A cross-border advertisement accurate delivery method, system and device based on user portrait

Abstract

The application discloses a cross-border advertisement accurate delivery method, system and device based on user portrait, relates to the technical field of cross-border digital marketing, and discloses the cross-border advertisement accurate delivery method, system and device based on user portrait, which generates accurate user portrait by integrating cross-regional behavior data, dynamically calculates an advertisement delivery strategy in combination with exchange rate, customs duty and logistics economic parameters, and is adapted in compliance with laws and cultural rules, thereby solving the problems of single data dimension, inaccurate user portrait and missing economic variables in the prior art, and having the advantages of improving cross-border advertisement delivery accuracy, conversion effect and reducing compliance risks.

Description

Technical Field

[0001] This application relates to the field of cross-border digital marketing technology, and in particular to a method, system and device for precise cross-border advertising delivery based on user profiles. Background Technology

[0002] Cross-border advertising refers to targeted marketing activities aimed at user groups in different countries or regions. Current cross-border advertising technologies suffer from limitations on three main levels: First, in terms of data collection, existing solutions typically rely solely on users' browsing history, search records, or basic geographic location information, failing to effectively integrate comprehensive data such as users' cross-regional behavioral patterns, multi-currency transaction characteristics, and device network environments. Second, in terms of user profiling, traditional methods lack the ability to identify cross-border specific characteristics such as cross-time zone activity patterns and multilingual interaction preferences, resulting in user profiles that fail to accurately reflect users' true needs in cross-border consumption scenarios. Third, in the economic parameter calculation stage, existing technologies generally ignore the impact of cross-border-specific economic variables such as exchange rate fluctuations, differences in tariff policies, and changes in logistics costs on advertising effectiveness. Furthermore, existing systems often use uniform templates when generating advertising content, failing to dynamically adapt to the laws, regulations, religious and cultural taboos of the target market, easily leading to compliance risks or cultural conflicts. These technical deficiencies collectively result in prominent problems such as low targeting accuracy, poor conversion rates, and high compliance risks in current cross-border advertising. Therefore, existing technologies urgently need improvement to address these issues.

[0003] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0004] The main purpose of this application is to provide a method, system, and device for precise cross-border advertising based on user profiles, aiming to improve the accuracy and conversion rate of cross-border advertising and reduce compliance risks.

[0005] To achieve the above objectives, this application proposes a method for precise cross-border advertising targeting based on user profiles, the method comprising:

[0006] Acquire user behavior data, transaction data, and device characteristic data;

[0007] The behavioral data, transaction data, and device feature data are subjected to cross-regional feature fusion processing to generate cross-border user profiles.

[0008] The matching degree between advertisements and users is determined based on the cross-border user profile, and data on exchange rate fluctuations, tariff rates, and logistics costs in the target region are obtained.

[0009] Calculate the advertisement delivery economic parameters based on the matching degree of the advertisement and the user, exchange rate fluctuation data, tariff rate data and logistics cost data;

[0010] Generate the original advertisement based on the cross-border user portrait and the advertisement delivery economic parameters, and filter and format convert the original advertisement according to the legal library and cultural library of the target region to generate the final advertisement;

[0011] Match the final advertisement to the advertisement server cluster of the target region for advertisement delivery.

[0012] In an embodiment, the step of obtaining the behavior data, transaction data and device feature data of the user comprises:

[0013] Collecting the behavior data through a browser API; the behavior data includes the page stay duration of the user in multiple countries or regions, the click behavior sequence and the language preference;

[0014] Obtaining the transaction data through a payment gateway interface; the transaction data includes the transaction currency, transaction amount and commodity category;

[0015] Obtaining the device feature data through the terminal device used by the user; the device feature data includes the device time zone, IP address jump frequency and network proxy feature.

[0016] In an embodiment, the step of performing cross-regional feature fusion processing on the behavior data, transaction data and device feature data to generate a cross-border user portrait comprises:

[0017] De-identifying the page stay duration of the user in multiple countries or regions and the language preference to generate a user behavior link code containing multiple language preferences;

[0018] Associating and matching the transaction data with the click behavior sequence to generate a user commodity category preference label;

[0019] Extracting the user cross-time zone active features based on the device time zone, IP address jump frequency and network proxy feature;

[0020] Inputting the user behavior link code, user commodity category preference label and user cross-time zone active features into a preset federated learning model to generate a cross-border user portrait.

[0021] In an embodiment, the step of obtaining the exchange rate data, tariff data and logistics cost data of the target region, and calculating the advertisement delivery economic parameters based on the exchange rate data, tariff data and logistics cost data comprises:

[0022] Real-time obtaining exchange rate fluctuation data through a central bank interface of the target region;

[0023] Calling the target country's General Administration of Customs API interface to obtain HS code corresponding customs duty rate data;

[0024] Obtaining logistics cost data through a logistics service provider interface; the logistics cost data includes real-time shipping fees and estimated delivery time.

[0025] In an embodiment, the step of calculating the advertising economic parameter based on the matching degree of the advertisement and the user, the exchange rate fluctuation data, the customs duty rate data, and the logistics cost data specifically includes:

[0026] The advertising economic parameter is calculated according to the following formula:

[0027] ;

[0028] Wherein, represents the advertising economic parameter; represents the matching degree of the advertisement and the user; represents the exchange rate fluctuation data; represents the customs duty rate data; represents the real-time shipping fees; represents the estimated delivery time; represents the exchange rate gain weight; represents the customs duty loss weight; represents the logistics shipping fee weight; represents the logistics time efficiency weight.

[0029] In an embodiment, the step of generating the original advertisement based on the cross-border user portrait and the advertising economic parameter, and filtering and converting the original advertisement according to the legal library and the cultural library of the target region to generate the final advertisement includes:

[0030] Based on the advertising economic parameter and the commodity category preference label in the cross-border user portrait, the corresponding advertisement template is called from the preset advertisement material library to generate the original advertisement;

[0031] Analyzing the advertisement review rules in the legal library of the target region, the original advertisement is detected for sensitive words and filtered for prohibited content;

[0032] Through the taboo color library, religious symbol library and festival custom library in the cultural library of the target region, the compliance of the visual elements of the advertisement is checked to generate the final advertisement.

[0033] In an embodiment, the step of matching the final advertisement to the target region's advertisement server cluster for advertising includes:

[0034] Adjusting the format parameters of the final advertisement according to the terminal device resolution distribution data of the target region, including video code rate, multilingual subtitle embedding position and interactive button size;

[0035] Converting the final advertisement into a standardized format meeting the requirements of the advertisement server cluster of the target region, and distributing it to the target time zone server node meeting the cross-time zone active characteristics of the user.

[0036] In an embodiment, after the step of matching the final advertisement to the advertisement server cluster of the target region for advertisement delivery, the method further comprises:

[0037] Collecting the exposure, click rate and conversion rate of the final advertisement;

[0038] Adjusting the user cross-time zone active characteristic weight of the federated learning model based on the exposure, click rate and conversion rate to generate an advertisement optimization coefficient;

[0039] Matching the exposure to the decay function of the active user amount of the target region time zone, analyzing the click rate with the cosine similarity of the product category preference label, and checking the conversion rate with the Pearson correlation of the advertisement delivery economic parameter to generate a multi-dimensional evaluation matrix;

[0040] Optimizing the final advertisement according to the advertisement optimization coefficient and the multi-dimensional evaluation matrix.

[0041] In addition, in order to achieve the above-mentioned purpose, the application also proposes a cross-border advertisement accurate delivery system based on user portrait, which comprises:

[0042] A first data acquisition module is used to acquire behavior data, transaction data and device feature data of a user;

[0043] A cross-border user portrait generation module is used to perform cross-regional feature fusion processing on the behavior data, transaction data and device feature data to generate a cross-border user portrait;

[0044] A second data acquisition module is used to determine the matching degree of an advertisement and a user according to the cross-border user portrait, and acquire exchange rate fluctuation data, tariff rate data and logistics cost data of a target region;

[0045] An advertisement delivery economic parameter calculation module is used to calculate the advertisement delivery economic parameter based on the matching degree of the advertisement and the user, the exchange rate fluctuation data, the tariff rate data and the logistics cost data;

[0046] A final advertisement generation module is used to generate an original advertisement based on the cross-border user portrait and the advertisement delivery economic parameter, and filter and format convert the original advertisement according to the legal library and the cultural library of the target region to generate a final advertisement;

[0047] The ad delivery module is used to match the final ad to the ad server cluster in the target region for ad delivery.

[0048] Furthermore, to achieve the above objectives, this application also proposes a cross-border advertising precision targeting device based on user profiles. The device includes: a memory, a processor, and a cross-border advertising precision targeting program based on user profiles stored in the memory and executable on the processor. The cross-border advertising precision targeting program based on user profiles is configured to implement the steps of the cross-border advertising precision targeting method based on user profiles.

[0049] The cross-border advertising precision targeting method, system, and device proposed in this application integrate cross-regional behavioral data to generate accurate user profiles, dynamically calculate advertising targeting strategies by combining exchange rates, tariffs, and logistics economic parameters, and make compliance adaptations based on legal and cultural rules. This solves the problems of single data dimensions, inaccurate user profiles, and missing economic variables in the prior art, and has the advantages of improving the accuracy of cross-border advertising, conversion effect, and reducing compliance risks. Attached Figure Description

[0050] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0051] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a flowchart illustrating an embodiment of the cross-border advertising precision targeting method based on user profiles provided in this application.

[0053] Figure 2 For this application Figure 1 A detailed flowchart of step S100;

[0054] Figure 3 For this application Figure 1 A detailed flowchart of step S200;

[0055] Figure 4 For this application Figure 1 Detailed flowchart of step S300;

[0056] Figure 5 For this application Figure 1 A detailed flowchart of step S500;

[0057] Figure 6 For the detailed flowchart of step S600 in the present application Figure 1 For the detailed flowchart of step S600 in the present application

[0058] Figure 7 For the detailed flowchart of step S600 in the present application

[0059] Figure 8 For the detailed flowchart of step S600 in the present application

[0060] Figure 9 For the detailed flowchart of step S600 in the present application

[0061] Explanation of reference numerals:

[0062] 10, cross-border advertisement precision delivery system based on user portrait; 100, first data acquisition module; 200, cross-border user portrait generation module; 300, second data acquisition module; 400, advertisement delivery economic parameter calculation module; 500, final advertisement generation module; 600, advertisement delivery module; 20, memory; 30, processor.

[0063] The purpose of the present application, functional characteristics and advantages will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0064] The technical solutions in the present application will be described clearly and completely below in combination with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. The components of the present application described and shown in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.

[0065] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. Meanwhile, in the description of the present application, the terms "first", "second", etc. are only used for differentiation, and cannot be understood as indicating or implying relative importance.

[0066] In the prior art, cross-border ad placement mainly relies on the user's geographic location or single-dimensional browsing records for targeting. Due to the significant differences in user behavior patterns and consumption preferences in different regions, ad pushing based on simple location information cannot accurately capture the user's real needs. For example, when a cross-border e-commerce platform faces users in multiple regions, it may fail to match the ad content with the user's actual willingness to pay due to not considering the impact of exchange rate fluctuations on product pricing. Meanwhile, without integrating device feature data, it may not be able to identify the user's real location, resulting in incompatible ad formats with the resolution of local terminal devices.

[0067] To solve the above problems, the skilled person realizes the need to build a multi-dimensional user portrait and introduce an economic factor optimization model. Through analysis, it is found that the prior art does not correlate user behavior data with device features across regions, making it impossible to identify the active features of users when switching between multiple regions. Meanwhile, ad placement strategies do not dynamically incorporate economic parameters such as exchange rates and tariffs, leading to an imbalance between cost and benefit. Based on this, the inventors propose to fuse user behavior link encoding with device features and establish an economic parameter calculation model, thereby simultaneously achieving content matching degree improvement and placement cost optimization during ad generation.

[0068] Therefore, the present application proposes a cross-border ad precise placement method based on user portrait, referring to Figure 1 , the method comprises steps S100-S600, wherein:

[0069] Step S100, obtaining user behavior data, transaction data, and device feature data;

[0070] Step S200, performing cross-regional feature fusion processing on the behavior data, transaction data, and device feature data to generate a cross-border user portrait;

[0071] Step S300, determining the matching degree of the ad and the user according to the cross-border user portrait, and obtaining exchange rate fluctuation data, tariff rate data, and logistics cost data of the target region;

[0072] Step S400, calculating ad placement economic parameters based on the matching degree of the ad and the user, the exchange rate fluctuation data, the tariff rate data, and the logistics cost data;

[0073] Step S500, generating an original ad based on the cross-border user portrait and the ad placement economic parameters, and filtering and format-converting the original ad according to the target region's legal library and cultural library to generate a final ad;

[0074] Step S600, matching the final ad to the target region's ad server cluster for ad placement.

[0075] In this embodiment, cross-regional feature fusion processing refers to de-identification processing of behavior data in multiple countries or regions to generate user behavior link encoding containing multi-language preferences. For example, a federated learning model can be used to perform encrypted calculation on user behavior data scattered in different servers, preserving user cross-time zone active features while avoiding data privacy leakage. The advertisement placement economic parameter refers to a quantitative index formed by combining the advertisement matching degree and the economic factors of the target region. Specifically, the exchange rate fluctuation data, the tariff rate data, and the logistics cost data can be weighted and calculated by preset weight coefficients to dynamically evaluate the benefit-risk ratio of advertisement placement. The legal library and the cultural library refer to databases that store advertisement compliance rules and visual taboos in the target region. For example, sensitive word rules in legal texts can be analyzed by natural language processing technology, or religious symbol taboos in advertisement materials can be detected by image recognition algorithms.

[0076] Specifically, the behavior data such as page dwell time and click sequence generated by users in multiple regions are de-identified to form behavior link encoding reflecting the real interest preferences of users. Device time zone and IP jump frequency data are used to extract user cross-time zone active features, such as identifying the user's behavior pattern of using European IP addresses at night in the GMT+8 time zone. After associating the product categories in the transaction data with the click behavior, product category preference labels are generated, forming the core dimension of the user portrait. When calculating the advertisement placement economic parameter, exchange rate fluctuation data can affect the accuracy of the conversion of product prices in advertisements in real time, tariff rate data determines whether the customs clearance cost is within the acceptable range of users, and logistics cost is related to the influence of the expected delivery time on user purchase decisions. In the final advertisement generation stage, the prohibited goods keywords filtering mechanism in the legal library will delete the product descriptions that do not comply with local import and export controls, and the holiday custom data in the cultural library will adjust the visual elements of the advertisement to conform to the cultural habits of the target region.

[0077] Compared with the prior art, the existing scheme usually processes user behavior data and economic environment parameters independently, resulting in a mismatch between advertisement content and user payment ability. For example, traditional methods may only recommend high-tariff goods based on user click records, without combining tariff data to calculate the actual take-home price, resulting in high click-through rate but low conversion rate. The present scheme can identify potential consumer demand of users in multiple regions by federated learning model fusion of multi-regional features, and at the same time, economic parameters are included in the matching degree calculation to ensure that the recommended advertisements meet user expectations in terms of price sensitivity, logistics timeliness, etc.

[0078] By the technical solution, the real demand of the user in the multi-region switching can be accurately identified, and the price information and the delivery scheme in the advertisement content are dynamically adjusted. For example, when the user equipment displays frequent switching of IP addresses in China, Japan and South Korea, the system automatically matches the commodity version with the lowest tariff rate in the three countries, and highlights the cross-border free shipping information in the advertisement. At the same time, the legal database filtering mechanism is used to avoid pushing the commodity advertisement prohibited to be imported in the target region, and the risk of violation is reduced. Finally, the generated advertisement format is adapted to the resolution of the mainstream equipment in the target region, the display effect of the advertisement material is ensured, and the user interaction experience is improved.

[0079] In a possible implementation, the reference Figure 2 , the step S100 comprises steps S110-S130, wherein

[0080] In step S110, the behavior data is collected through a browser API; the behavior data comprises page staying time, click behavior sequence and language preference of the user in multiple countries or regions;

[0081] In step S120, the transaction data is obtained through a payment gateway interface; the transaction data comprises transaction currency, transaction amount and commodity category;

[0082] In step S130, the device feature data is obtained through a terminal device used by the user; the device feature data comprises device time zone, IP address jump frequency and network proxy feature.

[0083] In the embodiment, the browser API collects the behavior data, which means that the application program interface provided by the browser is used to capture the interactive information of the user on the webpage. Specifically, the JavaScript event listening mechanism can be used to realize the tracking of the page staying time and the click event sequence, for example, the addEventListener function is used. The payment gateway interface obtains the transaction data, which means that the data channel is established with the third-party payment platform to extract the transaction record. Specifically, the OAuth 2.0 protocol can be used for authorized access, so as to obtain the encrypted transaction currency, amount and commodity category information. The terminal device obtains the device feature data, which means that the device properties are analyzed through the underlying interface of the operating system or the network protocol. Specifically, the Settings.System class provided by the Android system can be used to read the device time zone, and the IP address jump frequency and the proxy server identifier can be analyzed through the TCP / IP data packet.

[0084] Specifically, the behavior data is collected in real time through a browser API, which can cover the dynamic interaction behavior of the user in multiple regional pages. For example, the difference between the user's stay time in the European region page and the click sequence in the Asian region page can reflect the cross-regional preference. The transaction data is obtained through a payment gateway interface, ensuring the integrity and timeliness of the transaction currency and commodity category information. For example, the record of the user using US dollars to purchase electronic products and using euros to purchase clothing products can be associated with different regional consumption habits. The device feature data is directly extracted from the terminal device. The device time zone can be used to determine the user's residence area. The IP address jump frequency and network proxy features can identify whether the user uses a cross-border network access tool, such as a VPN jump, which produces a higher IP address change frequency than regular users.

[0085] Compared with the prior art, the existing cross-border advertising data collection usually relies on single geographic location information or static browsing records, such as determining the user's country of residence only through IP address, which cannot capture the differences in multi-page interaction behavior and transaction preferences. The present scheme can more accurately identify the user's real consumption intention in different regions by integrating multi-dimensional dynamic data sources, such as associating page stay time with transaction currency. In addition, the prior art lacks detection of network proxy features, resulting in placement bias when the user's actual location does not match the IP address. The present scheme can effectively exclude false geographic location interference through IP jump frequency and proxy feature analysis.

[0086] Through the above technical solutions, the present application can solve the problem of one-sidedness of user behavior data in cross-border advertising placement. By integrating dynamic interaction behavior, transaction preferences, and device network features, a more comprehensive data foundation is provided for subsequent generation of cross-border user portraits. For example, the association of language preference and transaction currency can assist in determining the user's multilingual ability and consumption regional tendency. The combination of device time zone and IP jump features can improve the accuracy of identifying the user's real geographic location.

[0087] In a feasible implementation manner, with reference to Figure 3 , step S200 includes steps S210-S240, wherein:

[0088] Step S210, de-identifying the page stay time and language preference of the user in multiple countries or regions to generate a user behavior link code containing multiple language preferences;

[0089] Step S220, associating and matching the transaction data with the click behavior sequence to generate a user commodity category preference label;

[0090] Step S230, extracting the user's cross-time zone active features based on device time zone, IP address jump frequency, and network proxy features;

[0091] In step S240, the user behavior link encoding, the user commodity category preference label, and the user cross-time zone active feature are input into a preset federated learning model to generate a cross-border user portrait.

[0092] In this embodiment, the user behavior link encoding of the multilingual preference refers to the serialization encoding of the language selection and the stay duration of the user on the pages in different countries by a hash algorithm. Specifically, the SHA-256 algorithm can be used to combine the page language code and the stay timestamp to generate an irreversible identifier, which is used to eliminate the user identity sensitive information while retaining the cross-regional behavior features. The user commodity category preference label refers to the time sequence matching of the commodity categories in the transaction records and the commodity display pages in the click behavior. Specifically, the Apriori association rule algorithm can be used to mine the probability threshold of the transaction generated after the user clicks the commodity page to generate the strongly associated commodity category label. The user cross-time zone active feature refers to the calculation of the user active period distribution through the deviation value of the device time zone and the IP address geographical location. Specifically, the sliding time window can be used to count the time period in which the IP address switching frequency exceeds the set threshold as the cross-border activity hot period. The preset federated learning model refers to a distributed machine learning framework deployed in multiple geographical region servers. Specifically, the transverse federated learning architecture can be used to realize the encrypted aggregation of the user feature vectors in each region, and the shared model parameters are updated through the gradient exchange method.

[0093] Specifically, first, the language preference and the stay duration of the user on the pages in multiple countries are de-identified, for example, the stay duration of the user selecting English on the US page and the stay duration of the user selecting Japanese on the Japanese page are respectively hashed encoded to form the behavior link sequence in the cross-language dimension. Then, the click timestamp of the user on the commodity detail page is matched with the transaction timestamp returned by the payment gateway, for example, the user completes the cross-border payment of the mobile phone commodity within three minutes after browsing the electronic product page, and then the preference label of the electronic product category is generated. At the same time, based on the difference between the device time zone and the actual geographical location of the IP address, for example, the device is set to East 8 zone but the IP address frequently switches between UTC+1 to UTC+3 time zone, combined with the VPN usage record, it is judged that the user has the activity feature of cross-European time zone. Finally, the feature vectors processed above are input into the federated learning model, for example, the homomorphic encryption technology is used to jointly train the user features in the European Union and the user features in Southeast Asia to generate the user portrait that can reflect the cross-border consumption habits.

[0094] Compared with the prior art, the traditional method only constructs the user portrait through the browsing data of a single geographic location, and cannot identify the real preferences of the user in a multi-language environment, for example, misjudging the English browsing behavior of the user on the page in the United Kingdom as the local user demand. However, the present scheme can effectively associate the behavior track of the user on the page in multiple countries with the real transaction data through the fusion of cross-regional features, for example, identifying that the user uses English to browse the goods but the actual purchase preference points to the characteristic goods of a specific region. The user active period analysis in the prior art is usually based on the local time of the device, while the present scheme can accurately capture the real active period of the user in cross-country travel or cross-border shopping through the IP address hopping frequency and time zone deviation detection.

[0095] Through the above technical scheme, the present application solves the problem of insufficient portrait accuracy caused by the fragmentation of cross-border user behavior features, and realizes the organic integration of multi-language behavior data and cross-time zone activity features. Specifically, the balance between user privacy protection and behavior feature availability is achieved through de-identification processing, and the completeness of the cross-border user portrait is improved under the premise of protecting data privacy by using a federated learning model, so that the generated user portrait can accurately reflect the consumption preferences and activity rules of the user in multiple countries or regions, providing reliable data support for subsequent advertisement matching.

[0096] In a feasible implementation manner, with reference to Figure 4 , the step S300 includes steps S310-S330, in which:

[0097] In step S310, exchange rate fluctuation data is acquired in real time through a central bank interface of the target region;

[0098] In step S320, HS code corresponding customs duty rate data is acquired by calling a customs general administration API interface of the target country;

[0099] In step S330, logistics cost data is acquired through a logistics service provider interface; the logistics cost data includes real-time freight and predicted delivery time length.

[0100] In the embodiment, the exchange rate fluctuation data refers to the change amount of the real-time exchange rate between the currency of the target region and the local currency of the advertiser, which can be achieved by periodically grabbing data through the exchange rate API interface provided by the central bank, and is used for dynamically evaluating the currency exchange loss and gain in the advertisement launching cost. The customs duty rate data refers to the import tax rate of a specific commodity category in the target country, which can be matched with the tax rate by using the HS code query interface provided by the customs general administration, and is used for calculating the tax cost of the advertisement promotion goods entering the target market. The logistics cost data refers to the real-time freight and delivery time of the goods from the warehouse of the advertiser to the target region, which can be acquired through the logistics service provider interface to obtain the cost and time efficiency parameters under different transportation modes, and is used for comprehensively evaluating the influence of the delivery link on the advertisement launching decision.

[0101] Specifically, exchange rate fluctuation data is collected through a central bank interface at preset time intervals, such as every 5 minutes, ensuring the real-time nature of the advertising launch economic parameters. Tariff rate data is mapped to the advertising commodity category through HS coding, such as clothing commodities corresponding to HS coding chapters 61-63, and the corresponding tariff value is returned by calling the General Administration of Customs API. Logistics cost data is input through a logistics service provider interface, including the weight, volume, and destination coordinates of the commodity, and returns the freight and estimated delivery time of the currently available transportation channel. After integrating these three types of data, the advertising and user matching degree is combined, and the advertising launch economic parameters are generated through weighted calculation, providing cost-benefit evaluation basis for subsequent advertising generation.

[0102] Compared with the prior art, traditional cross-border advertising usually only uses fixed exchange rates and historical tariff data for cost estimation, resulting in economic parameter calculation lagging behind market changes. However, the present scheme realizes real-time synchronous acquisition of multi-dimensional data through the central bank interface, the General Administration of Customs API, and the logistics service provider interface, which can dynamically reflect the impact of exchange rate fluctuations, policy adjustments, and changes in logistics capacity on advertising launch. For example, when the currency of the target region suddenly depreciates, real-time exchange rate data can immediately trigger the recalculation of advertising launch economic parameters, avoiding errors in launch costs caused by delayed updates of exchange rates.

[0103] Through the above technical scheme, the present application can accurately quantify the economic impact factors in cross-border advertising launch, solving the cost calculation deviation problem caused by data update lag in traditional methods. By real-time acquisition of exchange rate, tariff, and logistics data, the advertising launch economic parameters can dynamically reflect the changes in the market environment of the target region, enabling the advertising generation link to automatically avoid high-cost launch regions caused by sudden economic fluctuations, while preferentially selecting promotion paths with stable logistics time efficiency and low tax fees, improving the input-output ratio of advertising launch.

[0104] In one possible implementation, the step of calculating the advertising launch economic parameters based on the matching degree of the advertisement and the user, the exchange rate fluctuation data, the tariff rate data, and the logistics cost data specifically includes:

[0105] The advertising launch economic parameters are calculated according to the following formula:

[0106] ;

[0107] wherein, represents the advertising launch economic parameters; represents the matching degree of the advertisement and the user; represents the exchange rate fluctuation data; represents the tariff rate data; represents the real-time freight; represents the estimated delivery time. Indicates the exchange rate gain weight; Indicates tariff loss weight; Indicates the weight of logistics freight costs; This indicates the weight of logistics timeliness.

[0108] In this embodiment, the matching degree between advertisements and users refers to the degree of correlation between advertisement content and user preferences. Specifically, it can be calculated using a cosine similarity algorithm between user behavior link encoding and advertisement tags, reflecting the advertisement's attractiveness to the target user. Exchange rate fluctuation data refers to the real-time exchange rate change between the target region's currency and the base currency. Specifically, it can be calculated using the exchange rate midpoint difference from the central bank interface, measuring the impact of cross-border transaction cost fluctuations on advertising revenue. Tariff rate data refers to the import tariff rates levied by the target country on specific product categories. Specifically, it can be obtained by querying the tariff rates corresponding to HS codes through the General Administration of Customs API interface, assessing the compliance costs after goods enter the country. Real-time shipping costs refer to the logistics costs required to transport goods from the advertiser's location to the target region. Specifically, it can be obtained by obtaining real-time quotes through the logistics service provider interface, calculating the impact of logistics costs on user purchase intentions. Estimated delivery time refers to the time period from goods being shipped to user receipt, predicted using historical data from logistics service providers combined with current capacity status, assessing the negative effect of logistics timeliness on advertising conversion rates. The weights of exchange rate gains, tariff losses, logistics costs, and logistics timeliness refer to the influence coefficients of different economic factors on the economics of advertising. These can be determined through multiple regression analysis of historical advertising data and used to dynamically adjust the proportion of each factor in the evaluation of advertising economics.

[0109] Specifically, in calculating the economic parameters for ad placement, the matching degree between the ad and the user is first used as a basic parameter, reflecting the degree to which the ad aligns with user needs. Next, exchange rate fluctuation data is collected in real-time through the central bank interface of the target region, and the impact of exchange rate changes on the economic efficiency of the ad is calculated using exchange rate gain weights. Simultaneously, the current tariff rate data is obtained from the General Administration of Customs API, and the tax cost of goods entering the country is assessed based on tariff loss weights. Furthermore, real-time freight costs and estimated delivery times are collected through logistics service provider interfaces, and the negative impact of logistics costs on user decisions is quantified by combining logistics cost weights and logistics timeliness weights. Finally, these parameters are substituted into a preset economic parameter calculation formula to dynamically generate the economic parameters for ad placement, providing a quantitative basis for subsequent ad generation and placement decisions.

[0110] Compared with the prior art, the traditional advertisement economy evaluation usually only relies on a single index such as advertisement click rate or conversion rate, and does not consider economic factors specific to cross-border scenarios such as exchange rate fluctuations, tariff rates and logistics costs. For example, the prior art may only adjust the advertisement placement strategy according to the user click behavior, but does not integrate the influence of changes in tariff rates on the final selling price of goods, resulting in that the actual conversion rate after advertisement placement is lower than expected. The scheme introduces a multi-dimensional economic parameter calculation model, and integrates the exchange rate risk, tariff cost and logistics time limit in cross-border transactions into the advertisement economy evaluation system, so as to realize dynamic optimization of the advertisement placement strategy.

[0111] Through the above technical scheme, the application can effectively solve the problem of advertisement revenue loss caused by ignoring the fluctuation of economic environment in traditional cross-border advertisement placement. By quantifying the influence of exchange rate fluctuations on transaction costs, the restriction of tariff rates on commodity pricing and the negative effect of logistics time limit on user experience, the advertisement placement economic parameter can dynamically reflect the changes of the economic environment of the target market, thereby helping the advertiser to avoid high-risk areas in advance when generating advertisements and optimizing the allocation of advertisement placement resources.

[0112] In a feasible implementation manner, with reference to Figure 5 , the step S500 includes steps S510-S530, wherein:

[0113] Step S510, based on the advertisement placement economic parameter and the commodity category preference label in the cross-border user portrait, a corresponding advertisement template is called from a preset advertisement material library, and an original advertisement is generated in combination with the advertisement placement economic parameter;

[0114] Step S520, the advertisement review rules in the legal library of the target area are analyzed, and the original advertisement is subjected to sensitive word detection and prohibited content filtering;

[0115] Step S530, the advertisement visual elements are subjected to compliance verification through the taboo color library, religious symbol library and festival custom library in the culture library of the target area, and a final advertisement is generated.

[0116] In this embodiment, the advertisement material library refers to a database storing different advertisement templates and materials, which can be implemented by using a distributed storage system combined with a label classification index, for quickly matching advertisement templates according to commodity category preference labels. The legal library refers to a data set containing target region advertisement regulations, which can be implemented by obtaining government public files or accessing a third-party compliance database, for detecting whether the advertisement content meets local legal requirements. The culture library refers to a knowledge base storing target region cultural taboo information, which can be implemented by using a multi-modal data collection tool to integrate religious symbols, holiday customs and color taboo data, for verifying the compliance of advertisement visual elements. The sensitive word detection refers to identifying illegal words in the advertisement text, which can be implemented by using a natural language processing model combined with regular expression matching rules, for filtering content that violates laws or cultural taboos. The compliance verification refers to verifying whether the advertisement image, color and symbol meet the target region cultural habits, which can be implemented by using an image recognition algorithm combined with a rule engine, for avoiding conflicts caused by cultural differences.

[0117] Specifically, this step first matches user preferences with advertisement templates through the advertisement material library, and generates original advertisement content combined with economic parameters. Then, the advertisement text is scanned for sensitive words and filtered for prohibited goods using the review rules of the legal library, for example, to detect whether there is false propaganda or information about restricted sales goods. Next, the advertisement main tone is verified by calling the taboo color library in the culture library, and if the target region taboo color is found, it is automatically replaced with a safe color value; at the same time, potential conflict elements in the advertisement pattern are identified by the religious symbol library, for example, to remove graphics that are not consistent with local beliefs. Finally, the time nodes in the advertisement script are adjusted according to the holiday custom library, for example, to insert relevant blessing words during specific festivals. After multiple filtering and format conversion, the final advertisement that meets the legal specifications and cultural customs of the target region is generated.

[0118] Compared with the prior art, the existing cross-border advertisement generation method usually only relies on simple keyword matching or geographic location information, without considering the synergistic effect of legal review and cultural adaptation. For example, the traditional scheme may only push advertisements according to IP addresses, but cannot identify religious symbol conflicts or color taboos implied in the advertisement content. However, the present scheme realizes multi-dimensional matching of advertisement content and target region specifications through the dual verification mechanism of the legal library and the culture library, solving the compliance risk problem caused by single-dimensional filtering.

[0119] Through the above technical solutions, the present application can effectively avoid the problem of advertisement violation caused by legal differences or cultural misunderstandings, and improve the legality and cultural adaptability of advertisement content in the target region. At the same time, through the automatic filtering and format conversion process, the cost of manual review is reduced, and the efficiency of cross-border advertisement delivery is improved.

[0120] In a feasible implementation manner, reference is made to Figure 6The step S600 includes steps S610-S620, wherein:

[0121] In step S610, the format parameters of the final advertisement are adjusted according to the terminal device resolution distribution data of the target region, including video code rate, multi-language subtitle embedding position, and interactive button size.

[0122] In step S620, the final advertisement is converted into a standardized format required by the target region advertisement server cluster, and is distributed to a target time period server node that meets the cross-time zone active characteristics of the user.

[0123] In this embodiment, the terminal device resolution distribution data of the target region refers to a set of screen resolution information of the user terminal devices in the target region, which can be obtained by analyzing the screen parameters in the device feature data or by calling the interface of a third-party data analysis platform, and is used to dynamically match the display specifications of the advertisement to avoid interface adaptation abnormalities.

[0124] The video code rate adjustment refers to dynamically adjusting the compression rate of the video advertisement according to the resolution distribution peak value, which can be realized by using the hierarchical compression algorithm of the H.265 encoding protocol to ensure the smoothness of the playback on devices with different resolutions.

[0125] The multi-language subtitle embedding position optimization refers to automatically adjusting the subtitle display area according to the screen ratio of the device, which can be realized by using the edge detection algorithm of the OpenCV image processing library to prevent the subtitles from blocking key visual elements.

[0126] The interactive button size calibration refers to setting the minimum pixel threshold of the touch button according to the resolution interval, which can be realized by using the breakpoint adaptation mechanism of the responsive layout framework to ensure the accuracy of user operation.

[0127] The cross-time zone active characteristics of the user refer to the active time period data of the user in multiple time zones, which can be realized by associating the device time zone with the timestamp of the server log, and is used to match the timeliness window of the advertisement delivery.

[0128] Specifically, in the advertisement delivery stage, first, the resolution statistical data of the terminal equipment in the target area is obtained, for example, the resolution distribution histogram is formed by sampling the screen width and height parameters in the equipment feature data. Based on the histogram, the mainstream resolution interval is determined, and the dynamic code rate adjustment technology is used to adaptively compress the video advertisement, for example, the 1080p video is switched to 720p code rate when a low-resolution device is detected. At the same time, according to the screen aspect ratio, the coordinate offset calculation of the subtitle position is carried out, for example, the subtitle is placed in the bottom central area on a 16:9 screen, and on an 18:9 screen, it is moved up by 5% of the display height. The size of the interactive button is scaled according to the physical size density of the resolution interval, for example, the button height is set to be not less than 48 pixels on a 5-inch screen. After completing the format adjustment, the advertisement content is packaged into a standardized transmission format, such as the HLS streaming protocol or the AMP HTML format, and then according to the active period recorded in the user's cross-time zone active characteristics, the advertisement is distributed to the edge server node of the corresponding time zone, for example, when the user is active at 20:00-22:00 Beijing time, the advertisement is preloaded into the East Eight Zone CDN node cache queue.

[0129] Compared with the prior art, the traditional advertisement delivery usually adopts a fixed resolution template or a single time zone delivery strategy, resulting in picture cropping on low-resolution devices or pushing invalid advertisements in high-latency periods. The present scheme realizes dynamic format adaptation through resolution distribution data analysis, for example, when 40% of the users in a certain area use 720x1280 resolution devices, the advertisement version adapted to this resolution is preferentially generated, and at the same time, the precise scheduling of the server node is realized in combination with the cross-time zone active characteristics, for example, according to the user's historical behavior, the advertisement for users in the western United States is pushed to the Los Angeles server 2 hours in advance.

[0130] Through the above technical scheme, the present application solves the problem of mispositioning of advertisement elements caused by poor display adaptation of devices in different areas, avoids the waste of advertisement exposure caused by time zone matching errors, significantly improves the rendering completeness of advertisement materials and the timeliness of user reach, thereby increasing effective click behavior and reducing server resource redundancy consumption.

[0131] In a feasible implementation manner, with reference to Figure 7 After step S600, the method further includes steps S710-S740, wherein:

[0132] Step S710, the exposure, click rate and conversion rate of the final advertisement are collected;

[0133] Based on the exposure, click rate and conversion rate, the user cross-time zone active characteristic weight of the federated learning model is adjusted to generate an advertisement optimization coefficient;

[0134] The exposure amount is matched with a target region period active user amount by a decay function, a click rate is analyzed by cosine similarity with a commodity category preference label, and a conversion rate is checked by Pearson correlation with an advertisement launching economic parameter to generate a multi-dimensional evaluation matrix;

[0135] The final advertisement is optimized according to the advertisement optimization coefficient and the multi-dimensional evaluation matrix.

[0136] In this embodiment, the exposure amount refers to the number of times an advertisement is displayed in a target region, which can be specifically realized by an advertisement server log analysis tool to collect data and reflect the advertisement coverage. The click rate refers to the ratio of the number of times an advertisement is clicked to the exposure amount, which can be specifically realized by front-end burying point technology to track and count and measure the advertisement content attraction. The conversion rate refers to the proportion of users who complete a purchase or registration behavior after clicking, which can be specifically realized by a payment interface callback or a behavior tracking SDK to monitor and evaluate the actual effect of the advertisement. The federal learning model weight adjustment refers to dynamically correcting the contribution of user cross-time zone active features in the portrait according to advertisement effect feedback, which can be specifically realized by a gradient descent algorithm to update parameters and optimize user behavior prediction accuracy. The multi-dimensional evaluation matrix refers to a comprehensive evaluation system integrating exposure decay matching, preference similarity analysis and economic parameter correlation checking, which can be specifically realized by a matrix operation engine to realize data correlation analysis and identify the optimization direction of the advertisement launching strategy.

[0137] Specifically, after the advertisement launching is completed, the exposure amount data is collected in real time through the log system of the advertisement server cluster, and the click rate and the conversion rate are counted by using the user behavior tracking module. The federal learning model automatically adjusts the weight parameters of the user cross-time zone active features according to the advertisement optimization coefficient, for example, when the conversion rate of a certain period is significantly improved, the weight proportion of the active features in the period is correspondingly increased. At the same time, the matching degree of the exposure amount and the period active user amount is calculated by using the decay function to identify whether the advertisement launching period overlaps with the real active period of the user; the relevance of the click rate and the commodity category preference label is analyzed by using the cosine similarity to verify whether the advertisement content matches the user interest; the relationship between the conversion rate and the advertisement launching economic parameter is checked by using the Pearson correlation to judge whether the cost control strategy affects the conversion effect. Finally, the multi-dimensional evaluation results are input into the optimization algorithm to generate advertisement material replacement, period adjustment or economic parameter correction instructions.

[0138] Compared with the prior art, the existing advertisement optimization method usually only relies on a single indicator of click rate or conversion rate for strategy adjustment, and cannot effectively identify complex problems such as period matching deviation, interest deviation or cost control overload. The present scheme dynamically optimizes the user behavior feature weight by using the federal learning model, constructs a multi-dimensional evaluation system by combining the decay function, the similarity analysis and the correlation checking, and can simultaneously solve the problems of advertisement launching period mispositioning, insufficient content attraction and cost-benefit imbalance.

[0139] Through the technical solution, the application can dynamically optimize the advertisement placement strategy according to real-time effect data, balance the economic cost constraint while ensuring the user interest matching degree, effectively improve the precision and conversion efficiency of cross-border advertisements, and avoid resource waste or user experience decline caused by single index optimization.

[0140] It should be noted that the above examples are only used to understand the application and do not constitute a limitation on the cross-border advertisement accurate placement method based on user portrait of the application. More forms of simple transformation based on this technical concept are within the protection scope of the application.

[0141] The application also provides a cross-border advertisement accurate placement system 10 based on user portrait, referring to Figure 8 , the system comprises:

[0142] The first data acquisition module 100 is configured to acquire the behavior data, transaction data and device feature data of the user.

[0143] The cross-border user portrait generation module 200 is configured to perform cross-regional feature fusion processing on the behavior data, transaction data and device feature data to generate a cross-border user portrait.

[0144] The second data acquisition module 300 is configured to determine the matching degree of the advertisement and the user according to the cross-border user portrait, and acquire the exchange rate fluctuation data, tariff rate data and logistics cost data of the target region.

[0145] The advertisement placement economic parameter calculation module 400 is configured to calculate the advertisement placement economic parameter based on the matching degree of the advertisement and the user, the exchange rate fluctuation data, the tariff rate data and the logistics cost data.

[0146] The final advertisement generation module 500 is configured to generate an original advertisement based on the cross-border user portrait and the advertisement placement economic parameter, and filter and format convert the original advertisement according to the legal library and the cultural library of the target region to generate a final advertisement.

[0147] The advertisement placement module 600 is configured to match the final advertisement to the advertisement server cluster of the target region for advertisement placement.

[0148] In this embodiment, the first data acquisition module 100 refers to the device for collecting user multi-territory behavior and transaction records through the browser interface and payment gateway. Specifically, it can be implemented by adopting the joint calling mechanism of browser API and payment gateway interface, which is used to eliminate the regional coverage bias caused by single data source. The cross-border user portrait generation module 200 refers to the processor 30 that fuses cross-territory behavior links and transaction preferences. Specifically, it can be implemented by adopting the multi-modal feature fusion algorithm under the federal learning framework, which is used to solve the problem of user behavior fragmentation in cross-border scenarios. The advertisement delivery economic parameter calculation module 400 refers to the operation unit that dynamically integrates exchange rate fluctuations and logistics costs. Specifically, it can be implemented by adopting the weighted economic indicator dynamic adjustment algorithm, which is used to avoid the implicit cost risk in cross-border transactions. The final advertisement generation module 500 refers to the advertisement compliance device combined with legal rules and cultural taboos. Specifically, it can be implemented by adopting multi-level semantic filtering and visual element replacement technology, which is used to meet the content review requirements of different jurisdictions.

[0149] Specifically, the system realizes accurate cross-border advertisement delivery through the collaborative processing mechanism of multi-source heterogeneous data. The first data acquisition module 100 collects multi-dimensional cross-border behavior characteristics of users from browsers, payment systems and terminal devices. The cross-border user portrait generation module 200 uses a federal learning model to convert discrete cross-territory behavior codes into a unified portrait. The second data acquisition module 300 synchronously accesses central bank and customs data interfaces to obtain real-time economic fluctuation parameters of the target region. The advertisement delivery economic parameter calculation module 400 couples user preferences with economic variables through a dynamic weight distribution model to generate delivery parameters containing cost-benefit forecasts. The final advertisement generation module 500 performs double compliance modification on the advertisement content based on the text review rules of the legal library and the visual verification rules of the culture library. The advertisement delivery module 600 implements automatic adaptation and distribution of advertisement formats according to the device characteristics and network environment of the target region.

[0150] Compared with the prior art, the traditional cross-border advertisement system only relies on static geographic location information for delivery, while the present system establishes a multi-dimensional user preference and economic environment correlation model through cross-territory feature fusion and dynamic economic parameter integration. The prior art lacks an automatic adaptation mechanism for legal and cultural differences. The present system realizes real-time compliance verification of advertisement content through a structured legal rule library and a cultural taboo library. The existing system uses a fixed cost accounting mode, and the present system improves the economic benefit prediction accuracy of advertisement delivery through a dynamic weighting algorithm of exchange rate fluctuation data and logistics costs.

[0151] Through the technical solution, the three technical bottlenecks of incomplete user portrait, lagging economic parameters and legal cultural conflicts in cross-border advertising are effectively solved. The cross-regional feature fusion mechanism can accurately identify the cross-border consumption preferences of users, the dynamic economic parameter model can reflect the changes of the implicit cost of the target market in real time, and the multi-level compliance verification system can avoid the failure of advertising due to regional cultural differences. The system significantly improves the accuracy and compliance of cross-border advertising, reduces the economic risks caused by exchange rate fluctuations or tariff adjustments, and reduces the cost of manual adaptation through automatic format conversion.

[0152] The application also provides a cross-border advertising precision placement device based on user portrait, referring to Figure 9 The device comprises a memory 20, a processor 30, and a cross-border advertising precision placement program based on user portrait stored on the memory 20 and executable on the processor 30, which is configured to implement the steps of the cross-border advertising precision placement method based on user portrait.

[0153] The cross-border advertising precision placement device based on user portrait provided by the application adopts the cross-border advertising precision placement method based on user portrait in the above embodiments, which can improve the accuracy of cross-border advertising, conversion effect and reduce compliance risk. Compared with the prior art, the beneficial effects of the cross-border advertising precision placement device based on user portrait provided by the application are the same as those of the cross-border advertising precision placement method based on user portrait provided by the above embodiments, and other technical features in the cross-border advertising precision placement device based on user portrait are the same as those disclosed in the above embodiment method, which will not be repeated here.

[0154] The above only describes some embodiments of the application, and does not limit the patent scope of the application. Any equivalent structural transformation made by using the content of the application specification and drawings, or direct / indirect application in other related technical fields is included in the patent protection scope of the application.

Claims

1. A method for precise cross-border advertising targeting based on user profiles, characterized in that, The method includes: Acquire user behavior data, transaction data, and device characteristic data; The behavioral data, transaction data, and device feature data are subjected to cross-regional feature fusion processing to generate cross-border user profiles. The matching degree between advertisements and users is determined based on the cross-border user profile, and data on exchange rate fluctuations, tariff rates, and logistics costs in the target region are obtained. The economic parameters for advertising placement are calculated based on the matching degree between advertisements and users, exchange rate fluctuation data, tariff rate data, and logistics cost data. The original advertisement is generated based on cross-border user profiles and advertising economic parameters. The original advertisement is then filtered and format-converted according to the legal and cultural databases of the target region to generate the final advertisement. The final advertisement is matched to the ad server cluster in the target region for ad delivery; The steps for obtaining user behavior data, transaction data, and device characteristic data include: The behavioral data is collected via browser API; the behavioral data includes the duration of user stay on pages in multiple countries or regions, click behavior sequences, and language preferences. The transaction data is obtained through a payment gateway interface; the transaction data includes the transaction currency, transaction amount, and product category. The device characteristic data is obtained through the terminal device used by the user; the device characteristic data includes the device time zone, IP address hopping frequency, and network proxy characteristics. The step of performing cross-regional feature fusion processing on the behavioral data, transaction data, and device feature data to generate a cross-border user profile includes: The user's page dwell time and language preferences in multiple countries or regions are de-identified to generate user behavior link codes containing multi-language preferences; The transaction data is correlated and matched with the click behavior sequence to generate user product category preference tags; Based on device time zone, IP address hopping frequency, and network proxy characteristics, extract user cross-time zone activity characteristics; The user behavior link encoding, user product category preference tags, and user cross-time zone activity characteristics are input into a preset federated learning model to generate a cross-border user profile. The steps of obtaining exchange rate data, tariff data, and logistics cost data for the target region, and calculating the economic parameters for advertising placement based on the exchange rate data, tariff data, and logistics cost data, include: Real-time exchange rate fluctuation data can be obtained through the interface of the central bank of the target region. Call the API interface of the General Administration of Customs of the target country to obtain the tariff rate data corresponding to the HS code; Obtain logistics cost data through the logistics service provider's interface; the logistics cost data includes real-time freight rates and estimated delivery time. The steps for calculating the economic parameters of advertising placement based on the matching degree between advertisements and users, exchange rate fluctuation data, tariff rate data, and logistics cost data specifically include: Calculate the economic parameters of advertising placement using the following formula: ； in, Indicates the economic parameters of advertising spending; This indicates the match between the advertisement and the user; This indicates exchange rate fluctuation data; This indicates tariff rate data; Indicates real-time shipping costs; Indicates the estimated delivery time; Indicates the exchange rate gain weight; Indicates tariff loss weight; Indicates the weight of logistics freight costs; This indicates the weight of logistics timeliness.

2. The cross-border advertising precision targeting method based on user profiles as described in claim 1, characterized in that, The steps of generating original advertisements based on cross-border user profiles and advertising economic parameters, and filtering and format-converting the original advertisements according to the legal and cultural databases of the target region to generate the final advertisement include: Based on advertising economic parameters and product category preference tags in cross-border user profiles, the corresponding advertising templates are called from the preset advertising material library, and the original advertisement is generated by combining the advertising economic parameters. Analyze the advertising review rules in the legal database of the target region, and perform sensitive word detection and filtering of prohibited content in the original advertisement; The visual elements of the advertisement are verified for compliance using the taboo color library, religious symbol library, and festival custom library in the target region's cultural database, and the final advertisement is generated.

3. The cross-border advertising precision targeting method based on user profiles as described in claim 1, characterized in that, The step of matching the final advertisement to the ad server cluster in the target region for ad delivery includes: Adjust the format parameters of the final advertisement based on the resolution distribution data of terminal devices in the target area, including video bitrate, embedding position of multilingual subtitles, and size of interactive buttons; The final advertisement is converted into a standardized format that meets the requirements of the target region's ad server cluster and distributed to target time period server nodes that match the user's cross-time zone activity characteristics.

4. The cross-border advertising precision targeting method based on user profiles as described in claim 1, characterized in that, After the step of matching the final advertisement to the ad server cluster in the target region for ad delivery, the method further includes: Collect the final ad impressions, click-through rate, and conversion rate; The weights of user cross-time zone activity features in the federated learning model are adjusted based on the exposure, click-through rate, and conversion rate to generate advertising optimization coefficients. The exposure volume is matched with the number of active users in the target region during the time period using a decay function, the click-through rate is analyzed with the product category preference tags using cosine similarity analysis, the conversion rate is verified with the advertising economic parameters using Pearson correlation, and a multi-dimensional evaluation matrix is ​​generated. The final advertisement is optimized based on the advertising optimization coefficient and the multi-dimensional evaluation matrix.

5. A cross-border advertising precision targeting system based on user profiles, characterized in that, The system includes: The first data acquisition module is used to acquire user behavior data, transaction data, and device characteristic data; The cross-border user profile generation module is used to perform cross-regional feature fusion processing on the behavioral data, transaction data and device feature data to generate cross-border user profiles; The second data acquisition module is used to determine the matching degree between advertisements and users based on the cross-border user profile, and to acquire exchange rate fluctuation data, tariff rate data and logistics cost data of the target region. The advertising placement economic parameter calculation module is used to calculate advertising placement economic parameters based on the matching degree between ads and users, exchange rate fluctuation data, tariff rate data, and logistics cost data. The final ad generation module is used to generate original ads based on cross-border user profiles and advertising economic parameters, and to filter and convert the original ads according to the legal and cultural databases of the target region to generate the final ad. The ad delivery module is used to match the final ad to an ad server cluster in the target region for ad delivery. When the system is running, it can realize the cross-border advertising precision delivery method based on user profiles as described in any one of claims 1 to 4.

6. A cross-border advertising precision targeting device based on user profiles, characterized in that, The apparatus includes: a memory, a processor, and a user profile-based cross-border advertising precision targeting program stored in the memory and executable on the processor, the user profile-based cross-border advertising precision targeting program being configured to implement the steps of the user profile-based cross-border advertising precision targeting method as described in any one of claims 1 to 4.

Images