Advertisement putting control method and device and storage medium

Abstract

The invention provides an advertisement putting control method and device and a storage medium, and the method comprises the steps: collecting input data, including advertisement platform data, internal business data and competitor data, according to a set frequency, carrying out the analysis and calculation of the input data, obtaining a risk score and a corresponding risk level, and carrying out the calculation of the risk score; and judging whether a risk condition is met or not according to the risk level, executing a corresponding adjustment action according to the risk level when the risk condition is met, and starting a corresponding protection mechanism and a corresponding notification mechanism. According to the method, multi-condition comprehensive evaluation is realized by quantifying risk scores, potential high risks are identified through superposition logic and intervention is performed in advance, and the accuracy of advertisement putting risk judgment is improved; differentiated adjustment actions are matched for different risks in different ranges through a risk level strategy matrix, and fine control over risk and action strength matching is achieved; through the cooling mechanism bound with the risk level, the strategy can be adapted in real time according to the risk change, and the adaptive ability to the complex putting scene is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of advertisement delivery, and in particular to an advertisement delivery control method and device and a storage medium. BACKGROUND

[0002] In the field of digital advertisement delivery, in order to reduce invalid consumption, advertisement delivery control systems with basic automatic control functions have appeared in the industry. Such systems can monitor some indicators in the advertisement delivery process and perform simple suspension or price adjustment operations when the indicators are abnormal. For example, some systems can monitor conversion costs and automatically suspend delivery when the conversion costs exceed a preset value; another system can track click rates and lower the bid when the click rate is below a certain threshold. These systems reduce the frequency of manual intervention to some extent and save some management costs for advertisers.

[0003] However, the current advertisement delivery control technology still cannot meet the needs of precise control and risk avoidance in complex delivery scenarios, which is specifically manifested in the following aspects:

[0004] The protection actions of the existing advertisement delivery control system are relatively fixed, and usually only two states of "suspension" or "non-suspension" are set, without formulating a hierarchical response strategy according to the risk level. For example, the same suspension or price adjustment range is used for different severity of risks, which cannot achieve the fine control of "slight risk fine-tuning, serious risk strong intervention". At the same time, the existing advertisement delivery control system lacks a scientific cooling mechanism and state locking logic after executing the intervention action, which may prematurely resume delivery when the risk is not completely removed, or still maintain unnecessary restrictions when the risk has been eliminated, affecting the advertisement delivery effect. SUMMARY

[0005] Therefore, the present application provides an advertisement delivery control method, device and storage medium, which can improve the control accuracy of advertisement delivery risk and the delivery effect.

[0006] In a first aspect, an advertisement delivery control method is provided, which comprises:

[0007] Collecting input data at a set frequency, the input data comprising advertisement platform data, internal business data and competitor data;

[0008] Analyzing and calculating the input data to obtain a risk score and a corresponding risk level, and determining whether a risk condition is met according to the risk level;

[0009] When it is determined that the risk condition is met, performing a corresponding adjustment action according to the risk level;

[0010] Starting a protection mechanism and a notification mechanism corresponding to the risk level.

[0011] Further, when it is determined that the normal state, then the current advertising settings are maintained.

[0012] Further, before collecting the input data at a set frequency, the method further comprises:

[0013] automatically loading a basic parameter library, a risk judgment rule set, and a risk level strategy matrix, wherein,

[0014] the basic parameter library includes at least two standard indicators: a conversion cost target price and a conversion cost threshold, an LTV expected value, a normal fluctuation range of a click rate, and a large promotion identification feature of a competitor;

[0015] the risk judgment rule set includes weights and superposition logic when each standard indicator is abnormal;

[0016] the risk level strategy matrix stores a risk score corresponding to each risk level, a corresponding adjustment action, and a protection mechanism.

[0017] Further, the input data is analyzed and calculated to obtain a risk score and a corresponding risk level, and whether the risk condition is met is determined according to the risk level, including:

[0018] a plurality of risk indicators are calculated according to the input data;

[0019] each risk indicator is compared with a corresponding standard indicator to obtain an abnormal indicator;

[0020] the abnormal indicators are cooperatively calculated to obtain a risk score and a corresponding risk level;

[0021] when the risk score is within the range corresponding to the normal risk level and there is no abnormal indicator, it is determined that the normal state is met;

[0022] when the risk score is not within the range corresponding to the normal risk level, or the risk score is within the range corresponding to the normal risk level but the same risk indicator is an abnormal indicator for a continuous preset number of times, it is determined that the risk condition is met.

[0023] Further, the abnormal indicators are cooperatively calculated to obtain a risk score and a corresponding risk level, including:

[0024] all abnormal indicators are extracted;

[0025] a risk score is calculated according to the weights and superposition logic of each abnormal indicator in the risk judgment rule set;

[0026] the risk score is compared with the risk level strategy matrix to obtain a risk level.

[0027] Further, when it is judged that the risk condition is met, corresponding adjustment actions are performed according to the risk level, including:

[0028] When the risk score is in the score range of the first risk level, a bid API interface of an advertising platform is called to reduce the current bid of the advertisement by a first preset proportion, and a second preset proportion is transferred from the current plan budget to a backup plan to obtain parameters after the first risk level adjustment;

[0029] When the risk score is in the score range of the second risk level, non-core keywords are selected from all keywords currently launched according to a preset core keyword library, and a keyword state interface of the advertising platform is called to suspend the non-core keywords in batches;

[0030] When the risk score is in the score range of the third risk level, a plan state interface of the advertising platform is called to suspend the current advertising plan.

[0031] Further, a protection mechanism and a notification mechanism corresponding to the risk level are started, including:

[0032] When the risk level is the first risk level, a first duration cooling timer is started, the parameters after the first risk level adjustment are locked during the first duration, and the parameters after the first risk level adjustment are written into a monitoring dashboard without active alarm;

[0033] When the risk level is the second risk level, a second duration cooling timer is started, the non-core keywords are in a suspended state during the second duration, and alarm information is sent through an internal alarm interface, wherein the second duration is greater than the first duration;

[0034] When the risk level is the third risk level, a third duration cooling timer is started, the current advertising plan is resumed until a user's resuming instruction is received, and a voice notification interface is called to trigger a telephone alarm and generate a diagnosis report, wherein the third duration is greater than the second duration, and the second duration is greater than the first duration.

[0035] Further, when the risk score drops to the score range corresponding to a lower risk level during the cooling duration corresponding to each risk level, adjustment actions corresponding to the risk level after the drop are performed.

[0036] In a second aspect, an embodiment of the present application provides an advertising launching control device, which comprises:

[0037] A collection module is configured to collect input data at a set frequency, wherein the input data includes advertising platform data, internal business data, and competitor data;

[0038] The analysis and judgment module is configured to analyze and calculate the input data, obtain a risk score and a corresponding risk level, and determine whether the risk condition is met according to the risk level.

[0039] The adjustment module is configured to perform a corresponding adjustment action according to the risk level when it is determined that the risk condition is met.

[0040] The protection and notification module is configured to start a protection mechanism and a notification mechanism corresponding to the risk level.

[0041] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, and the storage medium stores a computer program. When the computer program is executed, the method in any one of the first aspect is performed.

[0042] In a fourth aspect, an embodiment of the present application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor is configured to execute the computer program to perform the method in any one of the first aspect.

[0043] The technical solution provided by the present application can load a risk judgment rule set containing multi-index weight and superposition logic, perform collaborative analysis on the advertisement platform data, internal business data (such as LTV), and competitor data (such as competitor promotion information), realize multi-condition comprehensive evaluation through quantitative risk score, and identify potential high risk and intervene in advance through superposition logic. On the contrary, when a single index is temporarily abnormal, the system determines it as low risk through risk score calculation, avoids the “one-size-fits-all” type of false suspension in the prior art, and significantly improves the accuracy of advertisement delivery risk judgment.

[0044] Meanwhile, the technical solution provided by the present application can match different adjustment actions for different risk score ranges through a risk level strategy matrix. Low risk (such as the first risk level) is only adjusted flexibly through “price reduction + partial budget transfer” to maintain basic delivery; medium risk (the second risk level) is accurately suspended for non-core keywords, and core traffic is reserved; and high risk (the third risk level) is only suspended for the whole plan. This hierarchical response avoids the traffic loss caused by “small risk and strong intervention” in the prior art, solves the problem of expanding invalid consumption caused by “high risk and weak intervention”, and realizes fine control of “risk and action intensity matching”.

[0045] Meanwhile, the technical solution provided by the present application can set a short cooling period for low risk to avoid frequent price adjustment, set manual recovery for high risk to ensure manual intervention in key decisions, and automatically execute a lower level action if the risk score decreases during the cooling period. Compared with the static control in the prior art, the present application can adapt the strategy in real time according to the risk change, and significantly improve the self-adaptation ability to complex delivery scenarios. Attached Figure Description

[0046] Figure 1 This is a flowchart of an advertising placement control method provided in an embodiment of the present invention.

[0047] Figure 2 This is a flowchart illustrating the process of obtaining risk scores and corresponding risk levels in this embodiment of the invention, and determining whether risk conditions are met based on the risk levels.

[0048] Figure 3 This is a flowchart illustrating the process of obtaining risk scores and corresponding risk levels in an embodiment of the present invention.

[0049] Figure 4 This is a flowchart illustrating the implementation of corresponding adjustment actions based on the risk level in an embodiment of the present invention.

[0050] Figure 5 This is a flowchart illustrating the implementation of the protection and notification mechanisms corresponding to the risk level in this embodiment of the invention.

[0051] Figure 6 This is a schematic diagram of the structure of an advertising placement control device provided in an embodiment of the present invention.

[0052] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0054] See Figure 1 , Figure 1 This is a flowchart of an advertising delivery control method provided by an embodiment of the present invention. The method includes the following steps:

[0055] Step 11: Collect input data at a set frequency. The input data includes advertising platform data, internal business data, and competitor data.

[0056] In step 11, the "set frequency" can be flexibly configured according to the advertising scenario, for example, 5 minutes / time or 10 minutes / time, taking into account the real-time data and the load of the advertising system; the "input data" is the basis for judgment by the advertising system, wherein the advertising platform data can include data such as consumption amount, conversion number, click rate, etc. directly reflecting the effect of advertising, internal business data covers user LTV and other indicators reflecting long-term benefits, and competitor data includes competitor promotion activities, keyword competition heat and other external environmental information. The implementation can be completed by automatically collecting through the interfaces of the advertising platform API, internal business system and competitor monitoring tool, without manual intervention.

[0057] The effect of this step is to ensure that the system can continuously obtain comprehensive and real-time raw data, providing complete input for subsequent risk judgment, and avoiding judgment deviation due to data missing or lag.

[0058] Step 12, analyzing and calculating the input data to obtain a risk score and a corresponding risk level, and judging whether the risk condition is met according to the risk level.

[0059] In step 12, the core process of converting the raw input data collected in step 11 into risk assessment results includes calculating risk indicators such as conversion cost, click rate fluctuation, LTV target rate, etc., and then identifying abnormal indicators by comparing standard indicators in the basic parameter library; the risk score is a quantitative representation of multi-dimensional risk indicators, calculated through the weight in the risk judgment rule set (such as higher abnormal weight of conversion cost than click rate) and the superposition logic (such as amplification of multiple abnormal risks); the risk level is the classification result of the risk score, used to determine the severity of the risk.

[0060] The implementation relies on a pre-set basic parameter library, a risk judgment rule set and a risk level strategy matrix to automatically complete index calculation, abnormal identification, risk score collaborative operation and level matching. The effect of this step is to convert scattered indicators into quantifiable and comparable risk levels, providing accurate judgment basis for subsequent actions, and solving the problem of one-sidedness of single indicator judgment in the prior art.

[0061] Step 13, when it is judged that the risk condition is met, performing corresponding adjustment actions according to the risk level.

[0062] The risk condition is met when the risk level reaches a degree requiring intervention (e.g., risk score ≥ 70); the corresponding adjustment action is strictly bound to the risk level, for example, a low risk level (e.g., 70 ≤ risk score < 80) performs bid reduction and budget transfer, a medium risk level (e.g., 80 ≤ risk score < 90) suspends non-core keywords, and a high risk level (e.g., risk score ≥ 90) directly suspends the plan. The implementation can automatically execute the preset action by calling the bid adjustment interface, budget allocation interface, keyword state interface, and plan state interface of the advertising platform. The effect of this step is to take differentiated intervention for different risk levels, avoiding traffic loss caused by excessive adjustment in low risk, and quickly stopping loss in high risk, thereby achieving precise matching of risk and action.

[0063] Step 14, starting the protection mechanism and notification mechanism corresponding to the risk level.

[0064] In this step, the protection mechanism includes a cooling time corresponding to the risk level (e.g., 30 minutes for low risk and 1 hour for medium risk) and state locking (maintaining the current adjusted parameters during the cooling period) to prevent repeated triggering of the same action in a short period of time; the notification mechanism pushes information according to the risk level, for example, low risk is only recorded to the monitoring board, medium risk sends an alarm through an internal system, and high risk triggers a phone alarm and generates a diagnosis report. The implementation manages the cooling period through a timer, locks the adjustment result through a state marker, and completes notification pushing through interfaces such as SMS and voice. The effect of this step is to ensure the effectiveness of the adjustment action, avoid frequent operations that interfere with the stability of the placement, and ensure that relevant personnel are aware of the risk status in a timely manner, allowing for manual intervention if necessary, thereby balancing automation and controllability.

[0065] Step 15, when it is determined that the risk condition is not met, i.e., normal state, then the current advertising placement settings are maintained.

[0066] The normal state refers to a risk score within a safe range and no abnormal indicators accumulated, at which time the current advertising placement settings are maintained, i.e., the existing bid, budget allocation, keyword state, and other parameters remain unchanged. The implementation is completed through a system internal state marker without the need for additional adjustment instructions, and only the normal state is recorded to the monitoring board. The effect of this step is to avoid unnecessary adjustments that interfere with the advertising placement rhythm, ensure the continuous acquisition of normal traffic, and ensure that the system does not affect the placement effect when the risk is controllable, thereby complementing the intervention action under the risk condition and achieving a balance between placement efficiency and risk control.

[0067] In some other embodiments of the present application, before step 11, the method can further include:

[0068] The automatic loading of the basic parameter library, the risk judgment rule set and the risk level strategy matrix, wherein the basic parameter library includes at least two standard indexes: conversion cost target price and conversion cost threshold, LTV expected value, click rate normal fluctuation range, competitor big promotion identification feature; the risk judgment rule set includes the weight and superposition logic when each standard index is abnormal; the risk level strategy matrix stores the risk score corresponding to each risk level, the corresponding adjustment action and the protection mechanism.

[0069] The automatic loading of the basic parameter library, the risk judgment rule set and the risk level strategy matrix before step 11 is a core configuration link for realizing the "intelligent pause" of the advertisement launching control, and provides a unified quantitative standard and logical basis for subsequent data collection, risk judgment and action execution.

[0070] The "basic parameter library" is a benchmark for the advertisement launching system to judge whether the index is abnormal, and the standard indexes contained therein are all directly related to the core target of the advertisement launching: the conversion cost target price and the conversion cost threshold clearly define the acceptable upper limit of the cost (such as a target price of 100 yuan, a 150% threshold of 150 yuan, and a 200% threshold of 200 yuan), which is the core standard for measuring the rationality of immediate consumption; the LTV expected value sets the minimum standard for the long-term value of the user (such as an expected 30-day LTV of 200 yuan), which is used to evaluate the long-term income of the launching; the click rate normal fluctuation range (such as allowing a ±30% fluctuation) defines the reasonable change interval of the click rate, avoiding the misjudgment of short-term random fluctuations as abnormal; the competitor big promotion identification feature (such as an activity marker containing keywords such as "618" and "limited discount", and a threshold of competition heat of 60 points) provides a quantitative basis for identifying the external competitive environment.

[0071] The implementation is to read the above-mentioned standard indexes from the preset database or configuration file through the system initialization interface, and support the user to configure in advance or modify later through a visual interface. The effect of this setting is to provide a "scale" for subsequent risk judgment, ensuring that the abnormal determination of different indexes has a unified, traceable standard, avoiding the judgment deviation caused by ambiguous benchmark.

[0072] The "risk judgment rule set" is the core algorithm logic carrier for realizing the "multi-condition cooperative judgment", wherein the "weight of each standard index abnormality" reflects the risk contribution degree of different indexes (for example, the conversion cost abnormality weight is 0.4, and the click rate abnormality weight is 0.2), and it is clear which index abnormality is more dangerous; the superposition logic defines the risk amplification rule when multiple indexes are abnormal (for example, 2 abnormal risk points x 1.3, and 3 abnormal indexes x 1.6), and solves the problem of how to quantify the risk after the superposition of multiple slight abnormalities. The implementation manner is to store the weight coefficient and the superposition formula in the form of code logic or a configuration table, and the system can directly call the risk score calculation after loading. The effect of this rule set is to convert scattered abnormal indexes into a comparable risk score, so that the system can upgrade from "single index judgment" to "multi-dimensional comprehensive evaluation", and avoids the misjudgment caused by ignoring the correlation of indexes in the prior art.

[0073] The "risk level strategy matrix" is a mapping bridge between the risk score and the execution action, wherein the "risk score corresponding to each risk level" delimits the interval of the risk degree (for example, 70≤risk score<80 is the first risk level); the "corresponding adjustment action" clearly defines the specific operation under different risk levels (for example, the first risk level reduces the price by 20%+transfers 30% of the budget); and the "corresponding protection mechanism" includes the cooling time (for example, the first risk level cools for 30 minutes), the state locking rule and the like. The implementation manner is to store the risk level, the risk score range, the action instruction and the protection parameter as an associated data table, and the system can quickly match the corresponding execution scheme through the risk score. The effect of this matrix is to realize the precise binding of "risk degree and coping intensity", so that the system can take differentiated measures according to the risk level, and avoids the extreme processing mode of "no intervention or strong pause" in the prior art, and balances the risk control and the continuity of the delivery.

[0074] In some embodiments of the present application, as shown in Figure 2 Figure 2 is an implementation flowchart for obtaining the risk score and the corresponding risk level, and judging whether the risk condition is met according to the risk level in the embodiments of the present application, that is, step 12 can include the following steps:

[0075] Step 121, calculating a plurality of risk indexes according to the input data.

[0076] ​This step is to convert the original data into quantitative indicators that can be used for risk judgment. The input data is the raw information collected from the advertising platform, internal business systems and competitor monitoring tools, such as consumption amount, conversion number, click volume, exposure volume, user LTV value, competitor promotion marker, etc. Through the preset calculation formula, these raw data can be converted into specific risk indicators, such as conversion cost obtained by dividing consumption amount by conversion number, click rate obtained by dividing click volume by exposure volume, LTV compliance rate obtained by the ratio of actual LTV value to expected LTV value, competitor risk index calculated according to competitor promotion marker and competition heat, etc. These risk indicators are the direct basis for subsequent judgment of whether it is abnormal or not, and provide specific quantitative carriers for risk assessment.

[0077] Step 122, compare each risk indicator with the corresponding standard indicator to obtain an abnormal indicator.

[0078] Step 122 is the key link of risk identification. Among them, the standard indicators come from the basic parameter library, such as the target price and threshold of conversion cost, the normal fluctuation range of click rate, etc. Each risk indicator calculated in step 121 is compared with the corresponding standard indicator one by one, and if a certain risk indicator exceeds the reasonable range set by the standard indicator, it will be marked as an abnormal indicator. For example, when the calculated conversion cost exceeds the conversion cost threshold set in the basic parameter library, the conversion cost indicator will be determined as an abnormal indicator. This step realizes the preliminary screening of various risk signals in the advertising process, and provides abnormal objects for subsequent collaborative calculation.

[0079] Step 123, perform collaborative calculation on the abnormal indicators to obtain a risk score and a corresponding risk level.

[0080] Step 123 is the core step of realizing multi-condition collaborative judgment. Based on the weight and superposition logic of each abnormal indicator in the risk judgment rule set, all abnormal indicators obtained in step 122 are comprehensively calculated. First, calculate the contribution value of each abnormal indicator in the risk score according to its weight, then calculate the total risk score according to the superposition logic (such as amplification coefficient when multiple abnormalities occur). Then, compare the calculated risk score with the risk score range corresponding to each risk level in the risk level strategy matrix, to determine the corresponding risk level.

[0081] Suppose that in the risk judgment rule set of a certain advertising delivery control system, three key indicators and their weights when abnormal are preset: conversion cost (weight 40%), click rate (weight 30%), competitor promotion intensity (weight 30%), and the superposition logic is "the sum of the contribution values of each abnormality x the amplification coefficient of multiple abnormalities" (when there are 2 or more abnormal indicators, the amplification coefficient is 1.2).

[0082] For example, if the input data in an analysis shows that the conversion cost exceeds the threshold value (determined to be abnormal), the click rate is lower than the lower limit of the normal fluctuation (determined to be abnormal), and the competitor's promotion intensity does not reach the abnormal standard (no abnormality), first, the single abnormal contribution value is calculated according to the weight: the conversion cost abnormality contribution value is 40 points (the upper limit value of the weight), the click rate abnormality contribution value is 30 points (the upper limit value of the weight), and the sum is 70 points. Since there are two abnormal indicators, the amplification coefficient 1.2 is triggered, and the final risk score is 70x1.2=84 points.

[0083] For another example, only the conversion cost indicator is abnormal, and there is no other abnormal indicator, at this time, the amplification coefficient does not need to be triggered, and the risk score is directly 40 points (only the weight contribution value of the indicator).

[0084] Through such a synergistic calculation, the difference in the influence degree of different abnormal indicators (weight difference) can be reflected, and the risk superposition effect (amplification coefficient) when multiple abnormal indicators coexist can be reflected. The final risk score can objectively quantify the overall risk level and provide accurate basis for subsequent matching of risk levels.

[0085] For example, if the risk score obtained through synergistic calculation is 75 points, and the risk score range corresponding to the first risk level in the risk level strategy matrix is 70≤risk score<80, then the risk level corresponding to the risk score is the first risk level. This step enables the system to comprehensively evaluate the overall risk degree from multiple abnormal indicators, avoiding the one-sidedness of single indicator judgment.

[0086] Step 124, when the risk score is in the range corresponding to the normal risk level and there is no abnormal indicator, it is determined to be a normal state.

[0087] Step 124 defines the determination standard of the normal state. The range corresponding to the normal risk level is the safety interval set in the risk level strategy matrix. When the risk score is in this interval and there is no abnormal indicator, it means that each indicator in the advertisement launching process is within a reasonable range, and there is no risk that needs to be intervened. The system therefore determines that it is in a normal state. At this time, the system will maintain the current advertisement launching settings and not perform adjustment actions to ensure the continuity and stability of the advertisement launching.

[0088] Step 125, when the risk score is not in the range corresponding to the normal risk level, or the risk score is in the range corresponding to the normal risk level but the same risk indicator is continuously abnormal for a preset number of times, it is determined that the risk condition is met.

[0089] Step 125 defines the risk condition that needs to be intervened. When the risk score exceeds the range corresponding to the normal risk level, it means that there is obvious risk in the advertisement delivery, and the system needs to take corresponding adjustment actions; and when the risk score is within the normal range, but the same risk indicator appears abnormally for many times in succession, it may mean that the indicator has potential and continuous risk. In order to avoid the accumulation and expansion of risk, the system will also judge that the risk condition is met. This step ensures that the system can identify various risk conditions in time, whether it is obvious risk or potential risk, which can trigger the subsequent adjustment action, so as to effectively control the risk of advertisement delivery.

[0090] In some embodiments of the present application, as shown in Figure 3 Figure 3 In some embodiments of the present application, the implementation flowchart of obtaining the risk score and the corresponding risk level is as follows, that is, step 123 cooperatively calculates the abnormal indicators to obtain the risk score and the corresponding risk level, which can be realized by the following steps:

[0091] Step 1231 extracts all abnormal indicators.

[0092] Step 1231 is a pre-screening link of risk score calculation. In step 122, the system has compared each risk indicator with the standard indicator and marked the abnormal indicators (such as conversion cost exceeding the threshold, click rate decreasing amplitude exceeding the normal range, etc.). This step extracts the abnormal indicators from the original indicator set by traversing all the marked abnormal indicators, forming a subset containing only abnormal indicators. For example, if it is found in a certain analysis that the conversion cost, click rate and LTV target rate are abnormal, and the exposure, CPC and other indicators are normal, this step only extracts the three abnormal indicators of conversion cost, click rate and LTV target rate, providing a clear input object for subsequent cooperative calculation, avoiding irrelevant indicators interfering with risk assessment.

[0093] Step 1232 calculates the risk score according to the weight and superposition logic of each abnormal indicator in the risk judgment rule set.

[0094] ​Steps 1232 constitute the core calculation process for quantifying risk. The risk judgment rule set predefines the weight of each abnormal indicator (e.g., conversion cost weight 0.4, click-through rate weight 0.2, LTV achievement rate weight 0.3) and the superposition logic for multiple abnormal indicators (e.g., risk score = ∑(single abnormal base score × weight) × superposition coefficient, where the superposition coefficient increases with the number of abnormal indicators: 1 abnormal indicator = 1.0, 2 abnormal indicators = 1.3, 3 abnormal indicators = 1.6). The system first converts the degree of deviation from the standard indicator for each abnormal indicator into a base score (e.g., a conversion cost exceeding the threshold by 50% corresponds to a base score of 80), then multiplies it by the weight of the indicator to obtain the single contribution value; then it sums all the single contribution values, and finally applies the corresponding amplification coefficient according to the superposition logic to obtain the final risk score. For example, if the abnormal contribution value of conversion cost is 32 points (80×0.4) and the abnormal contribution value of click rate is 12 points (60×0.2), and the superposition coefficient of the two abnormalities is 1.3, then the risk score = (32+12)×1.3 = 57.2 points.

[0095] Step 1233: Compare the risk score with the risk level strategy matrix to obtain the risk level.

[0096] Step 1233 is the matching process of mapping quantified risk to specific handling levels. The risk level strategy matrix predefines the correspondence between risk score ranges and risk levels (e.g., 0-69 points is level 0 (normal state), 70-79 points is the first risk level, 80-89 points is the second risk level, and ≥90 points is the third risk level). The system compares the risk score calculated in step 1232 with each interval in the matrix to determine its corresponding risk level. For example, if the calculated risk score is 57.2 points, it falls within the range of level 0, so the current risk level is determined to be level 0. This mapping relationship is directly related to subsequent adjustment actions and protection mechanisms (e.g., level 0 only marks abnormality, while the first risk level requires price reduction and budget transfer), ensuring that the advertising placement control system can take differentiated responses according to the degree of risk, achieving precise control.

[0097] In other embodiments of the present invention, such as Figure 4 As shown, Figure 4 This is a flowchart illustrating the implementation of corresponding adjustment actions based on the risk level in this embodiment of the invention. Specifically, step 13 can be achieved through the following steps:

[0098] Step 131: When the risk score is within the score range of the first risk level, call the bidding API interface of the advertising platform to reduce the current bid of the advertisement by a first preset percentage, and at the same time transfer a second preset percentage from the current plan budget to the backup plan to obtain the parameters after the first risk level is adjusted.

[0099] Step 131 focuses on "flexible intervention" to balance risk control and continuity of delivery for the first risk level (e.g., risk score 70≤score<80). When the risk score falls within this range, the system reduces the current bid by a first preset proportion (e.g., 20%) by calling the bid API interface of the advertising platform, directly reducing the cost per unit traffic, and slowing down the speed of ineffective consumption from the source; at the same time, a second preset proportion (e.g., 30%) is transferred from the current plan budget to the standby plan, which not only avoids continuous waste of budget in high-risk plans, but also retains part of the delivery opportunity through the standby plan to improve resource utilization efficiency. The "first risk level adjusted parameters" (e.g., adjusted bid, remaining budget, standby plan budget) generated finally will serve as the benchmark for subsequent protection mechanisms to ensure that the adjustment action is traceable and reusable. The core of this step is to maintain basic delivery under the premise of controllable risk through the combination strategy of "cost reduction + diversion", to avoid missing potential effective traffic due to excessive intervention.

[0100] Step 132, when the risk score is in the score range of the second risk level, according to the preset core keyword library, non-core keywords are selected from all keywords of the current delivery, and the keyword state interface of the advertising platform is called to suspend the non-core keywords in batches.

[0101] Step 132 corresponds to the adjustment of the second risk level (e.g., 80≤score<90), which aims to focus on core value traffic through "precise stop loss". The system pre-stores a "core keyword library" (e.g., keywords strongly related to the core conversion path of the business), and when the risk score falls within this range, non-core keywords (i.e., keywords not in the core library) are first selected from the keywords of the current delivery, and then the keyword state interface of the advertising platform is called to suspend these non-core keywords in batches. The logic of this operation is that core keywords usually contribute higher conversion value, and suspending non-core keywords can quickly cut off the consumption of low-value traffic while preserving the delivery of core traffic, maximizing the exposure opportunity of core business while controlling risk. Compared with full plan suspension, this "partial suspension" is more targeted and reduces the interference with the overall delivery rhythm.

[0102] Step 133, when the risk score is in the score range of the third risk level, the plan state interface of the advertising platform is called to suspend the current advertising plan.

[0103] Step 133 performs "strong stop loss" for high-risk scenarios of the third risk level (e.g., score ≥ 90) to avoid significant losses. When the risk score reaches this level, the system directly calls the plan status interface of the advertising platform to suspend the current advertising plan as a whole, instantly terminate all traffic acquisition and consumption, and fundamentally prevent the continuous waste of advertising fees during the high-risk period. This action is suitable for emergency situations such as sudden increase in conversion cost and severe abnormality of multiple indicators. By suspending the entire plan, it provides time for manual intervention to investigate problems and prevent further risk expansion. Compared with the adjustments of the previous two levels, this step has the highest intervention intensity and the highest priority, ensuring that the loss chain is cut off at the first time in extreme risk.

[0104] In some embodiments of the present application, as shown in Figure 5 Figure 5 is the implementation flowchart of starting the protection mechanism and the notification mechanism corresponding to the risk level in the embodiments of the present application, that is, step 14 can be implemented by the following steps:

[0105] Step 141, when the risk level is the first risk level, start a first duration cooling timer, lock the parameters adjusted by the first risk level during the first duration, write the parameters adjusted by the first risk level into a monitoring board, and do not actively alarm.

[0106] Step 141 is the protection mechanism for the first risk level (e.g., risk score 70 ≤ score < 80). It balances stability and flexibility through "short-term locking + silent recording". When this level is triggered, the system immediately starts a first duration (e.g., 30 minutes) cooling timer, during which it forces to maintain the parameters adjusted in step 131 (e.g., reduce the bid by 20%, transfer 30% of the budget to the standby plan), to prevent frequent adjustments due to data fluctuations. At the same time, write the adjusted parameters into the monitoring board for manual post-check, but do not actively trigger alarms (such as emails, SMS), to avoid excessive disturbance in low-risk scenarios. The logic of this design is that the first level of risk is relatively controllable, and short-term locking of adjusted parameters can allow the strategy to take full effect, while silent recording meets traceability and reduces manual intervention costs, ensuring that the system can still operate autonomously under light risk.

[0107] Step 142, when the risk level is the second risk level, start a second duration cooling timer, lock non-core keywords in a suspended state during the second duration, and send alarm information through an internal alarm interface, wherein the second duration is greater than the first duration.

[0108] ​Step 142 strengthens risk control through "medium-length lock + active alarm" for the protection mechanism of the second risk level (e.g., 80 ≤ score < 90). The system starts a second cooling timer (e.g., 1 hour, longer than the first length), during which the non-core keyword state suspended in step 132 remains unchanged, preventing risk rebound due to automatic recovery. At the same time, structured alarm information (including risk indicator details, suspended keyword list, etc.) is sent through an internal alarm interface (such as WeChat Enterprise or DingTalk) to prompt the operator to pay attention to potential problems. The combination of longer cooling time and active alarm not only gives the system enough time to digest the risk (e.g., waiting for the end of a competitor's promotion), but also ensures that manual intervention can be optimized in a timely manner. For example, if it is found that the core indicators have improved after a non-core keyword is suspended, the operator can choose to permanently exclude the keyword after the cooling period ends.

[0109] Step 143, when the risk level is a third risk level, a third length cooling timer is started, and when a user's recovery instruction is received, the current advertising plan is unpaused, and a voice notification interface is called to trigger a phone alarm and generate a diagnosis report, wherein the third length is greater than the second length, and the second length is greater than the first length.

[0110] Step 143, for the protection mechanism of the third risk level (e.g., score ≥ 90), through "indefinite lock + strong alarm + diagnosis support" to deal with extreme risk. The system starts a third cooling timer (e.g., "until manual recovery", longer than the second length), during which the advertising plan remains suspended, and only when the user's explicit recovery instruction is received will the suspension be lifted. At the same time, the system calls a voice notification interface to trigger a phone alarm (directly contacting the responsible person) and automatically generates a diagnosis report containing abnormal indicator trends, adjustment history, and possible causes. This combination of "indefinite lock + strong alarm" ensures that there is no secondary loss due to automatic recovery of the system in high-risk scenarios, and the diagnosis report provides data support for manual decision-making. For example, when the conversion cost of a plan suddenly rises to 200% of the target price, the system suspends the plan and generates a report, prompting "possible due to competitor malicious clicks", helping the operator to quickly locate the problem.

[0111] Steps 141 to 143 achieve precise matching of the protection mechanism and the risk level through the three gradient designs of "increasing cooling time + increasing alarm intensity + increasing lock range": short-term lock is used to maintain stability in low-risk scenarios, active alarm is used to promote manual optimization in medium-risk scenarios, and indefinite lock + strong intervention is used to prevent major losses in high-risk scenarios, forming a complete risk closed-loop management system.

[0112] In some embodiments of the present application, the method can further include the following steps:

[0113] Step 16, when the risk score during the cooling period corresponding to each risk level falls within the score range corresponding to a lower risk level, the adjustment action corresponding to the lowered risk level is performed.

[0114] Step 16 is an adaptive response mechanism for dynamic changes in risk during cooling, aiming to solve the problem of how to adjust the strategy after the risk level is lowered, and ensure that the protection mechanism always matches the real-time risk state.

[0115] Specifically, when the system is within the cooling period of a certain risk level (such as the 30-minute cooling period of the first risk level, the 1-hour cooling period of the second risk level), data will be continuously collected and risk scores will be recalculated at a set frequency. If the recalculated risk score falls within the score range corresponding to a lower risk level (such as the risk score of the original second risk level falling from 85 to 75, falling within the range of the first risk level), the system will automatically terminate the current cooling state and perform the corresponding adjustment action according to the rules of the lowered risk level (such as switching from "suspend non-core keywords" to "reduce bid by 20% + transfer 30% budget to backup plan").

[0116] For example: an ad plan triggered the second risk level (cooling for 1 hour, suspending non-core keywords) due to a risk score of 85, and after 30 minutes of cooling, real-time data showed that the risk score had fallen to 75 (falling within the range of the first risk level). At this time, the system immediately terminated the remaining 30 minutes of cooling, performed the adjustment action of the first risk level (reducing the bid by 20% and transferring the budget), and started the 30-minute cooling corresponding to the first risk level.

[0117] This step breaks the limitation of "strategy solidification during the cooling period" and realizes real-time adaptation of "risk downgrade → strategy downgrade" by dynamically tracking risk changes, avoiding waste of traffic caused by maintaining high-intensity intervention when the risk has already decreased (such as continuing to suspend non-core keywords when it should have been downgraded), ensuring the continuity of risk control, and further improving the system's adaptive ability to complex deployment scenarios.

[0118] Referring to Figure 6 , Figure 6 is a structural schematic diagram of an advertisement deployment control device provided by an embodiment of the present application, the device comprising:

[0119] The collection module 21 is configured to collect input data at a set frequency, and the input data includes ad platform data, internal business data, and competitor data.

[0120] The analysis and judgment module 22 is configured to analyze and calculate the input data to obtain a risk score and a corresponding risk level, and to determine whether a risk condition is met according to the risk level.

[0121] The adjusting module 23 is configured to perform a corresponding adjusting action according to the risk level when it is determined that the risk condition is met.

[0122] The protection and notification module 24 is configured to start a protection mechanism and a notification mechanism corresponding to the risk level.

[0123] In some embodiments of the present application, the device can further comprise a maintaining module 25 configured to maintain the current advertising delivery setting when it is determined that the risk condition is not met, i.e., in a normal state.

[0124] In some embodiments of the present application, the device can further comprise a loading module 21a configured to automatically load a basic parameter library, a risk judgment rule set and a risk level strategy matrix, wherein the basic parameter library comprises at least two standard indicators, including a conversion cost target price and a conversion cost threshold, an LTV expected value, a normal fluctuation range of a click rate and a large promotion identification feature of a competitor; the risk judgment rule set comprises a weight and a superposition logic when each standard indicator is abnormal; and the risk level strategy matrix stores a risk score corresponding to each risk level, a corresponding adjusting action and a protection mechanism.

[0125] In some embodiments of the present application, the analysis and judgment module 22 can comprise:

[0126] A calculating unit 221 configured to calculate a plurality of risk indicators according to the input data;

[0127] A comparing unit 222 configured to compare each risk indicator with a corresponding standard indicator to obtain an abnormal indicator;

[0128] A collaborative calculating unit 223 configured to collaboratively calculate the abnormal indicators to obtain a risk score and a corresponding risk level;

[0129] A first state unit 224 configured to determine a normal state when the risk score is within a range corresponding to a normal risk level and there is no abnormal indicator;

[0130] A second state unit 225 configured to determine that a risk condition is met when the risk score is not within the range corresponding to the normal risk level, or the risk score is within the range corresponding to the normal risk level but a same risk indicator is continuously abnormal for a preset number of times.

[0131] In some embodiments of the present application, the collaborative calculating unit 223 can comprise:

[0132] An extracting sub-unit 2231 configured to extract all abnormal indicators;

[0133] A calculating sub-unit 2232 configured to calculate a risk score according to a weight and a superposition logic of each abnormal indicator in the risk judgment rule set;

[0134] The comparison subunit 2233 is configured to compare the risk score with the risk level strategy matrix to obtain a risk level.

[0135] In some embodiments of the present application, the adjustment module 23 can include:

[0136] The first adjustment unit 231 is configured to, when the risk score is within a score range of a first risk level, call an API interface of an advertising platform to reduce a current bid of an advertisement by a first preset proportion, and transfer a second preset proportion from a current plan budget to a backup plan to obtain an adjusted parameter of the first risk level.

[0137] The second adjustment unit 232 is configured to, when the risk score is within a score range of a second risk level, filter non-core keywords from all keywords of a current campaign according to a preset core keyword library, and call a keyword state interface of the advertising platform to suspend the non-core keywords in batches.

[0138] The third adjustment unit 233 is configured to, when the risk score is within a score range of a third risk level, call a plan state interface of the advertising platform to suspend a current advertising plan.

[0139] In some embodiments of the present application, the protection and notification module 24 can include:

[0140] The first protection and notification unit 241 is configured to, when the risk level is the first risk level, start a first duration cooling timer, lock the adjusted parameter of the first risk level during the first duration, write the adjusted parameter of the first risk level into a monitoring dashboard, and send no active alarm.

[0141] The second protection and notification unit 242 is configured to, when the risk level is the second risk level, start a second duration cooling timer, lock the non-core keywords in a suspended state during the second duration, and send an alarm information through an internal alarm interface, wherein the second duration is greater than the first duration.

[0142] The third protection and notification unit 243 is configured to, when the risk level is the third risk level, start a third duration cooling timer, and when a user's resuming instruction is received, release the suspension of the current advertising plan, and call a voice notification interface to trigger a phone alarm and generate a diagnosis report, wherein the third duration is greater than the second duration, and the second duration is greater than the first duration.

[0143] In some other embodiments of the present application, the adjustment module 23 is further configured to, when the risk score decreases to a score range corresponding to a lower risk level during a cooling duration corresponding to each risk level, perform an adjustment action corresponding to the decreased risk level.

[0144] It should be noted that the advertisement putting control device in the embodiments of the present application and the advertisement putting control method in the above embodiments belong to the same inventive concept, and the technical details not described in detail in the device can be seen from the related description of the method above, and will not be described here.

[0145] In addition, the embodiments of the present application also provide a computer readable storage medium, the computer readable storage medium stores a computer program, wherein the computer program is set to execute the method described above when running.

[0146] Figure 7 is a structural schematic diagram of an electronic device 10 provided by the embodiments of the present application. The electronic device is intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular telephones, smart phones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are merely examples and are not intended to limit the implementations of the present application described and / or claimed herein.

[0147] As shown in Figure 7 The electronic device 10 includes at least one processor 11, and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., which is communicatively connected to the at least one processor 11, wherein the memory stores a computer program executable by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0148] A plurality of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunications networks.

[0149] The processor 11 can be various general and / or special purpose processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, and the like. The processor 11 performs various methods and processes described above, such as the idle detection method.

[0150] In some embodiments, the idle detection method can be implemented as a computer program tangibly embodied in a computer readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded onto the RAM 13 and executed by the processor 11, one or more steps of the idle detection method described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to perform the idle detection method by any other suitable means, such as by means of firmware.

[0151] Various implementations of the systems and techniques described above can be realized in digital electronic circuitry, integrated circuitry, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on a chip (SOC), a programmable logic device (PLD), a computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0152] Computer programs used to implement the methods of the present application can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the computer program, when executed by the processor, implements the functions / acts specified in the flowcharts and / or block diagrams. The computer program can be executed entirely on a machine, partially on a machine, partially on a machine as a stand-alone software package, partially on a machine and partially on a remote machine or entirely on a remote machine or server.

[0153] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. A computer-readable storage medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of a machine-readable storage medium will include one or more lines of a program of instructions in a transitory signal, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0154] To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0155] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0156] The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system, to solve the defects of large management difficulty and weak business scalability in traditional physical host and VPS service.

[0157] It should be understood that the order of the steps shown above can be re-executed, steps can be added or deleted. For example, the steps described in the present application can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions of the present application can be achieved, which are not limited herein.

[0158] The above detailed description does not constitute a limitation on the scope of protection of the present application. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. An advertising placement control method, characterized in that, The method includes: Input data is collected at a set frequency, including advertising platform data, internal business data, and competitor data. The input data is analyzed and calculated to obtain a risk score and a corresponding risk level, and the risk level is used to determine whether the risk conditions are met. When it is determined that the risk conditions are met, corresponding adjustment actions are performed according to the risk level. Activate the protection and notification mechanisms corresponding to the risk level.

2. According to the method of claim 1, when it is determined that the risk conditions are not met, i.e., the normal state, the current advertising settings are maintained.

3. The method according to claim 1, characterized in that, Before collecting input data at a set frequency, the method further includes: Automatically load the basic parameter library, risk assessment rule set, and risk level strategy matrix. The basic parameter library includes at least two of the following standard indicators: target price for conversion cost and threshold for conversion cost, expected value of LTV, normal fluctuation range of click-through rate, and characteristics for identifying competitors' major promotional events. The risk assessment rule set includes the weights and overlay logic for abnormalities in each standard indicator; The risk level strategy matrix stores the risk score, corresponding adjustment actions, and protection mechanisms for each risk level.

4. The method according to claim 3, characterized in that, The input data is analyzed and calculated to obtain a risk score and corresponding risk level, and the risk level is used to determine whether the risk conditions are met, including: Calculate multiple risk indicators based on the input data; Each risk indicator is compared with its corresponding standard indicator to identify abnormal indicators. The abnormal indicators are calculated collaboratively to obtain risk scores and corresponding risk levels; When the risk score is within the range corresponding to the normal risk level and there are no abnormal indicators, it is judged to be in a normal state; When the risk score is not within the range corresponding to the normal risk level, or when the risk score is within the range corresponding to the normal risk level but the same risk indicator is consecutively preset as an abnormal indicator, it is determined that the risk condition is met.

5. The method according to claim 4, characterized in that, The abnormal indicators are collaboratively calculated to obtain risk scores and corresponding risk levels, including: Extract all abnormal indicators; The risk score is calculated based on the weights and superposition logic of each abnormal indicator in the risk judgment rule set. The risk score is compared with the risk level strategy matrix to obtain the risk level.

6. The method according to claim 4, characterized in that, When the risk conditions are determined to be met, corresponding adjustment actions are performed according to the risk level, including: When the risk score is within the range of the first risk level, the bidding API interface of the advertising platform is called to reduce the current bid of the advertisement by a first preset percentage, and at the same time, a second preset percentage is transferred from the current plan budget to the backup plan to obtain the parameters after the first risk level is adjusted. When the risk score is within the range of the second risk level, non-core keywords are selected from all currently advertised keywords based on the preset core keyword library, and the keyword status interface of the advertising platform is called to pause the non-core keywords in batches. When the risk score falls within the range of the third risk level, the advertising platform's plan status interface is invoked to pause the current advertising plan.

7. The method according to claim 6, characterized in that, Activate the protection and notification mechanisms corresponding to the risk level, including: When the risk level is the first risk level, a first-duration cooldown timer is started, and the parameters adjusted for the first risk level are locked during the first duration. The parameters adjusted for the first risk level are written to the monitoring dashboard, and no active alarm is triggered. When the risk level is the second risk level, a second-duration cooldown timer is started. During the second duration, non-core keywords are locked in a paused state, and alarm information is sent through the internal alarm interface. The second duration is longer than the first duration. When the risk level is the third risk level, a third-duration cooldown timer is started until a user's instruction to resume advertising is received. Then, the current advertising plan is unsuspended, and a voice notification interface is called to trigger a telephone alarm and generate a diagnostic report. The third duration is longer than the second duration, and the second duration is longer than the first duration.

8. The method according to claim 7, characterized in that, If the risk score drops to the range of the lower risk level during the cooldown period corresponding to each risk level, then the adjustment action corresponding to the lower risk level will be executed.

9. An advertising delivery control device, characterized in that, The device includes: The data acquisition module is used to collect input data at a set frequency. The input data includes advertising platform data, internal business data, and competitor data. The analysis and judgment module is used to analyze and calculate the input data to obtain the risk score and the corresponding risk level, and to determine whether the risk conditions are met based on the risk level. The adjustment module is used to perform corresponding adjustment actions based on the risk level when it is determined that the risk conditions are met. The protection and notification module is used to activate the protection and notification mechanisms corresponding to the risk level.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, wherein the computer program is configured to execute the method of any one of claims 1 to 8 when it is run.

Images