key data information output by the preset neural network model and the landing page associated key data information in the first set object corresponding to each of the style features, includes:

[0095] In step S1251, according to the correlation between each of the key data information output by the preset neural network model and the landing page associated key data information in the first set object corresponding to each of the style features, the prior elements in the second set object corresponding to each of the style features whose correlation exceeds a preset correlation threshold are determined from the key data information;

[0096] The second set object is a set of elements selected from the first set object and whose correlation exceeds a preset threshold. The prior element is a key data information in the second set object that is associated with a specific style feature and has a high correlation. The data feature vector is a feature vector used to represent key data information, such as word frequency, TF-IDF value, etc.

[0097] In the disclosed embodiment, each key data information and each style feature combination are used as input and sent to the neural network model. The model will output a correlation score for each pair of combinations, and this score represents the degree of correlation between the key data information and the style feature. The combinations that exceed the preset correlation threshold are screened, and a set correlation threshold is used as the screening criterion. Only when the correlation score between the key data information and the style feature exceeds this threshold, the pair of combinations is considered to be highly correlated. All key data information whose correlation score exceeds the threshold is screened out. These screened key data information are called prior elements because they are highly correlated with specific style features. These prior elements are classified into the second set object corresponding to the style feature.

[0098] In the disclosed embodiment, by comparing these correlations with a preset correlation threshold, key data information and style feature combinations with high correlations can be screened out and placed as prior elements in the second set object.

[0099] For example, the correlation between the key data information "price" and the style feature "title" is very high. At the same time, it is also found that the correlation between the key data information "brand" and the style feature "description text" also exceeds the preset threshold. Therefore, the combination of "price" and "title" and the combination of "brand" and "description text" are put into the second set object as prior elements.

[0100] In step S1252, the data feature vectors of the key data information corresponding to the a priori elements in the second set objects corresponding to the style features are determined, and the central feature vector corresponding to the second set objects is determined according to the data feature vectors of the key data information corresponding to the a priori elements in the second set objects corresponding to the style features;

[0101] The central feature vector is a second set object corresponding to each style feature, and the central feature vector can be obtained by calculating the average value of the data feature vectors of all the prior elements therein.

[0102] In the disclosed embodiment, for each second set object corresponding to a style feature, it is necessary to determine the data feature vector of each prior element therein. Then, by calculating the average value of these data feature vectors, the central feature vector corresponding to the style feature can be obtained. This central feature vector represents the average feature of the key data information of all prior elements under the style feature.

[0103] Following the above-mentioned embodiment, the central feature vectors of the second set object corresponding to the "title" (which mainly contains key data information related to the "price") and the second set object corresponding to the "description text" (which mainly contains key data information related to the "brand") are calculated respectively. In this way, two central feature vectors are obtained, which represent the average features of the key data information of all a priori elements under the "title" and "description text", respectively. In the subsequent landing page optimization, the display content of the "title" and "description text" can be adjusted according to these two central feature vectors to improve the user's purchase conversion rate.

[0104] In step S1253, a first Euclidean distance between data feature vectors corresponding to the key data information corresponding to each of the prior elements in the second set object is calculated, and a second Euclidean distance between the data feature vector corresponding to the key data information corresponding to each of the prior elements in the second set object and the corresponding central feature vector is calculated;

[0105] Among them, the first Euclidean distance refers to the distance between the data feature vectors of the key data information corresponding to each prior element in the second set object. Euclidean distance is a commonly used distance measurement method used to measure the similarity or difference between two vectors. The second Euclidean distance refers to the distance between the data feature vector of the key data information corresponding to each prior element in the second set object and the corresponding central feature vector. Similarly, this is also measured using Euclidean distance.

[0106] In step S1254, based on the first Euclidean distance and the second Euclidean distance of the key data information corresponding to each prior element in the second set object, the probability of each key data information appearing in the display area corresponding to each style feature of the basic landing page in each basic landing page is determined.

[0107] In the disclosed embodiment, the two Euclidean distances calculated above are used to determine the probability of key data information appearing in the display area corresponding to each style feature in the basic landing page. Specifically, the probability of each key data information appearing under each style feature can be determined based on the size relationship between the first Euclidean distance and the second Euclidean distance, and the comparison results between them and the preset threshold or probability distribution function.

[0108] For example, suppose you are optimizing the landing page of a news website to increase the user's click-through rate. In this scenario, key data information may include the title, summary, author, etc. of the news, while style features may include the main title, subtitle, news picture, etc. of the page.

[0109] The first Euclidean distance between the data feature vectors of the key data information (such as the specific content of the “hot news”) corresponding to each prior element (such as the title of “hot news”) in the second set of objects, as well as the second Euclidean distance between the data feature vectors of these prior elements and the corresponding central feature vector (such as the average feature vector of the “hot news” category) were calculated.

[0110] Based on the two calculated Euclidean distances, we can determine the probability that the key data information "Hot News" appears in the display area corresponding to the style features such as the main title, subtitle or news picture in the basic landing page. For example, if the second Euclidean distance between the data feature vector of "Hot News" and the central feature vector of the main title is small, then the probability of "Hot News" appearing in the main title area is higher. In this way, we can optimize the layout and content of the landing page based on these probabilities to increase the user's click-through rate.

[0111] In one embodiment, in step S1254, determining the probability that each of the key data information appears in the display area corresponding to each of the style features of the basic landing page in each of the basic landing pages according to the first Euclidean distance and the second Euclidean distance of the key data information corresponding to each of the prior elements in the second set object includes:

[0112] In step S12541, the average of the first Euclidean distance and the second Euclidean distance of the key data information corresponding to each of the prior elements in the second set object is calculated;

[0113] In the disclosed embodiment, for each a priori element in the second set object, first, all the first Euclidean distances and all the second Euclidean distances are used to participate in the calculation of the sum of the first Euclidean distance and the second Euclidean distance of each of the a priori elements in the second set object, and then the average value is calculated based on the sum. For example, if the second set object includes the key data information corresponding to 3 a priori elements, there will be 3 first Euclidean distances and 2 or 3 second Euclidean distances (depending on whether the central feature vector is determined from the data feature vector of the key data information corresponding to the 3 a priori elements. If the central feature vector is determined from the data feature vector of the key data information corresponding to the 3 a priori elements, there will only be 2 second Euclidean distances. If the central feature vector is not determined from the data feature vector of the key data information corresponding to the 3 a priori elements, there will be 3 second Euclidean distances). Furthermore, the sum of the 3 first Euclidean distances and 2 or 3 second Euclidean distances can be calculated, and the average value is obtained to obtain the average value in this step.

[0114] In step S12542, the standard deviation of the first Euclidean distance and the second Euclidean distance of the key data information corresponding to each of the prior elements in the second set object is calculated;

[0115] Similarly, for each a priori element in the second set object, firstly all the first Euclidean distances and all the second Euclidean distances are used to participate in calculating the standard deviation of the first Euclidean distance and the second Euclidean distance of each of the a priori elements in the second set object.

[0116] In step S12543, the mean value and the standard deviation corresponding to the key data information corresponding to each of the prior elements in the second set object are substituted into a Gaussian distribution formula to obtain a probability density function corresponding to each of the prior elements in the second set object;

[0117] Among them, Gaussian distribution (normal distribution) is a continuous probability distribution, which is widely used to describe random variables in nature and social sciences. In this step, using the mean and standard deviation calculated in steps S12541 and S12542, a Gaussian distribution model (also called normal distribution) can be constructed to describe the distribution of key data information of each prior element in the second set object. For example, a Gaussian distribution model is constructed based on the mean and standard deviation of the Euclidean distance of the key data information of the product. This model can predict the possible performance range of the product in a certain period of time in the future, as well as the probability of these performances occurring.

[0118] In step S12544, the mean of the first Euclidean distance and the second Euclidean distance corresponding to each key data information corresponding to each prior element in the second set object is substituted into the corresponding probability density function to obtain the probability that each key data information appears in the display area corresponding to each style feature of the basic landing page in each basic landing page.

[0119] In the disclosed embodiment, the mean of the Euclidean distances corresponding to the key data information of each prior element in the second set object is substituted into the Gaussian distribution (probability density function) obtained in step S12543, and the probability of each key data information appearing in the display area corresponding to each basic landing page style feature is calculated.

[0120] For example, in the case of an e-commerce website, you may want to know the probability of a product's price or sales information appearing in a display area with a specific style feature (such as the top, middle, or bottom of the page). By calculating the probability density of the mean of the Euclidean distance of these key data information in the Gaussian distribution model, you can get the specific probability value.

[0121] In one implementation, in step S1252, determining the central feature vector corresponding to the second set object according to the data feature vector of the key data information corresponding to the priori element in the second set object corresponding to each of the style features includes:

[0122] In step S12521, the starting point of the data feature vector of the key data information corresponding to the priori element in the second set object corresponding to each of the style features is mapped to the same origin coordinate, so as to obtain the origin coordinate system constructed by the data feature vector of the key data information corresponding to the priori element in the second set object corresponding to each of the style features;

[0123] The data feature vectors of the key data information of the prior elements associated with each style feature in all the second set objects are unified into a common reference frame, namely the origin coordinate system. This helps in the aggregation and calculation of vectors in the subsequent steps.

[0124] In specific implementation, the starting point of each data feature vector (usually the starting coordinate of the vector) is mapped to the origin (0,0), thereby constructing an origin coordinate system centered on the origin. In this way, the directions and lengths of all vectors are compared and calculated in this unified coordinate system.

[0125] In step S12522, the centroid coordinates are determined according to the end point of the data feature vector of the key data information corresponding to the priori elements in the second set object corresponding to each of the style features in the origin coordinate system;

[0126] In the disclosed embodiment, in the origin coordinate system, the endpoints of all mapped data feature vectors constitute a point set. The "center of gravity" coordinates of these points represent the central tendency of these vectors as a whole. The calculation of the center of gravity coordinates usually involves weighted averaging of all these endpoint coordinates. Since all vectors are considered to be equally weighted in this scenario (unless there is additional information indicating different weights), the center of gravity coordinates can be obtained by simply taking the arithmetic average of all the endpoint coordinates.

[0127] In step S12523, the central feature vector corresponding to the second set object is determined according to the centroid coordinates.

[0128] In the implementation of the present disclosure, the barycentric coordinates are determined and the central feature vector of the second set object can be constructed based on the coordinates. This vector starts from the origin and points to the barycentric coordinates. Its length and direction represent the comprehensive characteristics of the key data information of all prior elements in the set object. The central feature vector can help understand the overall distribution and trend of the data in the feature space, thereby optimizing model training and improving prediction accuracy.

[0129] For example, the key data information (such as price, sales volume, etc.) of all products in each category is converted into data feature vectors, and the starting points of these vectors are mapped to the origin coordinate (0,0). In this way, the performance of different products in the same category is compared in a common coordinate system.

[0130] Next, the average coordinates of the end points of the feature vectors of all product data in each category are calculated, which is the centroid coordinates of the category. This coordinate reflects the overall trend of the category's products in key indicators such as price and sales volume. Finally, the central feature vector is constructed based on the centroid coordinates of each category. These vectors can not only intuitively understand the characteristics of each category, but can also be used for further data analysis and model training, such as comparing the central feature vectors of different categories to find similarities and differences, thereby optimizing product recommendation algorithms or inventory management strategies.

[0131] In one implementation, in step S12523, determining the central feature vector corresponding to the second set object according to the barycentric coordinates includes:

[0132] The data feature vector whose end point is closest to the center of gravity coordinates is determined as the central feature vector corresponding to the second set object.

[0133] In the disclosed embodiment, the centroid coordinates of each category are calculated based on the endpoint coordinates of the data feature vectors of each commodity. This coordinate represents the overall trend or center point of the category of commodities on the key indicators. Next, the data feature vectors of all commodities in the category are traversed to calculate the Euclidean distance (or other suitable distance measurement) between their respective endpoints and the centroid coordinates. This distance measures the degree of deviation of each vector from the overall trend in the feature space.

[0134] Suppose there is a clothing category that contains multiple items. The data feature vector of each item consists of quantitative indicators of its price, sales volume, and user evaluation. After calculating the barycentric coordinates of this category, start comparing the distance between the end point of each item's data feature vector and this barycentric coordinate. After comparing all distances, select the data feature vector closest to the barycentric coordinate as the central feature vector of the clothing category. This vector is closest to the overall trend in the feature space, so it can be considered to best represent the characteristics of this category.

[0135] For example, if the data feature vector of a product (representing moderate price, stable sales, and good user reviews) is closest to the centroid coordinate, then this vector is selected as the central feature vector of the clothing category. This vector will be used in subsequent data analysis, model training, or business decision-making to represent the characteristics of the entire category.

[0136] In this way, a representative central feature vector can be determined for each second set object (such as goods of different categories), so that large-scale data sets can be processed and analyzed more efficiently.

[0137] In one embodiment, in step S12, based on the prior page conversion data on the target mobile phone, determining the probability that each of the key data information appears in the display area corresponding to each of the style features of the basic landing page in each of the basic landing pages includes:

[0138] In step S1201, a plurality of prior page conversion data on the target mobile phone is obtained, wherein the prior page conversion data includes click behavior records, browsing behavior records and conversion behavior records in each of the basic landing pages;

[0139] Among them, the user's interaction data on the basic landing page is collected through log records, user behavior tracking systems, etc. These data should record in detail the user's specific operations on each page, such as which links are clicked, which areas are browsed, and whether the conversion is finally completed (such as purchase, registration, etc.). In this way, the behavior data of multiple users on the target mobile phone on the basic landing page can be collected, including click, browse and conversion behaviors.

[0140] In step S1202, for each of the basic landing pages, the display area of ​​each of the style features in the basic landing pages on the display page during the display process is analyzed;

[0141] Among them, web page layout analysis technology (such as DOM analysis, CSS style analysis, etc.) is used to identify and locate the display areas of all style features in each basic landing page. These areas will serve as key areas for subsequent analysis. In this way, the specific locations of different style features (such as pictures, text blocks, buttons, etc.) in each basic landing page can be clearly identified on the display page.

[0142] In step S1203, based on the click behavior records, browsing behavior records and conversion behavior records in the prior page conversion data in each of the basic landing pages, the frequency of each of the key data information appearing in the display area corresponding to each of the style features is counted;

[0143] Among them, all prior page conversion data are traversed, and for each basic landing page, whether the key data information appears in the display area of ​​a certain style feature is checked and recorded. By cumulative counting, the frequency of each key data information in each display area is obtained. Therefore, based on the prior page conversion data, the frequency of key data information (such as product name, price, promotion information, etc.) in the corresponding display area of ​​each style feature can be counted.

[0144] In step S1204, the occurrence probability of each key data information in the display area corresponding to each style feature of each basic landing page is calculated, wherein the occurrence probability is obtained based on the ratio of the frequency of the key data information in the display area to the total frequency;

[0145] For each key data information and each display area, the frequency of the key data information in the area is divided by the total frequency of the key data information in all display areas to obtain the probability of the key data information appearing in the display area. This probability reflects the exposure and importance of the key data information in the area. Therefore, by calculating the ratio of the frequency to the total frequency, the probability of each key data information appearing in the display area corresponding to each basic landing page style feature can be obtained.

[0146] In step S1205, each occurrence probability is determined as a probability that each key data information appears in a display area corresponding to each style feature of the basic landing page in each basic landing page.

[0147] In the disclosed embodiment, the calculated occurrence probability is used as the final probability of each key data information appearing in the display area corresponding to each style feature in each basic landing page. For example, the e-commerce platform collects a large number of users' click, browse and purchase behavior data on the product details page in recent times through the user behavior tracking system. Using web page analysis tools, identify the key style features in the product details page, such as product pictures, price tags, purchase buttons, etc., and determine their display areas on the page. Analyze the collected user behavior data, and count the occurrence frequency of key data information such as product name, price, promotion information in the display area corresponding to style features such as product pictures, price tags, and purchase buttons. Calculate the occurrence probability of each key data information in each display area. For example, it is found that the probability of the product name appearing in the product picture area is much higher than that in the price tag area, and the probability of the price information appearing in the price tag area is the highest. According to the calculated occurrence probability, the e-commerce platform decides to adjust the page layout so that the product name is more prominently displayed in the product picture area, while ensuring that the price information is clearly visible in the price tag area, so as to improve the user's willingness to buy and conversion rate.

[0148] In one embodiment, in step S12, selecting a target style feature from the style features according to the probability in each of the basic landing pages to generate a backup landing page includes:

[0149] The style feature with the highest probability in each of the basic landing pages is selected as the target style feature to generate a backup landing page.

[0150] In the disclosed embodiment, the probability of occurrence of key data information in the display area corresponding to different style features in each basic landing page is calculated. These probability data should clearly reflect the user's attention and preference for each style feature during browsing, clicking and conversion.

[0151] For each basic landing page, traverse all its style features and the probability of occurrence of their corresponding key data information. Find the style feature with the highest probability in the page, that is, the style feature that user behavior data shows that users pay the most attention to or interact with most frequently. Determine this style feature as the target style feature of the page. Based on the target style features of each selected basic landing page, design and generate new backup landing pages. In the backup landing page, these target style features should be highlighted, while considering the overall layout aesthetics and user experience of the page.

[0152] For example, you can use methods such as A / B testing to compare the original landing page with the alternative landing page to evaluate the effectiveness of the alternative landing page. Then, based on user feedback and behavior data on the new landing page, continue to optimize the design and content of the alternative landing page. Iterate the above process, continuously select new target style features and generate new alternative landing pages to continuously improve user experience and conversion rate.

[0153] For example, for each product detail page, the data analysis team found that the key data information (such as product name) in the product image area has the highest probability of appearing, indicating that users pay the most attention to product images and related information. Therefore, the product image area is determined as the target style feature of the product detail page. Based on the above analysis, the design team of the e-commerce platform designed a new product detail page as an alternative landing page. In this alternative landing page, the product image area is significantly enlarged and placed in the center of the page to attract users' attention. At the same time, the price tag and purchase button are also carefully designed and placed in an easy-to-click position. The e-commerce platform compares the original product detail page with the alternative landing page through A / B testing and displays it to different user groups. Based on user feedback and behavior data (such as click-through rate, conversion rate, etc.), the e-commerce platform found that the conversion rate of the alternative landing page was significantly higher than that of the original page. Therefore, the e-commerce platform decided to promote the alternative landing page as the new product detail page across the entire network and continue to collect user feedback for subsequent optimization and iteration.

[0154] In one embodiment, the key data information includes at least one of the product name, discount information, brand logo, display background and font color information, font size and font information corresponding to the target mobile phone advertisement.

[0155] Among them, the product name is the name of the specific product promoted or displayed in the target mobile phone advertisement, which is the basis for users to identify the product and understand the product information. Discount information is the information of promotions, discounts, special offers or other preferential measures provided in the advertisement to attract users to buy. This information is usually used to stimulate users' desire to buy and improve conversion rate. Brand logo is a general term for graphics, text, color and other elements designed to distinguish one's own products from other products. In advertisements, the display of brand logo helps to enhance users' recognition and memory of the brand. Display background is the background area used to display products, text, pictures and other content in advertisements. The design of display background can affect users' overall impression and attention allocation of advertisements. Font color information is the font color used in advertisements. Different colors can convey different emotions and atmospheres, which has an important impact on users' visual experience and reading comfort. Font size is the size of the font, that is, the font size. It determines the display size of the text in the advertisement and is crucial to the clarity and readability of the information. Font information is the specific type or style of the font, such as Songti, Heiti, Kaiti, etc., or a specially designed font. Font information also affects the overall visual effect and reading experience of the advertisement.

[0156] The present disclosure also provides a mobile phone advertising delivery system, including:

[0157] a memory having a computer program stored thereon;

[0158] A processor is used to execute the computer program in the memory to implement the steps of the method in any one of the aforementioned embodiments.

[0159] Figure 2 The mobile phone advertising delivery device 100 shown includes: a processor 1001 and a memory 1003. The processor 1001 and the memory 1003 are connected, such as through a bus 1002. Optionally, the mobile phone advertising delivery device 100 may also include a communication component, which can be used for data interaction between the device 100 and other devices, such as data transmission and / or data reception. It should be noted that the communication component is not limited to one in actual scheduling, and the structure of the mobile phone advertising delivery device 100 does not constitute a limitation on the embodiments of the present application.

[0160] Processor 1001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. Processor 1001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0161] The bus 1002 may include a path to transmit information between the above components. The bus 1002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 1002 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 2 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0162] The memory 1003 may be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), magnetic disk storage medium, other magnetic storage devices, or any other medium that can be used to carry or store program code and can be read by a computer, without limitation herein.

[0163] The memory 1003 is used to store program codes for executing the embodiments of the present disclosure, and the execution is controlled by the processor 1001. The processor 1001 is used to execute the program codes stored in the memory 1003 to implement the steps shown in the embodiment of the mobile phone advertising delivery method.

[0164] The embodiment of the present disclosure further provides a computer-readable storage medium having program codes stored thereon. When the program codes are executed by a processor, the steps and corresponding contents of the aforementioned mobile phone advertisement delivery method embodiment can be implemented.

[0165] The preferred embodiments of the present disclosure are described in detail above in conjunction with the accompanying drawings; however, the present disclosure is not limited to the specific details in the above embodiments; within the technical concept of the present disclosure, various changes, modifications, substitutions and variations may be made to these embodiments, and these changes, modifications, substitutions and variations all fall within the protection scope of the present disclosure.

[0166] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any appropriate manner without contradiction, and they should also be regarded as the contents disclosed in this disclosure. In order to avoid unnecessary repetition, this disclosure will not further describe various possible combinations. The technical scope of this application is not limited to the contents in the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. A method for placing mobile phone advertisements, characterized in that: include: According to a plurality of key data information displayed by the target mobile phone advertisement to be delivered, a plurality of basic landing pages when the target mobile phone advertisement is delivered on the mobile phone are matched from preset landing pages, wherein each of the key data information corresponds to a basic landing page, and the style features in each of the basic landing pages are different; Based on the prior page conversion data on the target mobile phone, determine the probability that each of the key data information appears in the display area corresponding to each of the style features of the basic landing page in each of the basic landing pages, select the target style feature from the style features according to the probability in each of the basic landing pages, and generate a backup landing page; Associating each of the key data information as landing page-associated key data information to the display area of ​​the target style feature corresponding to the backup landing page, generating a target landing page, and delivering the key data information of the target mobile phone advertisement corresponding to the target landing page to the target mobile phone; According to the acquired regional page conversion posterior data of the display area corresponding to each style feature on the target landing page, the probability of each key data information appearing in the display area corresponding to each style feature of the basic landing page in each basic landing page is updated, and the regional page conversion posterior data is used to characterize the probability of page conversion occurring in the access of the target mobile phone in the display area; The determining, based on the prior page conversion data on the target mobile phone, the probability that each of the key data information appears in the display area corresponding to each of the style features of the basic landing page in each of the basic landing pages includes: Converting the a priori style features corresponding to the a priori page conversion data on the target mobile phone into a first landing page feature vector, and converting each of the style features of the basic landing page into a second landing page feature vector; Calculate the characteristic Euclidean distance between the first landing page characteristic vector and the second landing page characteristic vector corresponding to each of the style characteristics of each of the basic landing pages; The priori page conversion data corresponding to the first landing page feature vector whose feature Euclidean distance corresponding to each style feature is less than the first preset distance threshold is used as the priori element in the first set object corresponding to the style feature; Inputting the landing page associated key data information corresponding to the priori element in the first set object corresponding to each of the style features and each of the key data information into a preset neural network model, and obtaining the correlation between each of the landing page associated key data information output by the preset neural network model and the priori element in the first set object corresponding to each of the style features; Based on the correlation between each of the key data information output by the preset neural network model and the landing page associated key data information in the first set object corresponding to each of the style features, the probability of each of the key data information appearing in the display area corresponding to each of the style features of the basic landing page is determined.

2. The method according to claim 1, characterized in that The determining, based on the correlation between each of the key data information output by the preset neural network model and the landing page associated key data information in the first set object corresponding to each of the style features, the probability that each of the key data information appears in the display area corresponding to each of the style features of the basic landing page in each of the basic landing pages comprises: According to the correlation between each of the key data information output by the preset neural network model and the landing page-related key data information in the first set object corresponding to each of the style features, determining, from the key data information, a priori elements in the second set object corresponding to each of the style features whose correlation exceeds a preset correlation threshold; Determine the data feature vector of the key data information corresponding to the a priori element in the second set object corresponding to each of the style features, and determine the central feature vector corresponding to the second set object according to the data feature vector of the key data information corresponding to the a priori element in the second set object corresponding to each of the style features; Calculating a first Euclidean distance between data feature vectors corresponding to the key data information corresponding to each of the prior elements in the second set object, and calculating a second Euclidean distance between the data feature vector corresponding to the key data information corresponding to each of the prior elements in the second set object and the corresponding central feature vector; According to the first Euclidean distance and the second Euclidean distance of the key data information corresponding to each prior element in the second set object, the probability of each key data information appearing in the display area corresponding to each style feature of the basic landing page in each basic landing page is determined.

3. The method according to claim 2, characterized in that The determining, based on the first Euclidean distance and the second Euclidean distance of the key data information corresponding to each of the prior elements in the second set object, the probability that each of the key data information appears in the display area corresponding to each of the style features of the basic landing page in each of the basic landing pages comprises: Calculate the average of the first Euclidean distance and the second Euclidean distance of the key data information corresponding to each of the prior elements in the second set object; Calculate the standard deviation of the first Euclidean distance and the second Euclidean distance of the key data information corresponding to each of the prior elements in the second set object; Substituting the mean value and the standard deviation corresponding to the key data information corresponding to each of the prior elements in the second set object into a Gaussian distribution formula to obtain a probability density function corresponding to each of the prior elements in the second set object; Substitute the mean of the first Euclidean distance and the second Euclidean distance corresponding to each key data information corresponding to each prior element in the second set object into the corresponding probability density function to obtain the probability that each key data information appears in the display area corresponding to each style feature of the basic landing page in each basic landing page.

4. The method according to claim 2, characterized in that: The determining, according to the data feature vector of the key data information corresponding to the priori element in the second set object corresponding to each of the style features, a central feature vector corresponding to the second set object comprises: Mapping the starting point of the data feature vector of the key data information corresponding to the a priori element in the second set object corresponding to each of the style features to the same origin coordinate, to obtain an origin coordinate system constructed by the data feature vector of the key data information corresponding to the a priori element in the second set object corresponding to each of the style features; Determining the centroid coordinates according to the end point of the data feature vector of the key data information corresponding to the priori elements in the second set object corresponding to each of the style features in the origin coordinate system; According to the center of gravity coordinates, a central feature vector corresponding to the second set of objects is determined.

5. The method according to claim 4, characterized in that The step of determining the central feature vector corresponding to the second set of objects according to the barycentric coordinates includes: The data feature vector whose end point is closest to the center of gravity coordinates is determined as the central feature vector corresponding to the second set object.

6. The method according to any one of claims 1 to 5, characterized in that: The step of selecting target style features from the style features according to the probability in each of the basic landing pages to generate a backup landing page includes: The style feature with the highest probability in each of the basic landing pages is selected as the target style feature to generate a backup landing page.

7. The method according to any one of claims 1 to 5, characterized in that: The key data information includes at least one of the product name, discount information, brand logo, display background and font color information, font size and font information corresponding to the target mobile phone advertisement.

8. A mobile phone advertising system, characterized in that: include: a memory having a computer program stored thereon; A processor, configured to execute the computer program in the memory to implement the steps of the method according to any one of claims 1 to 7.

Images