An advertising placement method and system based on a multi-objective optimization algorithm

Abstract

The present invention discloses an advertising placement method and system based on a multi-objective optimization algorithm, which relates to the technical field of advertising placement. It includes setting optimization objectives for each stage, allocating initial weights to the optimization objectives, dividing the advertising placement cycle according to the initial weights and performing resource allocation; generating and executing an initial advertising placement strategy according to the optimization objectives and the initial weights; collecting real-time performance data and resource consumption during the initial advertising placement process in real time, and analyzing the advertising placement effect of the previous stage; dynamically adjusting the initial weights of the optimization objectives for the next stage according to the analysis results; and optimizing the initial advertising placement strategy based on the adjusted weights of the optimization objectives. By setting optimization objectives for each stage and allocating initial weights, the present invention realizes the reasonable division of the advertising placement cycle and the precise allocation of resources, ensures that advertising objectives in different stages such as exposure volume, click-through rate, and conversion rate can be fully concerned, and avoids waste of resources.

Description

Technical Field

[0001] The present invention relates to the technical field of advertising placement, and particularly to an advertising placement method and system based on a multi-objective optimization algorithm. Background Art

[0002] With the rapid development of the Internet advertising market, the complexity and diversity of advertising placement have increased day by day. Traditional single-objective optimization methods have difficulty effectively meeting the advertising placement requirements of multiple scenarios and multiple objectives. Advertisers not only hope to maximize the exposure of advertisements in each advertising placement stage, but also hope to adjust advertising strategies according to different time periods, market environments, and user behaviors to maximize click-through rates and conversion rates. Therefore, multi-objective optimization algorithms have gradually become a research hotspot in the field of advertising placement. In recent years, advertising placement methods based on multi-objective optimization have been gradually introduced. By setting multiple optimization objectives and comprehensively considering the performance of advertisements in different stages, efforts are made to achieve better optimization effects simultaneously in multiple dimensions such as exposure, click, and conversion. Such methods not only improve the flexibility of advertising placement, but also provide more refined placement strategies for advertisers, thus receiving wide attention.

[0003] However, the existing advertising placement technologies based on multi-objective optimization algorithms still have some deficiencies in practical applications. First of all, these technologies often lack flexibility in resource allocation and fail to fully consider the dynamic demand changes in each stage of the advertising placement cycle. Specifically, existing technologies usually adopt a method of evenly distributing resources when dividing the placement cycle. This strategy ignores the differences in market environment, user behavior, and stage objectives, resulting in problems such as resource waste or improper allocation. Secondly, there is a lag in the weight adjustment of existing advertising placement strategies. Usually, a fixed weight allocation is executed within a fixed cycle, and the strategy for subsequent stages cannot be dynamically adjusted according to the real-time advertising placement effect, making it difficult to fully respond to the rapid changes in the market environment and the immediate feedback of user behavior. These deficiencies limit the flexibility and intelligence level of advertising placement strategies to a certain extent. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides an advertising placement method based on a multi-objective optimization algorithm to solve the problems of lack of flexibility in resource allocation and lag in weight adjustment in the prior art.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides an advertising placement method based on a multi-objective optimization algorithm, which includes setting optimization objectives for each stage, assigning preliminary weights to the optimization objectives, dividing the advertising placement cycle according to the preliminary weights and allocating resources; generating and executing a preliminary advertising placement strategy according to the optimization objectives and the preliminary weights; collecting real-time performance data and resource consumption during the preliminary advertising placement process in real time, and analyzing the advertising placement effect of the previous stage; dynamically adjusting the preliminary weights of the optimization objectives in the next stage according to the analysis results; and optimizing the preliminary advertising placement strategy based on the adjusted weights of the optimization objectives.

[0008] As a preferred embodiment of the advertising placement method based on the multi-objective optimization algorithm of the present invention, wherein: the steps of setting optimization objectives for each stage and assigning preliminary weights to the optimization objectives are as follows:

[0009] Determine the total advertising placement cycle and the total amount of advertising resources according to business requirements and the market promotion plan;

[0010] Divide the placement cycle into a preheating stage, a main promotion stage, and a sprint stage according to non-uniform time according to the total advertising placement cycle;

[0011] In the preheating stage, define maximizing the exposure volume as the optimization objective;

[0012] In the main promotion stage, define maximizing the click-through rate as the optimization objective;

[0013] In the sprint stage, define maximizing the conversion rate as the optimization objective;

[0014] Assign preliminary weights to the optimization objectives by performing exponential, sine processing, and difference adjustment on the information entropy of each objective. The expression is:

[0015]

[0016] where w j is the preliminary weight of the jth optimization objective, E j is the information entropy value of the jth optimization objective, n is the total number of optimization objectives, j is the index variable of the optimization objective, λ is the adjustment coefficient for adjusting the difference in the weight distribution of the information entropy values of the optimization objectives, γ is the adjustment coefficient for controlling the weakening strength of the weights of high-entropy value objectives, and E max is the maximum value among the information entropies of all optimization objectives.

[0017] As a preferred embodiment of the advertising placement method based on the multi-objective optimization algorithm of the present invention, wherein: the steps of dividing the advertising placement cycle according to the preliminary weights and allocating resources are as follows:

[0018] Advertising resources for each stage with non-uniform time division are non-uniformly allocated according to the market demand and user behavior changes in each stage. The expression is as follows:

[0019]

[0020] Among them, R i is the advertising resource volume in the i-th stage, R is the total advertising resource volume, D i is the market activity in the i-th stage, U i is the user behavior characteristics in the i-th stage, and i is the index variable of the advertising placement stage.

[0021] As a preferred solution of the advertising placement method based on the multi-objective optimization algorithm described in the present invention, wherein: the preliminary advertising placement strategy includes advertising channel selection, advertising material design, advertising targeting setting, as well as budget and frequency.

[0022] As a preferred solution of the advertising placement method based on the multi-objective optimization algorithm described in the present invention, wherein: the real-time performance data during the advertising placement process includes the exposure volume, click volume, conversion volume, click-through rate, and conversion rate;

[0023] The resource consumption situation includes the advertising budget consumption, click cost, display cost, and conversion cost within each time period;

[0024] To analyze the advertising placement effect in the pre-analysis stage, the specific steps are as follows.

[0025] By using the mutual relationship among the exposure volume, click volume, conversion volume, and budget consumption, logarithmic transformation is used to weaken the influence of the maximum values of the exposure volume and resource consumption volume. At the same time, the advertising placement effect is measured by the click-through rate and conversion rate. The expression is as follows:

[0026]

[0027] Among them, I i (t) is the exposure volume at time point t in the i-th stage, C i (t) is the click volume at time point t in the i-th stage, V i (t) is the conversion volume at time point t in the i-th stage, L i (t) is the resource consumption volume at time point t in the i-th stage, ∈ is a small positive number to prevent the denominator from being zero, EI i (t) is the stage advertising placement effect index at time point t in the i-th stage, R i (t) represents the advertising resource input volume at time point t in the i-th stage.

[0028] As a preferred solution of the advertising placement method based on the multi-objective optimization algorithm of the present invention, wherein: the preliminary weights of the optimization objectives in the next stage are dynamically adjusted according to the analysis results, and the specific steps are as follows.

[0029] By using the effect index of the current stage and the effect thresholds of each optimization objective, the hyperbolic tangent function is used to smooth the difference between the effect index and the threshold, and the preliminary weights of the optimization objectives are dynamically adjusted. The expression is:

[0030]

[0031] where w′ j (i + 1) is the weight of the jth optimization objective after dynamic adjustment in the i + 1 stage, μ is the adjustment coefficient that controls the adjustment range of the preliminary weights of the optimization objectives, and θ j is the effect threshold corresponding to the jth optimization objective, and w j (t) represents the preliminary weight of the jth optimization objective at time point t.

[0032] As a preferred solution of the advertising placement method based on the multi-objective optimization algorithm of the present invention, wherein: based on the weights of the adjusted optimization objectives, the preliminary advertising placement strategy is optimized, and the specific steps are as follows.

[0033] Based on the weight w′ j (t + 1) of the adjusted optimization objective, the advertising resource amounts in each stage are reallocated;

[0034] According to the advertising resource amount R' i of the adjusted ith stage, the preliminary advertising placement strategy is optimized.

[0035] In a second aspect, the present invention provides an advertising placement system based on a multi-objective optimization algorithm, including a resource allocation module, a strategy generation module, an effect analysis module, a weight optimization module, and a strategy optimization module; the resource allocation module is used to set the optimization objectives for each stage, allocate preliminary weights to the optimization objectives, divide the advertising placement cycle according to the preliminary weights, and perform resource allocation; the strategy generation module is used to generate and execute a preliminary advertising placement strategy according to the optimization objectives and the preliminary weights; the effect analysis module is used to collect the real-time performance data and resource consumption situation in the preliminary advertising placement process in real time, and analyze the advertising placement effect of the previous stage; the weight optimization module is used to dynamically adjust the preliminary weights of the optimization objectives in the next stage according to the analysis results; the strategy optimization module is used to optimize the preliminary advertising placement strategy based on the weights of the adjusted optimization objectives.

[0036] In a third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the advertising placement method based on the multi-objective optimization algorithm as described in the first aspect of the present invention is implemented.

[0037] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the advertising placement method based on the multi-objective optimization algorithm as described in the first aspect of the present invention is implemented.

[0038] The beneficial effects of the present invention are as follows: By setting the optimization objectives for each stage and allocating preliminary weights, a reasonable division of the advertising placement cycle and precise allocation of resources are achieved, ensuring that advertising objectives in different stages, such as exposure, click-through rate, and conversion rate, can receive sufficient attention and avoiding waste of resources; during the advertising placement process, advertising performance data and resource consumption are collected in real time, the effects of the previous stage are analyzed, and the optimization objective weights for the next stage are dynamically adjusted based on the analysis results, enabling the advertising strategy to flexibly respond to changes in the market and user behavior, ultimately improving the overall effect of advertising placement and the resource utilization efficiency. Description of the Drawings

[0039] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for description in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts.

[0040] Figure 1 It is a flowchart of the advertising placement method based on the multi-objective optimization algorithm in Embodiment 1.

[0041] Figure 2 It is a schematic diagram of the advertising placement system based on the multi-objective optimization algorithm in Embodiment 1. Detailed Embodiments

[0042] To make the above objects, features, and advantages of the present invention more obvious and understandable, the detailed embodiments of the present invention will be described in detail below with reference to the drawings in the specification.

[0043] Many specific details are set forth in the following description to facilitate a thorough understanding of the present invention, but the present invention may be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention, so the present invention is not limited by the specific embodiments disclosed below.

[0044] Second, the "one embodiment" or "embodiment" referred to herein means a specific feature, structure, or characteristic that may be included in at least one implementation manner of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or alternative embodiments that are mutually exclusive with other embodiments.

[0045] Embodiment 1, referring to Figure 1 and Figure 2 , is the first embodiment of the present invention. This embodiment provides an advertising placement method based on a multi-objective optimization algorithm, including the following steps:

[0046] S1: Set the optimization objectives for each stage, assign preliminary weights to the optimization objectives, divide the advertising placement cycle according to the preliminary weights, and allocate resources.

[0047] S1.1: Determine the total cycle of advertising placement and the total amount of advertising resources according to business requirements and marketing plans;

[0048] For example, determine a 30-day promotion activity with a total advertising budget of 1 million yuan. Then, divide the placement cycle into a preheating stage (the first 5 days), a main promotion stage (the next 15 days), and a sprint stage (the last 10 days), and reasonably allocate advertising resources according to the objectives of each stage: For example, in the preheating stage, allocate 20% of the budget to maximize exposure; in the main promotion stage, allocate 50% of the budget to increase the click-through rate; in the sprint stage, allocate the remaining 30% of the budget to improve the conversion rate.

[0049] S1.2: Divide the placement cycle into a preheating stage, a main promotion stage, and a sprint stage according to non-uniform time according to the total cycle of advertising placement;

[0050] In the preheating stage, define maximizing exposure as the optimization objective;

[0051] In the main promotion stage, define maximizing the click-through rate as the optimization objective;

[0052] In the sprint stage, define maximizing the conversion rate as the optimization objective;

[0053] Specifically: Example of stage division:

[0054] It can also be divided according to marketing activities. For example:

[0055] The first stage: The first 5 days are the preheating stage.

[0056] The second stage: The next 15 days are the main promotion stage.

[0057] The third stage: The last 10 days are the sprint stage.

[0058] S1.3: By performing exponential, sine processing, and differential adjustment on the information entropy of each target, assign preliminary weights to the optimization targets. The expression is as follows:

[0059]

[0060] where w j is the preliminary weight of the j-th optimization target, E j is the information entropy value of the j-th optimization target, n is the total number of optimization targets, j is the index variable of the optimization target, λ is the adjustment coefficient for regulating the difference in weight distribution of the information entropy values of the optimization targets, γ is the adjustment coefficient for controlling the degree of weakening of the weights of high-entropy targets, and E max is the maximum value among the information entropies of all optimization targets.

[0061] Furthermore, the information entropy value of the optimization target is expressed as:

[0062]

[0063] where m is the total number of advertising placement stages, p ij is the proportion of the optimization target j in the i-th stage, and i is the index variable of the advertising placement stage.

[0064] It should also be noted that this process performs complex mathematical processing on each optimization target, adjusts it in combination with the differences between the targets, and reasonably assigns the preliminary weights of each optimization target. This way of weight assignment ensures that during the advertising placement process, resources can be allocated according to the priorities of different targets, balancing the requirements of each stage, and ultimately optimizing the overall advertising placement effect.

[0065] S1.4: Allocate the advertising resources in each stage divided according to non-uniform time non-uniformly according to the market demand and user behavior changes in each stage. The expression is as follows:

[0066]

[0067] where R i is the amount of advertising resources in the i-th stage, R is the total amount of advertising resources, D i is the market activity in the i-th stage, U i is the user behavior characteristics in the i-th stage, and i is the index variable of the advertising placement stage.

[0068] Furthermore, user behavior characteristics are obtained by the advertising platform collecting real-time interaction data between users and advertisements and analyzing and processing it. For example, the platform will record whether the user clicks on the advertisement, the time spent on the advertisement page, and whether further purchase or registration operations are performed. Taking one advertisement placement as an example, assume that after seeing the advertisement, the user clicks on the link, stays on the target page for 30 seconds, and then completes the purchase. The platform will extract and analyze the user's click behavior, stay duration, and conversion behavior as user behavior characteristics, calculate the user's responsiveness and engagement with the advertisement, and thus provide a basis for optimizing the advertisement at this stage.

[0069] S2: Generate and execute a preliminary advertisement placement strategy based on the optimization goal and preliminary weights.

[0070] S2.1: The preliminary advertisement placement strategy includes advertisement channel selection, advertisement material design, advertisement targeting settings, as well as budget and frequency.

[0071] Furthermore, based on the optimization goal and resource volume of each stage, select the advertisement placement channel;

[0072] For example, in the preheating stage, select display advertisements or video advertisements with high exposure, in the main promotion stage, select search advertisements or social media advertisements that improve click-through rate, and in the sprint stage, select remarketing advertisements or promotional email advertisements with high conversion rate.

[0073] Design corresponding advertisement materials according to the optimization goal of each stage;

[0074] For example, in the preheating stage, the advertisement materials should focus on brand exposure and use pictures or videos with strong visual impact; in the main promotion stage, the materials should highlight the product advantages and promotional information to encourage users to click; in the sprint stage, the advertisement materials should emphasize the sense of urgency and preferential intensity to prompt users to complete the purchase as soon as possible.

[0075] Advertisement targeting settings and budget frequency control should also be adjusted according to the goals of each stage. In the preheating stage, the advertisement targeting can be broader to cover more potential users; in the main promotion stage, it should focus on user groups with higher click potential; in the sprint stage, the targeting should be more precise to lock in users with purchase intent. The budget and frequency are reasonably controlled according to the resource allocation of each stage to ensure the maximization of advertisement effects.

[0076] S3: Real-time collect the real-time performance data and resource consumption during the preliminary advertisement placement process, and analyze the advertisement placement effect of the previous stage.

[0077] S3.1: The real-time performance data during the advertisement placement process includes the exposure volume, click volume, conversion volume, click-through rate, and conversion rate;

[0078] Furthermore, the real-time performance data during the advertisement placement process is automatically collected through the monitoring system of the advertisement platform. The platform records the interaction behaviors between users and advertisements, such as the number of advertisement displays (exposure volume), user click behaviors (click volume), and completed conversion actions (conversion volume), and calculates the click-through rate by computing the ratio of the click volume to the exposure volume, and the conversion rate by computing the ratio of the conversion volume to the click volume.

[0079] S3.2: The resource consumption situation includes the advertisement budget consumption, click cost, display cost, and conversion cost within each time period;

[0080] Furthermore, the real-time resource consumption situation during the advertisement placement process is obtained through the billing system of the advertisement platform. The platform automatically calculates and records the corresponding advertisement budget consumption, cost per click (click cost), cost per display (display cost), and cost per conversion (conversion cost) based on the advertisement display, click, and conversion events in each time period.

[0081] S3.3: Analyze the advertisement placement effect in the previous stage. The specific steps are as follows.

[0082] By using the mutual relationship among the exposure volume, click volume, conversion volume, and budget consumption, the logarithmic transformation is used to weaken the influence of the maximum values of the exposure volume and resource consumption volume. At the same time, the advertisement placement effect is measured through the click-through rate and conversion rate. The expression is as follows:

[0083]

[0084] Among them, I i (t) is the exposure volume at time point t in the i-th stage, C i (t) is the click volume at time point t in the i-th stage, V i (t) is the conversion volume at time point t in the i-th stage, L i (t) is the resource consumption volume at time point t in the i-th stage, ∈ is a tiny positive number to prevent the denominator from being zero, EI i (t) is the stage advertisement placement effect index at time point t in the i-th stage, R i (t) represents the advertisement resource input volume at time point t in the i-th stage.

[0085] Furthermore, EI i (t) ∈ (0, +∞). The higher the index, the better the effect. Usually, the value ranges from 0 to 10. The specific value range interpretation is as follows:

[0086] EI i (t) < 1: The stage effect is poor, and the weight of the optimization target needs to be significantly adjusted.

[0087] 1 ≤ EI i(t) < 5: The stage effect is medium, and the weights of the optimization objectives should be adjusted appropriately.

[0088] EI i (t) ≥ 5: The stage effect is good, and the weights of the optimization objectives can be maintained or slightly adjusted.

[0089] It should be noted that this expression comprehensively processes the exposure volume, click volume, conversion volume, and budget consumption, and uses logarithmic transformation to weaken the influence of extreme values in the exposure volume and resource consumption, ensuring that it is not interfered by the outliers of a single indicator during the calculation process. At the same time, the actual effect of the advertisement is further measured by the ratio of the click-through rate and the conversion rate. The algorithm performs refined calculations on the mutual relationships among the exposure volume, click volume, conversion volume, and resource consumption to obtain the advertising placement effect index EI i (t) to more accurately evaluate the advertising performance of each stage. It can dynamically and objectively reflect the effect of advertising placement, avoid the distortion of the overall effect by extreme data, and thus provide a reliable basis for subsequent strategy optimization.

[0090] S4: Dynamically adjust the preliminary weights of the optimization objectives for the next stage according to the analysis results.

[0091] S4.1: Through the effect index of the current stage and the effect thresholds of each optimization objective, use the hyperbolic tangent function to smooth the difference between the effect index and the threshold, and dynamically adjust the preliminary weights of the optimization objectives. The expression is:

[0092]

[0093] where, w′ j (i + 1) is the weight of the jth optimization objective dynamically adjusted in the i + 1 stage, μ is the adjustment coefficient that controls the adjustment range of the preliminary weights of the optimization objectives, θ j is the effect threshold corresponding to the jth optimization objective, and w j (t) represents the preliminary weight of the jth optimization objective at time point t.

[0094] Furthermore, θ j is defined based on historical data and business objectives. The definition process first analyzes the past advertising placement data, calculates the average effect level and fluctuation range of each optimization objective (such as exposure volume, click-through rate, conversion rate, etc.). Combining specific business requirements and market conditions, a reasonable target benchmark is set. For example, by analyzing past advertising placements, the historical average click-through rate is determined to be 2%, and it is set as the threshold θ j .

[0095] It should be noted that this expression combines the effect index EI i (t) of the current stage with the effect thresholds θ of each optimization objectivej The difference between them is smoothed using the hyperbolic tangent function tanh to avoid drastic fluctuations during weight adjustment. The initial weight w j (t) is dynamically adjusted according to this smoothed difference, and the adjustment coefficient μ controls the adjustment amplitude to ensure the flexibility and stability of weight adjustment. Finally, the weights of all objectives are reallocated through normalization (the denominator part) so that the weights of each optimization objective can dynamically adapt according to the real-time effect during the advertisement placement process, optimizing the allocation of advertisement resources. The beneficial effect of this algorithm is that through smoothing, it avoids overresponding to short-term fluctuations and at the same time ensures that the advertisement placement strategy can be flexibly adjusted according to the actual effect.

[0096] S5: Optimize the preliminary advertisement placement strategy based on the weights of the adjusted optimization objectives.

[0097] S5.1: Based on the weights w′ j (t + 1) of the adjusted optimization objectives, reallocate the amount of advertisement resources in each stage;

[0098] Specifically: First, calculate the priority of each optimization objective in advertisement placement according to the current weight adjustment results in each stage. Then, reallocate the total advertisement resources according to the new weight ratio to each stage to ensure that the resource allocation more precisely matches the current optimization objectives.

[0099] For example, through the adjusted weights w′ j (t + 1) of the optimization objectives, the specific process of reallocating the amount of advertisement resources in each stage is as follows:

[0100] Suppose there are a preheating stage, a main promotion stage, and a sprint stage, and the total advertisement budget is 1 million yuan. In the initial resource allocation, the weights of the three stages are w1(t)=0.3, w2(t)=0.4, w3(t)=0.3 respectively. Therefore, the initially allocated resource amounts are:

[0101] Preheating stage: 300,000 yuan

[0102] Main promotion stage: 400,000 yuan

[0103] Sprint stage: 300,000 yuan

[0104] After advertisement placement, based on the performance of each stage, the weights of the optimization objectives are recalculated, and the adjusted weights are w'1(t + 1)=0.2, w'2(t + 1)=0.5, w'3(t + 1)=0.3.

[0105] When reallocating resources, according to the new weight ratio, the advertisement resources will be reallocated as:

[0106] Preheating stage: 1 million yuan × 0.2 = 200,000 yuan

[0107] Main promotion stage: 1 million yuan × 0.5 = 500,000 yuan

[0108] Sprint stage: 1 million yuan × 0.3 = 300,000 yuan

[0109] Therefore, the resource allocation has changed. In the main promotion stage, due to the increase in its weight, it gets more advertising resources (from 400,000 yuan to 500,000 yuan), while in the preheating stage, due to the decrease in its importance, the resource allocation is reduced (from 300,000 yuan to 200,000 yuan). In this way, the advertising resource allocation can dynamically adapt to the actual needs and target weights of each stage, ensuring the maximization of the advertising placement effect.

[0110] S5.2: Optimize the preliminary advertising placement strategy according to the adjusted advertising resource volume of the i-th stage.

[0111] Furthermore, first, use the reallocated advertising resource volume to re-examine and adjust the advertising placement strategy of this stage to ensure the effective utilization of resources.

[0112] Specifically, the optimization strategies include: advertising channel adjustment, according to the change of resources, select more suitable advertising channels for the current budget, such as choosing high-conversion channels when increasing the budget, and giving priority to low-cost exposure channels when reducing the budget;

[0113] Advertising material optimization, ensure that the design of advertising materials matches the goals of the current stage, increasing user clicks and conversion rates;

[0114] For example, in the main promotion stage, the goal is to increase the click-through rate, and the advertising resource volume increases to 500,000 yuan. To attract more users to click, the advertising materials can highlight the unique selling points of the product and time-limited discount information. If promoting a smartwatch, the advertising materials can be designed to emphasize its core functions (such as "real-time heart rate monitoring" and "all-weather waterproof"), and add promotional information like "50% off limited time, only today" in the copywriting. Such a design can attract users to click and improve the click-through rate.

[0115] Optimization of targeting settings, by analyzing user behavior data, adjust the accuracy of the target audience, and lock in higher-value user groups;

[0116] For example, in the sprint stage, the goal is to increase the conversion rate, and the advertising resource volume remains at 300,000 yuan. By analyzing user behavior data, it is found that certain users (such as those who have visited the product page but not completed the purchase) have higher conversion potential. Therefore, the advertising targeting settings can be adjusted to specifically target these user groups with remarketing ads. For example, show personalized promotional ads (such as "The items in your shopping cart now enjoy an extra discount") to these users, thus effectively improving the conversion rate.

[0117] Budget and frequency control, reasonably arrange the daily budget and ad exposure frequency according to the new resource volume, avoid resource waste or insufficient exposure, and ensure the maximization of the continuous ad delivery effect.

[0118] For example, in the preheating stage, the goal is to increase brand exposure, and the ad resource volume is reduced to 200,000 yuan. To avoid resource waste, the ad budget and exposure frequency need to be reasonably controlled. Suppose the daily budget is 20,000 yuan. If the ad frequency is too high (the same user sees the ad multiple times in a day), it may cause user fatigue and lead to a decline in the effect. Therefore, an ad frequency upper limit can be set. For example, each user can see the ad at most 3 times a day. This can not only ensure the wide exposure of the ad but also avoid overexposing the same user, improving the resource utilization efficiency.

[0119] This embodiment also provides an ad delivery system based on a multi-objective optimization algorithm, including: a resource allocation module, a strategy generation module, an effect analysis module, a weight optimization module, and a strategy optimization module; the resource allocation module is used to set the optimization goal for each stage, allocate preliminary weights to the optimization goal, divide the ad delivery cycle according to the preliminary weights, and perform resource allocation; the strategy generation module is used to generate and execute a preliminary ad delivery strategy according to the optimization goal and the preliminary weights; the effect analysis module is used to collect the real-time performance data and resource consumption in the preliminary ad delivery process in real time and analyze the ad delivery effect of the previous stage; the weight optimization module is used to dynamically adjust the preliminary weights of the optimization goal for the next stage according to the analysis result; the strategy optimization module is used to optimize the preliminary ad delivery strategy based on the weights of the adjusted optimization goal.

[0120] This embodiment also provides a computer device applicable to the case of the ad delivery method based on a multi-objective optimization algorithm, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the ad delivery method based on a multi-objective optimization algorithm as proposed in the above embodiment.

[0121] The computer device may be a terminal, which includes a processor, a memory, a communication interface, a display screen, and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner, and the wireless manner can be implemented through WIFI, carrier networks, NFC (Near Field Communication), or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0122] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the advertising placement method based on the multi-objective optimization algorithm as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM for short), Electrically Erasable Programmable Read-Only Memory (EEPROM for short), Erasable Programmable Read-Only Memory (EPROM for short), Programmable Read-Only Memory (PROM for short), Read-Only Memory (ROM for short), magnetic memory, flash memory, a magnetic disk, or an optical disc.

[0123] In summary, through the following steps: setting the optimization objectives for each stage and allocating preliminary weights, the present invention realizes a reasonable division of the advertising placement cycle and precise allocation of resources, ensuring that advertising objectives in different stages, such as exposure volume, click-through rate, and conversion rate, can be fully concerned, and avoiding waste of resources; during the advertising placement process, advertising performance data and resource consumption are collected in real time, the effects of the previous stage are analyzed, and the optimization objective weights for the next stage are dynamically adjusted based on the analysis results, enabling the advertising strategy to flexibly respond to changes in the market and user behavior, and ultimately improving the overall effect of advertising placement and resource utilization efficiency.

[0124] Example 2. Referring to Table 1, this is the second example of the present invention. To further verify the technical solution of the present invention, experimental simulation data of the advertising placement method based on the multi-objective optimization algorithm is given.

[0125] The total advertising placement period is 30 days, and the advertising budget is 1 million yuan. According to the market promotion requirements, the advertising placement period is divided into three stages: the preheating stage (the first 5 days), the main promotion stage (the next 15 days), and the sprint stage (the last 10 days). The optimization objectives for each stage are as follows: maximizing the exposure volume in the preheating stage, maximizing the click-through rate in the main promotion stage, and maximizing the conversion rate in the sprint stage. The allocation ratio of advertising resources is: 20% in the preheating stage, 50% in the main promotion stage, and 30% in the sprint stage. Weight allocation is performed on the optimization objectives as follows:

[0126] E1 = 0.8, E2 = 0.7, E3 = 0.9, λ = 0.5, γ = 2, E max = 0.9, μ = 0.1,;

[0127] Substitute the above data into the following expression (only w1 is shown here):

[0128]

[0129] The calculation results are shown in Table 1 below:

[0130] Table 1 Preliminary weight allocation table for optimization objectives

[0131] <![CDATA[Initial weight w1 of the optimization objective]]> 1.599 <![CDATA[Initial weight w2 of the optimization objective]]> 1.7077 <![CDATA[Initial weight w3 of the optimization objective]]> 1.313

[0132] First, in this experiment, according to the overall advertising placement period and market promotion requirements, the advertising placement is divided into three stages: the preheating stage, the main promotion stage, and the sprint stage. The optimization objectives for each stage are the exposure volume, click-through rate, and conversion rate respectively. The advertising budget is initially allocated according to the requirements of different stages as follows:

[0133] R = 1 million yuan, D1 = 0.6, D2 = 0.8, D3 = 0.5, U1 = 0.4, U2 = 0.7, U3 = 0.9;

[0134] Substitute the above data into the following expression (only R1 is shown here):

[0135]

[0136] The calculation results are shown in Table 2 below:

[0137] Table 2 Advertising resource allocation table

[0138] <![CDATA[Advertising resource allocation amount R10,000 yuan]]> 14.45 <![CDATA[Advertising resource allocation amount R20,000]]> 32.73 <![CDATA[Advertising resource allocation amount R30,000 yuan]]> 22.35

[0139] Secondly, based on information entropy calculation and user behavior feature analysis, the advertising resource allocation weights for each stage are dynamically adjusted. Through the analysis of user click, dwell time, and conversion behavior collected in real time by the advertising platform, the resource allocation is further optimized, making the advertising placement more accurate and effective, as follows:

[0140] EI1(t) = 0.0685, EI2(t) = 0.075, EI3(t) = 0.082, θ1 = 0.05, θ2 = 0.04, θ3 = 0.06;

[0141] Substitute the above data into the following expression (only w'1(t + 1) is shown here):

[0142]

[0143] The calculation results are shown in Table 3 below:

[0144] Table 3 Initial weight update results of the optimization objective

[0145] <![CDATA[Updated weight w'1(t + 1)]]> 0.3459 <![CDATA[Updated weight w'2(t + 1)]]> 0.3700 <![CDATA[Updated weight w'3(t+1)]]> 0.2841

[0146] Finally, to verify the effectiveness of the optimization strategy proposed in the present invention, a comparative experiment is conducted in this embodiment to test the effects of the traditional advertising placement strategy and the optimization strategy proposed in the present invention respectively. The traditional advertising placement strategy uses a method of evenly allocating resources according to a fixed ratio, without considering the market demand differences and user behavior characteristics in each stage during the entire placement period. Specifically, the traditional strategy simply divides the advertising cycle into a preheating stage, a main promotion stage, and a sprint stage, and allocates a 1 million yuan advertising budget to each stage according to the ratio of 20%, 50%, and 30% without dynamic adjustment. This method lacks pertinence and cannot optimize the advertising effect based on real-time feedback. The main parameters of the comparative experiment include exposure, click-through rate, conversion rate, and advertising budget consumption, etc.

[0147] Specifically, it is shown in Table 4 below:

[0148] Table 4 Comparison table of advertising placement effects

[0149]

[0150]

[0151] Through the data analysis of the above table, it can be clearly seen that the advertising effect of the strategy of the present invention in each stage is significantly better than that of the traditional strategy. By improving the click-through rate and conversion rate, the present invention achieves higher advertising effects with the same or lower budget consumption. For example, in the sprint stage, the click-through rate of the strategy of the present invention is 1.85%, which is much higher than 1.12% of the traditional strategy; at the same time, the conversion rate reaches 2.47%, significantly exceeding 1.98% of the traditional strategy. In addition, the click cost and conversion cost of the strategy of the present invention are also lower than those of the traditional strategy, further indicating its advantage in cost-effectiveness.

[0152] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. An advertisement delivery method based on a multi-objective optimization algorithm, characterized in that: include, Set optimization goals for each stage, assign preliminary weights to the optimization goals, divide the advertising delivery cycle according to the preliminary weights, and allocate resources; Generate and execute preliminary advertising delivery strategies based on optimization goals and preliminary weights; Collect real-time performance data and resource consumption during the initial advertising process, and analyze the advertising effects of the previous stage; Dynamically adjust the initial weights of the next stage optimization objectives based on the analysis results; Optimize the initial advertising delivery strategy based on the adjusted initial weights of the optimization goals; The optimization goal of each stage is set, and preliminary weights are assigned to the optimization goals. The specific steps are as follows: Determine the total advertising cycle and total advertising resources based on business needs and marketing plans; According to the total advertising cycle, the advertising cycle is divided into the warm-up stage, the main promotion stage and the sprint stage according to uneven time; In the warm-up phase, maximizing exposure is defined as the optimization goal; In the main promotion stage, define maximizing click-through rate as the optimization goal; In the sprint phase, define maximizing conversion rate as the optimization goal; By performing exponential and sinusoidal processing and difference adjustment on the information entropy of each target, a preliminary weight is assigned to the optimization target. The expression is: ； in, It is The initial weights of the optimization objectives, It is The information entropy value of the optimization target is is the total number of optimization objectives, is the index variable of the optimization target, is the adjustment coefficient for adjusting the weight distribution difference of the information entropy value of the optimization target, It is the adjustment coefficient that controls the weakening strength of the weight of the high entropy target. It is the maximum value of all optimization target information entropy; The real-time performance data during the advertising delivery process includes exposure, clicks, conversions, click-through rate and conversion rate; The resource consumption includes advertising budget consumption, click cost, display cost and conversion cost in each time period; The specific steps of analyzing the advertising effect in the pre-stage are as follows: Through the relationship between exposure, clicks, conversions and budget consumption, logarithmic transformation is used to weaken the maximum impact of exposure and resource consumption. At the same time, the click-through rate and conversion rate are used to measure the advertising effect. The expression is: ； in, It is Stages at a time point Exposure, It is Stages at a time point of clicks, It is Stages at a time point The amount of conversion, It is Stages at a time point The resource consumption, is a small positive number that prevents the denominator from being zero, It is Stages at a time point The advertising effectiveness index of the stage, Indicates Stages at a time point The amount of advertising resources invested; The specific steps of dynamically adjusting the initial weight of the optimization target in the next stage according to the analysis results are as follows: Through the effect index of the current stage and the effect threshold of each optimization target, the hyperbolic tangent function is used to smooth the difference between the effect index and the threshold, and the initial weight of the optimization target is dynamically adjusted. The expression is: ； in, is the adjustment coefficient that controls the initial weight adjustment range of the optimization target. It is The effect threshold corresponding to the optimization goal, Indicates The optimization goal is at the time point The initial weight of .

2. The advertisement delivery method based on the multi-objective optimization algorithm according to claim 1, characterized in that: The specific steps of dividing the advertising delivery cycle and allocating resources according to the preliminary weights are as follows: The advertising resources of each stage divided by non-uniform time are allocated non-uniformly according to the market demand and user behavior changes in each stage. The expression is: ； in, It is The amount of advertising resources in each stage, Total advertising inventory, It is The market activity at each stage, It is in Behavioral characteristics of users at each stage, It is the index variable of the ad serving stage.

3. The advertisement delivery method based on the multi-objective optimization algorithm according to claim 2, characterized in that: The preliminary advertising delivery strategy includes advertising channel selection, advertising creative design, advertising targeting setting, and budget and frequency.

4. The advertisement delivery method based on a multi-objective optimization algorithm according to claim 3, characterized in that: The specific steps of optimizing the initial advertising delivery strategy based on the adjusted optimization target weight are as follows: Weights based on the adjusted optimization goal , reallocate the advertising resources in each stage; According to the adjusted The amount of advertising inventory in each stage , optimize the initial advertising strategy.

5. An advertisement delivery system based on a multi-objective optimization algorithm, based on the advertisement delivery method based on a multi-objective optimization algorithm according to any one of claims 1 to 4, characterized in that: Including resource allocation module, strategy generation module, effect analysis module, weight optimization module and strategy optimization module; Resource allocation module, which is used to set the optimization target for each stage, assign preliminary weights to the optimization targets, divide the advertising delivery cycle according to the preliminary weights, and allocate resources; A strategy generation module is used to generate and execute a preliminary advertising delivery strategy based on the optimization objectives and preliminary weights; The effect analysis module is used to collect real-time performance data and resource consumption during the initial advertising process and analyze the advertising effect of the previous stage; The weight optimization module is used to dynamically adjust the initial weight of the optimization target in the next stage according to the analysis results; The strategy optimization module is used to optimize the initial advertising delivery strategy based on the weights of the adjusted optimization objectives.

Images