A cross-platform advertisement intelligent decision and dynamic bidding method and system

Abstract

This invention proposes a cross-platform advertising intelligent decision-making and dynamic bidding method and system. The method includes: receiving real-time bidding requests and reading advertising plan operation data; mapping budget usage ratio and exposure frequency to interval numbers, and constructing a decision state space by combining platform and target identifiers; solidifying the decision state space into a read-only decision benchmark instance and generating a compact identifier; querying weight configuration to calculate a unified bidding intensity level, and determining the platform level number based on the platform type; selecting a parameter template based on the platform level number, calculating the actual bidding parameters, and constructing a bidding request object for sending. This invention achieves end-to-end tracking and deduplication through compact identifiers; and solves the semantic offset problem between heterogeneous platforms through unified decision expression and template mapping, significantly improving the response speed, decision stability, and delivery controllability of cross-platform advertising bidding.

Description

Technical Field

[0001] This invention belongs to the field of marketing, and in particular relates to a cross-platform advertising intelligent decision-making and dynamic bidding method and system. Background Technology

[0002] With the rapid development of internet advertising, ad placement has evolved from a single platform to cross-platform placement covering multiple media types such as search, news feeds, and short videos. Advertisers need to participate in real-time bidding on multiple platforms simultaneously to obtain as much effective exposure and conversion as possible within budget constraints. Existing cross-platform placement systems typically operate in a "separate connection, separate bidding" manner: on the one hand, they obtain the operational status and strategy configuration from the placement side and the platform side; on the other hand, when bidding is triggered, they generate bidding parameters based on the field system of each platform and initiate bidding. In practical engineering processes, such systems commonly suffer from unavoidable structural contradictions: First, the bidding process is highly concurrent and its state changes rapidly. When the same advertising campaign triggers bidding requests multiple times within a short period, key constraints such as budget consumption and frequency counts are continuously updated in different threads and processing stages. This results in inconsistent states referenced by different modules within a single bidding process, making it difficult to align and compare the behavior of the same strategy at different points in time. Second, different platforms have significant differences in bidding fields, billing methods, and optional parameter sets. If cross-platform systems directly output specific bid values ​​as the final decision result, extensive platform adaptation is often required on the execution side. This adaptation process can easily introduce semantic offsets, leading to uncontrollable differences in the execution effect of the same decision intent on different platforms. Third, existing systems lack a transferable structured intermediate expression for the bidding decision process, making it difficult to form unified constraints from "decision formation" to "platform parameter assembly and invocation." When performance fluctuations occur, it is difficult to pinpoint whether the problem stems from inconsistent state references, platform adaptation differences, or the decision logic itself. The aforementioned problems have become more prominent with the increase in the number of platforms, the faster pace of advertising, and the increase in concurrent users, which has constrained the stable operation and consistent execution of cross-platform dynamic bidding systems at the engineering level. Summary of the Invention

[0003] The purpose of this invention is to design a cross-platform advertising intelligent decision-making and dynamic bidding method and system, which is a technical solution that can uniformly constrain decision input, decision expression and execution assembly in cross-platform concurrent bidding scenarios, so as to reduce inconsistencies within the link and improve the controllability of cross-platform execution.

[0004] To achieve the above objectives, a first aspect of the present invention provides a cross-platform advertising intelligent decision-making and dynamic bidding method, the method comprising: Receive real-time bidding requests from advertising platforms and read the cumulative spending amount, total budget, exposure count within the time window, platform type identifier, and target category identifier of the advertising campaign; The ratio of the cumulative consumption amount to the total budget is mapped to a budget interval number, and the exposure count value is mapped to a frequency interval number, thus constructing a decision state space that includes the budget interval number, the frequency interval number, the platform type identifier, and the target category identifier; Write the decision state space into a read-only field of the bidding context object, generate a freeze time identifier, and calculate a compact identifier based on the decision state space to form a decision benchmark instance; Query the target weight configuration table to obtain the weight value corresponding to the target category identifier, calculate the unified bidding intensity level based on the budget interval number, the frequency interval number, the weight value and the compact identifier, and determine the platform level number by combining the platform type identifier, and generate an advertising bidding decision expression containing the compact identifier, the unified bidding intensity level and the platform level number; Select a platform parameter template based on the platform level number, calculate the actual bid parameters based on the unified bidding intensity level, the basic bid of the advertising plan and the template coefficient, and construct a platform bidding request object containing the actual bid parameters and the compact identifier concurrently.

[0005] Further, mapping the ratio of the cumulative consumption amount to the total budget to a budget interval number includes: The ratio of the cumulative consumption amount to the total budget is used to obtain the budget utilization rate. Based on the interval division rules pre-loaded into memory, determine the threshold interval into which the budget usage ratio falls; The preset number corresponding to the threshold interval that falls into the range is assigned to the budget interval number.

[0006] Further, the calculation of the compact identifier based on the decision state space includes: The budget interval number, the frequency interval number, the platform type identifier, and the target category identifier are arranged in a fixed order; The intermediate value is obtained by accumulating the numbered items one by one using a fixed base. Perform a fixed-modulus modulo operation on the intermediate value, and use the result as the compact identifier.

[0007] Furthermore, the calculation of the uniform bidding intensity level includes: The first product is obtained by multiplying the budget interval number by the first weight in the weight value; The second product is obtained by multiplying the frequency interval number by the second weight in the weight value; The base score is obtained by adding the first product to the second product; The perturbation value is obtained by performing a modulo operation on the compact identifier, and the uniform bidding intensity level is obtained by adding the base score and the perturbation value.

[0008] Furthermore, determining the platform level number includes: Extract the threshold set corresponding to the platform type identifier from the platform strength mapping table in memory; The unified bidding intensity level is compared sequentially with the thresholds in the threshold set; The level number corresponding to the threshold range into which the unified bidding intensity level falls is determined as the platform level number.

[0009] Furthermore, the calculation of the actual bid parameters includes: Read the template coefficients corresponding to the platform level number from the platform parameter template library; The scaling factor is obtained by calculating the ratio of the unified bidding intensity level to the maximum intensity constant configured in the system. The gain factor is obtained by adding the template coefficients to the scaling factor. The actual bid parameter is obtained by multiplying the base bid of the advertising plan by the gain factor.

[0010] Furthermore, the construction platform bidding request object includes: Call the message construction class that matches the platform type identifier to generate a fixed-field structure; Write the actual bid parameters into the bid value field of the fixed field structure; Write the compact identifier into the request tracking field of the fixed field structure; The fixed field structure of the written data is serialized to obtain the platform bidding request object.

[0011] Furthermore, the method also includes: Receive the bidding response returned by the platform and parse it to obtain the execution result; The execution result is associated with and stored in the compact identifier in the bidding context object; Perform ad display, billing, or log write operations based on the data stored in the associated database.

[0012] Furthermore, the decision benchmark instance is an immutable structure. The immutable structure is written into the decision state space, the freeze time identifier, and the compact identifier once during creation, and is passed in read-only reference mode in subsequent steps.

[0013] In a second aspect of the present invention, a cross-platform advertising intelligent decision-making and dynamic bidding system is provided, the system comprising: The state constraint module is used to receive real-time bidding requests from the advertising platform, read the cumulative spending amount, total budget, exposure count value within the time window, platform type identifier, and target category identifier of the advertising plan; map the ratio of the cumulative spending amount to the total budget to a budget interval number, map the exposure count value to a frequency interval number, and construct a decision state space containing the budget interval number, the frequency interval number, the platform type identifier, and the target category identifier; The benchmark freeze module is used to write the decision state space into a read-only field of the bidding context object, generate a freeze time identifier, and calculate a compact identifier based on the decision state space to form a decision benchmark instance. The decision expression module is used to query the target weight configuration table to obtain the weight value corresponding to the target category identifier, calculate the unified bidding intensity level based on the budget interval number, the frequency interval number, the weight value and the compact identifier, and determine the platform level number in combination with the platform type identifier, and generate an advertising bidding decision expression containing the compact identifier, the unified bidding intensity level and the platform level number. The dynamic bidding module is used to select a platform parameter template based on the platform level number, calculate the actual bidding parameters based on the unified bidding intensity level, the basic bid of the advertising plan and the template coefficient, and construct a platform bidding request object containing the actual bidding parameters and the compact identifier concurrently.

[0014] The beneficial technical effects of the present invention are at least as follows: To address the aforementioned issues, this invention provides a cross-platform advertising intelligent decision-making and dynamic bidding method and system. By discretizing necessary state elements such as budget constraints, frequency constraints, platform type, and target category when bidding is triggered, a unified decision state space is formed. This space is then frozen within the bidding context as the decision benchmark for the current bidding. A compact identifier is generated to bind this benchmark in multi-threaded and asynchronous call chains. Subsequently, a decision expression including intensity level and platform level is generated based on the frozen benchmark. This allows the execution side to complete parameter assembly and interface calls through a platform template library without rereading the running state, carrying the benchmark identifier in the call context to maintain consistency from decision to execution. Through this concept, this invention enables cross-platform advertising dynamic bidding to maintain consistent internal state references, transferable decision semantics, and constrained execution assembly even in concurrent and heterogeneous environments, thereby reducing the risk of semantic offset in cross-platform execution and improving engineering controllability and stability. Attached Figure Description

[0015] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0016] Figure 1 This is a flowchart of a cross-platform advertising intelligent decision-making and dynamic bidding method according to the present invention.

[0017] Figure 2 This is a framework diagram of a cross-platform advertising intelligent decision-making and dynamic bidding system according to the present invention. Detailed Implementation

[0018] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0019] In one or more embodiments, such as Figure 1 As shown, a cross-platform advertising intelligent decision-making and dynamic bidding method is disclosed, the method comprising the following: S1: Receive real-time bidding requests from the advertising platform, read the cumulative spending amount, total budget, exposure count within the time window, platform type identifier, and target category identifier of the advertising plan; map the ratio of the cumulative spending amount to the total budget to a budget interval number, map the exposure count to a frequency interval number, and construct a decision state space containing the budget interval number, the frequency interval number, the platform type identifier, and the target category identifier; Specifically, upon receiving a real-time bidding request from an advertising platform, the system first reads the real-time running status of the current advertising campaign through the campaign execution control module. This module continuously maintains the cumulative spending and total budget for each advertising campaign during system operation. These two values ​​are stored in the running status structure fields in memory and are updated instantly each time a billing deduction occurs. Let the cumulative spending be... The total budget is Then the proportion of budget usage Calculated as follows: ; in, The current cumulative spending amount for the advertising campaign is maintained and cached in real time by the billing and deduction logic within the campaign operation control module. The total budget field, configured and persisted in the database during ad campaign creation, is loaded into memory via a database read interface when bidding is triggered. The aforementioned ratio... This is a numerical value in proportional form, used to depict the progress of budget consumption. The system then divides the data according to pre-loaded interval partitioning rules into... Mapped to discrete budget interval number For example, when A value of 1 is assigned when the value falls within the predefined first threshold interval, a value of 2 is assigned when the value falls within the second threshold interval, and so on. The interval division rules are loaded by the configuration file during system initialization and stored in a parameter table that resides in memory. They are directly called during runtime and are not dynamically modified.

[0020] After the budget ratio is processed, the system reads the exposure count value within the current time window from the frequency control submodule. This count value originates from a time-window-based counter structure. Each time an exposure record is written to the log system, the frequency control module increments the count unit within the corresponding time window. The time window is divided by an internal system timer for a fixed duration, such as a several-minute interval, and is identified by an array index. The read... For integer count values, the system maps them to discrete interval numbers according to preset frequency interval rules. For example, 0 exposures are mapped to 0, a certain number of exposures are mapped to 1, and a higher number of exposures are mapped to 2. This mapping rule is also loaded into memory when the system starts up to ensure the consistency of frequency representation across different platforms.

[0021] Meanwhile, the bidding interface listening module extracts the platform type identifier when parsing the bidding request message. This identifier comes from a number mapping table agreed upon when integrating with various advertising platforms. For example, during the system integration phase, fixed numbers are assigned to different platforms, and this mapping relationship is stored in a local configuration file. Whenever a bidding request is received, the corresponding number is obtained by parsing the platform field in the message and querying the mapping table. In addition, the system retrieves the target category identifier of the advertising campaign through database queries. This field, stored in the main ad campaign table, identifies the target type for the current ad campaign. Database queries are performed using the primary key index and are cached in the current bidding context object after being read.

[0022] After completing the above processing, the system constructs a cross-platform advertising bidding decision state space: ; in, The budget interval number is derived from the ratio value. The interval mapping result; The frequency interval number is derived from the exposure count value. The interval mapping result; The platform type number is obtained by parsing the bidding interface message; This is the target category number for the advertising campaign, retrieved from a database field. This collection... It is stored in memory as a fixed-length structure and associated with the current bid request object, serving as the sole decision input in subsequent steps. This is achieved by converting continuous runtime data into a discrete, numbered representation of the state space. While maintaining operational constraint information, it achieves structural uniformity, providing a stable and standardized input basis for subsequent generation of advertising bidding decision benchmarks.

[0023] S2: Write the decision state space into the read-only field of the bidding context object, generate a freeze time identifier, and calculate a compact identifier based on the decision state space to form a decision benchmark instance; Specifically, after obtaining the cross-platform advertising bidding decision state space output in step one... This step then solidifies it into a stable, referable decision benchmark instance within the current ad bidding process. This solidification action addresses the engineering realities of cross-platform dynamic bidding: the same ad campaign may be triggered multiple times under concurrent traffic, with budget progress and frequency counts continuously updated in different threads; if subsequent decision expression generation and cross-platform parameter generation read state combinations from different points in time, inconsistencies in the state within a single bidding chain will occur. This step, based on the principle of "forming a one-time decision benchmark within the bidding trigger thread," will... Write to a read-only field of the bidding context object and synchronously record the freeze time marker. This allows the "state combination on which the current bidding is based" and the "time when the state combination is formed" to be passed together within the same context.

[0024] Regarding the calculation of identifier generation, this step involves accumulating a set of ordered discrete values ​​item by item according to a fixed cardinality to ensure order sensitivity and combinatorial sensitivity; subsequently, modular arithmetic is used to compress the result into a fixed integer space, facilitating constant-time carrying and comparison in high-concurrency links. Due to the... All are discrete numbers; treating them as "symbol sequences" and encoding them using multinomial hashing is a direct migration from classical algorithms to this scenario. Considering the characteristics of cross-platform ad bidding, this step chooses to use the "platform type number"... "and target category number" "Incorporating them into the same ordered coding ensures that the same budget range and frequency range form different frozen baseline identifiers on different platforms or with different target types, thereby binding subsequent parameter template selection and constraint caliber to the same identifier. The specific calculation is as follows: first, sequentially..." Perform radix accumulation, then modulus compression to obtain : ; in, The budget interval number is derived from the result of step one, which uses a proportional interval to divide the budget. The frequency interval number is derived from the result of intervalizing the exposure count in step one; This is the platform type number, derived from the parsing and mapping table lookup of the request message fields by the bidding interface listening module; The ad campaign target category number is read from the ad campaign configuration database field. (Constant) A fixed base is used to implement order-sensitive term-by-term accumulation; a constant. A fixed modulus is used to compress the encoding to a fixed range and reduce the probability of collisions. The entire calculation process involves only integer addition, multiplication, and modulo operations, with both sides representing integer values. The calculation result... It is also an integer identifier, which facilitates storage and transmission as a fixed-length field in the bidding context object. (See the example in step one.) For example, the calculation process can be written as a continuous, debugging-style calculation: first obtain intermediate values. , and then get , and then get Finally, the modulus is taken to obtain ;Should In this bidding process, the compact identifier used as the freezing benchmark is passed along with the context, and is used to bind the subsequently generated decision expressions to the same set of platform-side parameter generation actions. .

[0025] After completion After calculation, the system forms a decision benchmark instance in the bidding context object. Its assembly method uses fixed-length tuple encapsulation, allowing subsequent modules to obtain the state combination, freeze time identifier, and compact identifier used in the current bidding without rereading the running state: ; in, The output of step one is written to the state space of the bidding context object; The freeze time marker is generated by the system's monotonically increasing timer source within the auction trigger thread and written to the auction context object to mark the order in which the freeze benchmark was generated; This is the integer compact identifier calculated by the above formula. In engineering implementation, this corresponds to an immutable structure: three fields are written at once during creation, and then passed to the next step as a read-only reference; the next step will... The time-bound field written to the decision expression is used to limit the validity of the expression within the lifecycle of the current bidding request; the next step reads this field directly when generating the ad bidding decision expression. In As a decision-making condition, and to take Write the binding fields into the decision expression to ensure a one-to-one correspondence between the expression and the current freeze baseline; the next step is to carry the same data when generating parameters across platforms and making API calls. Write to the platform call context to ensure that the link from "state condition" to "action" revolves around the same freeze baseline. The output of this step is the decision baseline instance for the current ad bidding. ,in From step one, Writes originating from the timer source within the auction-triggered thread. From the Polynomial rolling hash coding and modular compression calculation.

[0026] S3: Query the target weight configuration table to obtain the weight value corresponding to the target category identifier, calculate the unified bidding intensity level based on the budget interval number, the frequency interval number, the weight value and the compact identifier, and determine the platform level number in combination with the platform type identifier, and generate an advertising bidding decision expression containing the compact identifier, the unified bidding intensity level and the platform level number; Specifically, in obtaining the decision benchmark instance output from step two... This step then proceeds based on the frozen state space. Generate advertising bidding decision expression and combine this expression with compact identifiers Binding ensures consistent referencing throughout subsequent cross-platform parameter generation and API call chains. The expression's calculation employs a combination of discrete scoring and deterministic perturbation: the discrete scoring part is based on a linear weighted summation model in mathematics, superimposing several discrete features according to their weights to obtain a comparable score; the deterministic perturbation part is based on the classic hash bucketing and dithering concepts, using existing identifiers... A modulo operation is performed to obtain a small-range integer perturbation. This application combines and extrapolates these two methods to an intensity calculation for real-time bidding in cross-platform advertising: numbering the budget intervals... frequency interval numbering The two principal factors serving as discrete constraint strength are numbered according to the target category. Choose its weight, and then... This generates deterministic perturbations, enabling the same objective and similar constraints to be distributed across adjacent intensity levels under concurrent bidding, thereby preventing hotspot concentration in subsequent template selection and queue scheduling.

[0027] In practice, the system first stores the target weights in the "target weight configuration table" in memory according to... The query retrieves the integer weights. The key of this configuration table is the target category number. The values ​​are two integer weights, configured by the campaign management backend and written to the database, and loaded into the memory cache when the campaign service starts. They can be obtained by looking up the table during the bidding process. Then, a uniform bidding intensity level is calculated based on discrete weighted summation with a deterministic perturbation term. : ; in, The budget interval number comes from ; Frequency interval numbering, from ; and To be according to The retrieved integer weights; For compact identifiers in decision-making benchmark instances; For deterministic disturbance terms, first... Obtain by taking the modulus arrive The integer bucket numbers are then shifted to... arrive The perturbation value. Both sides of the above formula are discrete integer scores, and the operations only include integer multiplication, addition, subtraction, and modulo operations. The scale of the expression is consistent, which is convenient for implementation with integer instructions in high-concurrency bidding threads. This formula is derived from a linear weighted summation scoring model: Let the base score Adding deterministic disturbances get ,in The introduction of the same Combining different freeze indicators This creates a discrete distribution of adjacent gear positions, thereby providing a more granular and controllable distinction for subsequent template selection.

[0028] get Then, the system reads the platform type number from the current bidding request context object. (Should The bidding interface listening module writes the context object when parsing the bidding request message, and then... Select the corresponding threshold set in the "Platform Strength Mapping Table" in memory, and map the uniform strength level to the platform level number. The original source of this mapping table is a commonly used segmented threshold quantization method in engineering (dividing continuous or discrete scores into a finite number of levels according to a threshold). Each platform in the table is configured with a set of ordered thresholds, and the system compares them sequentially to obtain the values. This achieves a constant-time threshold scan or binary search. After mapping, a decision expression is formed: ; in, From Used for binding expressions and freezing benchmarks; For the reason The calculated uniform bidding intensity level; For the reason and The platform level number is obtained by quantification in the platform strength mapping table. The logical relationship between this formula and the previous formula is as follows: the previous formula first obtains the strength level. This formula will then be used to... and and quantified Encapsulated as a structure representation that can be directly referenced by the execution side, facilitating the next step. Select platform parameter template, to Select target caliber constraints, while carrying It runs through the entire execution chain.

[0029] A set of implementation examples calculated using proportional values ​​and configuration parameters are provided to demonstrate operability. The example state spaces from steps one and two are reused. And take the compact identifier obtained in the example of step two. When the target category number At that time, the system looks up the target weight configuration table. Substituting into the basic score, we get Then calculate the disturbance term. ,therefore Thus obtain Then number them according to platform type. Select the intensity mapping threshold set for platform 2. An example threshold selection rule is: when... Mapped to ,when Mapped to ,when Mapped to Then the current Mapping Ultimately, this leads to the formation of a decision expression. ,in Binding to the frozen baseline, This directly points to the "Level 3" parameter template for Platform 2 in the next step. This is used to select the corresponding set of strength level parameters within the template and assemble the execution constraints.

[0030] S4: Select a platform parameter template according to the platform level number, calculate the actual bid parameters based on the unified bidding intensity level, the basic bid of the advertising plan and the template coefficient, and construct a platform bidding request object containing the actual bid parameters and the compact identifier concurrently.

[0031] Specifically, in obtaining the advertising bidding decision expression output from step three... Then, this step uses this expression as the basis for execution to complete the cross-platform dynamic bidding for advertisements, where... This is used to determine the template set and interface adapter for the target platform. A key engineering challenge in cross-platform execution is that different platforms have different interface fields, bidding calibers, and sets of optional parameters; directly using a uniform numerical strategy can easily lead to semantic offsets on the platform side. This step will... As a selection key for platform template sets and interface adapters, As the index key for the platform's level template, As a strength scale inside the template, This frozen baseline binding identifier is written into the platform call context, thus stably mapping "uniform strength semantics" to "platform executable parameters." Platform-side parameter assembly is handled by the interface adaptation layer: during system deployment, the adaptation layer registers a set of message construction classes and field mapping tables for each platform. At runtime, it selects the corresponding construction class based on the target platform, writes the calculated bid parameters into the platform-defined fields, and... Write the extended field or request tracking field, then initiate a bidding request and receive the bidding results via the platform SDK or HTTP interface.

[0032] The quantitative calculation of bidding parameters uses a scaling model as the source. Mathematically, this model can be represented as a scaling factor over a benchmark value. Applying proportional gain yields ,in This is a dimensionless gain factor; in control and engineering systems, it is often used to linearly amplify or reduce a baseline control quantity by an intensity factor. Based on this, this application derives a modified form suitable for cross-platform bidding: using the base bid of the advertising plan. As a benchmark value, the platform level template coefficient With uniform strength level The ratio is used as a component of the gain and is introduced into the maximum intensity constant of the system configuration. right The gain term is proportionalized to keep it within a controllable range. This yields the actual bid parameters for the current auction. : ; in, The base bid value for the advertising campaign comes from the database field configured for the advertising campaign. It is written when the advertising campaign is created or modified, and loaded into the memory cache by querying the primary key when the bidding is triggered. The platform parameter template library has a level of The template coefficients are configured in the template library during the system deployment phase and loaded into an array of memory structures when the delivery service starts. The uniform bidding intensity level is calculated in step three; The maximum strength constant configured for the system is stored in a configuration file and loaded into the memory constant area when the service starts. (The left side of the equation...) With the right side All represent the bid values, with the value in parentheses being a dimensionless scaling factor. The coefficient is dimensionless. Since it is a dimensionless proportion, the expressions on both sides are consistent. The logical relationship between this formula and step three is as follows: step three first generates... This step is to... Select the template set and interface adapter for the corresponding platform, and... Extract the platform level template coefficients ,by Substitute the scaling factor and calculate This transforms the "uniform strength level" into a "platform-executable bid value".

[0033] Finish After calculation, the system constructs a platform bidding request object. And write it to the execution field. The message constructor is generated by the platform interface adaptation layer. This constructor is implemented as a fixed field structure and a serialization function: the platform that interfaces with the SDK uses the request object type provided by the SDK and assigns values ​​to the fields; the platform that interfaces with HTTP uses the JSON field mapping table to write the bid field to the specified key. It is written into the bidding fields specified by the platform (such as the bid value field or the maximum bid field). This information is written into the platform request tracking field or extended field so that the bidding results returned by the platform can be mapped to the freeze benchmark identifier within the system. Subsequently, the platform's sending interface is called to initiate the bidding request. The bidding response is received and parsed by the interface listening module into an execution result such as successful or unsuccessful bid, and the execution result is then compared with the request object. The data is saved in the same bidding context so that subsequent business processes can continue to handle display, billing, and log writing.

[0034] A set of examples calculated based on configuration parameters and intensity levels are provided to demonstrate operability. Let the base bid for a certain advertising plan be... The system is configured with a maximum strength constant. The decision expression output in step three is as follows: ,but , , The system is set to... Index the Level 3 template in the template array of the target platform and read the template coefficients. Substituting the scaling term, we get The item in parentheses is Therefore, the actual bid parameters The system then constructs a bid request object. Write the bid field And write the freeze flag field to The bidding request is initiated through the platform's API. After the platform returns a response, the API listening module parses the result and compares it with the response. By binding storage, a cross-platform dynamic advertising bidding process can be completed.

[0035] In one or more embodiments, such as Figure 2 As shown, a cross-platform advertising intelligent decision-making and dynamic bidding system is disclosed, the system comprising: The state constraint module is used to receive real-time bidding requests from the advertising platform, read the cumulative spending amount, total budget, exposure count value within the time window, platform type identifier, and target category identifier of the advertising plan; map the ratio of the cumulative spending amount to the total budget to a budget interval number, map the exposure count value to a frequency interval number, and construct a decision state space containing the budget interval number, the frequency interval number, the platform type identifier, and the target category identifier; The benchmark freeze module is used to write the decision state space into a read-only field of the bidding context object, generate a freeze time identifier, and calculate a compact identifier based on the decision state space to form a decision benchmark instance. The decision expression module is used to query the target weight configuration table to obtain the weight value corresponding to the target category identifier, calculate the unified bidding intensity level based on the budget interval number, the frequency interval number, the weight value and the compact identifier, and determine the platform level number in combination with the platform type identifier, and generate an advertising bidding decision expression containing the compact identifier, the unified bidding intensity level and the platform level number. The dynamic bidding module is used to select a platform parameter template based on the platform level number, calculate the actual bidding parameters based on the unified bidding intensity level, the basic bid of the advertising plan and the template coefficient, and construct a platform bidding request object containing the actual bidding parameters and the compact identifier concurrently.

[0036] It is worth noting that the specific workflow of the cross-platform advertising intelligent decision-making and dynamic bidding system provided in this embodiment of the invention is the same as that of the cross-platform advertising intelligent decision-making and dynamic bidding method described in the above embodiment, and will not be repeated here.

[0037] This invention also provides a cross-platform advertising intelligent decision-making and dynamic bidding device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the steps described in the above embodiment of a cross-platform advertising intelligent decision-making and dynamic bidding method, for example... Figure 1 The steps S1 to S4 described above; or, when the processor executes the computer program, it implements the functions of each module in the above system embodiments.

[0038] For example, the computer program may be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the cross-platform advertising intelligent decision-making and dynamic bidding device.

[0039] The aforementioned cross-platform advertising intelligent decision-making and dynamic bidding device can be a computing device such as a desktop computer, laptop, handheld computer, or cloud server. This device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the device may also include input / output devices, network access devices, buses, etc.

[0040] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the cross-platform advertising intelligent decision-making and dynamic bidding device, connecting all parts of the device via various interfaces and lines.

[0041] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the cross-platform advertising intelligent decision-making and dynamic bidding device by running or executing the computer programs and / or modules stored in the memory and calling the data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function, etc.; the data storage area may store data created according to the operation of the controller, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0042] The module integrating cross-platform advertising intelligent decision-making and dynamic bidding equipment, if implemented as a software functional unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0043] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0044] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A cross-platform advertisement intelligent decision and dynamic bidding method, characterized in that, The method includes: Receive real-time bidding requests from advertising platforms and read the cumulative spending amount, total budget, exposure count within the time window, platform type identifier, and target category identifier of the advertising campaign; The ratio of the cumulative consumption amount to the total budget is mapped to a budget interval number, and the exposure count value is mapped to a frequency interval number, thus constructing a decision state space that includes the budget interval number, the frequency interval number, the platform type identifier, and the target category identifier; Write the decision state space into a read-only field of the bidding context object, generate a freeze time identifier, and calculate a compact identifier based on the decision state space to form a decision benchmark instance; Query the target weight configuration table to obtain the weight value corresponding to the target category identifier, calculate the unified bidding intensity level based on the budget interval number, the frequency interval number, the weight value and the compact identifier, and determine the platform level number by combining the platform type identifier, and generate an advertising bidding decision expression containing the compact identifier, the unified bidding intensity level and the platform level number; Select a platform parameter template based on the platform level number, calculate the actual bid parameters based on the unified bidding intensity level, the basic bid of the advertising plan and the template coefficient, and construct a platform bidding request object containing the actual bid parameters and the compact identifier concurrently.

2. The cross-platform advertisement intelligent decision and dynamic bidding method according to claim 1, wherein, The step of mapping the ratio of the cumulative consumption amount to the total budget to a budget interval number includes: The ratio of the cumulative consumption amount to the total budget is used to obtain the budget utilization rate. Based on the interval division rules pre-loaded into memory, determine the threshold interval into which the budget usage ratio falls; The preset number corresponding to the threshold interval that falls into the range is assigned to the budget interval number.

3. The cross-platform ad intelligent decision and dynamic bidding method of claim 1, wherein, The calculation of the compact identifier based on the decision state space includes: The budget interval number, the frequency interval number, the platform type identifier, and the target category identifier are arranged in a fixed order; The intermediate value is obtained by accumulating the numbered items one by one using a fixed base. Perform a fixed-modulus modulo operation on the intermediate value, and use the result as the compact identifier.

4. The cross-platform ad intelligent decision and dynamic bidding method of claim 1, wherein, The calculation of the unified bidding intensity level includes: The first product is obtained by multiplying the budget interval number by the first weight in the weight value; The second product is obtained by multiplying the frequency interval number by the second weight in the weight value; The base score is obtained by adding the first product to the second product; The perturbation value is obtained by performing a modulo operation on the compact identifier, and the uniform bidding intensity level is obtained by adding the base score and the perturbation value.

5. The cross-platform advertising intelligent decision-making and dynamic bidding method according to claim 1, characterized in that, The determination of the platform level number includes: Extract the threshold set corresponding to the platform type identifier from the platform strength mapping table in memory; The unified bidding intensity level is compared sequentially with the thresholds in the threshold set; The level number corresponding to the threshold range into which the unified bidding intensity level falls is determined as the platform level number.

6. The cross-platform advertising intelligent decision-making and dynamic bidding method according to claim 1, characterized in that, The calculation of the actual bid parameters includes: Read the template coefficients corresponding to the platform level number from the platform parameter template library; The scaling factor is obtained by calculating the ratio of the unified bidding intensity level to the maximum intensity constant configured in the system. The gain factor is obtained by adding the template coefficients to the scaling factor. The actual bid parameter is obtained by multiplying the base bid of the advertising plan by the gain factor.

7. The cross-platform advertising intelligent decision-making and dynamic bidding method according to claim 1, characterized in that, The construction platform bidding request object includes: Call the message construction class that matches the platform type identifier to generate a fixed-field structure; Write the actual bid parameters into the bid value field of the fixed field structure; Write the compact identifier into the request tracking field of the fixed field structure; The fixed field structure of the written data is serialized to obtain the platform bidding request object.

8. The cross-platform advertising intelligent decision-making and dynamic bidding method according to claim 1, characterized in that, The method further includes: Receive the bidding response returned by the platform and parse it to obtain the execution result; The execution result is associated with and stored in the compact identifier in the bidding context object; Perform ad display, billing, or log write operations based on the data stored in the associated database.

9. The cross-platform advertising intelligent decision-making and dynamic bidding method according to claim 1, characterized in that, The decision benchmark instance is an immutable structure. The immutable structure is written to the decision state space, the freeze time identifier, and the compact identifier once during creation, and is passed in read-only reference mode in subsequent steps.

10. A cross-platform advertising intelligent decision-making and dynamic bidding system, characterized in that, The system includes: The state constraint module is used to receive real-time bidding requests from the advertising platform, read the cumulative spending amount, total budget, exposure count value within the time window, platform type identifier, and target category identifier of the advertising plan; map the ratio of the cumulative spending amount to the total budget to a budget interval number, map the exposure count value to a frequency interval number, and construct a decision state space containing the budget interval number, the frequency interval number, the platform type identifier, and the target category identifier; The benchmark freeze module is used to write the decision state space into a read-only field of the bidding context object, generate a freeze time identifier, and calculate a compact identifier based on the decision state space to form a decision benchmark instance. The decision expression module is used to query the target weight configuration table to obtain the weight value corresponding to the target category identifier, calculate the unified bidding intensity level based on the budget interval number, the frequency interval number, the weight value and the compact identifier, and determine the platform level number in combination with the platform type identifier, and generate an advertising bidding decision expression containing the compact identifier, the unified bidding intensity level and the platform level number. The dynamic bidding module is used to select a platform parameter template based on the platform level number, calculate the actual bidding parameters based on the unified bidding intensity level, the basic bid of the advertising plan and the template coefficient, and construct a platform bidding request object containing the actual bidding parameters and the compact identifier concurrently.

Images