Information delivery optimization method and system, electronic device, and storage medium
By analyzing historical data and optimization algorithms, and establishing objective functions, the inefficiency and mismatch problems caused by empirical solutions in the information delivery method are solved, and the best results of information delivery are achieved.
Patent Information
- Application Number
- CN201911051022.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2019-10-31
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2039-10-31
AI Technical Summary
The existing information delivery methods are formulated based on the experience of decision makers, resulting in low efficiency of information delivery and mismatch between the plan and the actual delivery results, and lack of theoretical basis.
By collecting historical delivery data, analyzing the correlation between historical information delivery parameters and actual delivery result parameters, establishing an objective function and using an optimization algorithm to solve it, and optimizing information delivery parameters.
It improves the efficiency of information delivery and the matching degree between the delivery parameters and the actual delivery results, provides quantifiable objective functions and constraints, and realizes the global optimal information delivery parameters.
Smart Images

Figure CN112749824B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of computers, and in particular relates to an information delivery optimization method and system, electronic equipment, and storage medium. Background Art
[0002] Earlier, some information platforms made profits by selling information positions, that is, the first information under category A required a fee of C yuan to place an information B hours later, and the input-output ratio of the sale was fixed. Later, information platforms mainly used competitive bidding when selling information positions. Because it is a real-time bidding, the information placed in each position will change at every moment, and the information publisher has no way to discover the change in the placement position in the first place and modify the bidding price to ensure the placement result. In addition, there are many existing information platforms, and the bidding rules of each information platform are different. In order to obtain better placement results, the information publisher needs to place information on more information platforms, but for the information publisher, it is difficult to coordinate and manage each information platform.
[0003] In order to adapt to bidding needs, the operators of information delivery need to spend a lot of time maintaining the delivery information. The key to maintaining the delivery information is to provide appropriate information delivery plans.
[0004] The existing information delivery methods are mainly formulated by operators, that is, decision makers, based on their own experience. However, different decision makers have different experiences and familiarity with the information platform, and may come to different conclusions, thus giving different information delivery plans.
[0005] The existing method of information delivery is to provide an empirical plan based on the decision maker's experience and familiarity with the information platform. This artificially formulated plan has many uncertainties and lacks theoretical basis, which leads to low efficiency of information delivery and mismatch between the plan and actual delivery results. Summary of the invention
[0006] The technical problem to be solved by the embodiments of the present invention is to overcome the defects in the prior art that the information delivery plan is derived by relying on the experience of decision makers and their familiarity with the information platform, resulting in low information delivery efficiency and mismatch between the plan and the actual delivery results, and to provide an information delivery optimization method and system, electronic device, and storage medium.
[0007] The embodiment of the present invention solves the above technical problems through the following technical solutions:
[0008] An embodiment of the present invention provides an information delivery optimization method, including:
[0009] Collecting historical delivery data, the historical delivery data includes historical information delivery parameters and actual delivery result parameters obtained by delivering information according to the historical information delivery parameters;
[0010] Obtaining delivery demand parameters, and establishing an objective function according to the delivery demand parameters;
[0011] Analyze the correlation between the historical information delivery parameters and the actual delivery result parameters according to the historical delivery data, and use the analysis results as optimization constraints;
[0012] Under the constraints, an optimization algorithm is used to solve the optimal information delivery parameters that make the objective function reach the delivery demand parameters.
[0013] Preferably, the actual delivery result parameter includes an actual value of at least one indicator used to measure the delivery result, and the delivery demand parameter includes a target value of the at least one indicator;
[0014] The step of establishing an objective function according to the delivery demand parameters comprises:
[0015] For each target value of the indicator, a sub-objective function is established respectively;
[0016] Combine each sub-objective function to establish the overall objective function;
[0017] The steps of analyzing the correlation between the historical information delivery parameters and the actual delivery result parameters according to the historical delivery data and using the analysis results as optimization constraints include:
[0018] According to the mapping relationship between the historical information delivery parameter and the actual value of each of the indicators, respectively analyzing the relationship function between the historical information delivery parameter and each of the indicators;
[0019] The relationship function is used as a constraint condition for optimization. Preferably, a weight method and a dimensionless method are used to combine each sub-objective function to establish an overall objective function;
[0020] And / or, using regression analysis method to analyze the relationship function between the historical information delivery parameter and each of the indicators.
[0021] Preferably, the information is delivered to at least one information platform, the delivery requirement parameters are set for each information platform or part of all information platforms, and the optimal information delivery parameters include information delivery parameters for each information platform or part of all information platforms;
[0022] Or, an optimization cycle is divided into at least one time period, the delivery demand parameter is set for each time period or part of the time period in an optimization cycle, and the optimal information delivery parameter includes the information delivery parameter for each time period or part of the time period in an optimization cycle;
[0023] Or, the information is delivered to at least one information platform, an optimization cycle is divided into at least one time period, the delivery demand parameters are set for each information platform or part of the information platforms among all the information platforms in each time period or part of the time period in an optimization cycle, and the optimal information delivery parameters include the information delivery parameters of each information platform or part of the information platforms among all the information platforms in each time period or part of the time period in an optimization cycle.
[0024] Preferably, the information delivery parameters include a bid for delivering the information and an information delivery position associated with the bid on the information platform;
[0025] And / or, the constraint conditions also include: at least one of: a limit on the platform's investment budget, an upper and lower limit on the information platform's delivery budget, and an upper and lower limit on the information platform's delivery bid.
[0026] Preferably, the information is delivered to at least one information platform, an optimization cycle is divided into at least one time period, the delivery demand parameter is set for each information platform among all information platforms in each time period in an optimization cycle, the optimal information delivery parameter includes the information delivery parameter of each information platform among all information platforms in each time period in an optimization cycle, and the information delivery parameter includes the bid for delivering the information;
[0027] Under the constraints, the step of using a stepwise optimization algorithm to solve the optimal information delivery parameters that make the objective function reach the delivery demand parameters includes:
[0028] Set the bid, number of iterations N and iteration step size corresponding to the initial solution of each information platform in all information platforms;
[0029] Initialize the number of iterations iter = 1;
[0030] Initialize calculation period t=1;
[0031] Fix the bids of each information platform in the t-1th period and the t+1th period and adjust the bids of the tth period according to the iteration step;
[0032] According to the delivery demand parameters of each information platform in the tth period, select the optimal bid of each information platform in the tth period;
[0033] Determine whether t < T - 1 holds. If it holds, calculate t = t + 1 and jump to the step of fixing the bids for the (t - 1)-th period and the (t + 1)-th period of each information platform and adjusting the bid for the t-th period according to the iteration step. If it does not hold, then:
[0034] Determine whether iter < N is met. If it is met, calculate iter = iter + 1 and jump to the step of initializing the calculation period t = 1. If it does not hold, then:
[0035] Calculate the placement demand parameters within the entire optimization period and take them as the optimal values;
[0036] Determine whether the accuracy of the optimal value meets the accuracy requirement. If so, the optimization ends, and the bids of each information platform in the optimal value state are the optimal information placement parameters. If not, then take the bids of each information platform in the optimal value state as the initial solution, and then jump back to the step of initializing the iteration number iter = 1.
[0037] An embodiment of the present invention further provides an information placement optimization system, including:
[0038] A data collection module for collecting historical placement data, where the historical placement data includes historical information placement parameters and actual placement result parameters obtained by placing information according to the historical information placement parameters;
[0039] A target function module for obtaining placement demand parameters and establishing a target function according to the placement demand parameters;
[0040] A constraint condition module for analyzing the correlation between the historical information placement parameters and the actual placement result parameters according to the historical placement data, and taking the analysis result as the constraint condition for optimization;
[0041] An optimization solving module for solving the optimal information placement parameters that make the target function reach the placement demand parameters under the constraint conditions..
[0042] Preferably, the information is placed on at least one information platform. An optimization period is divided into at least one period. The placement demand parameters are set for each information platform among all information platforms in each period of an optimization period. The optimal information placement parameters include the information placement parameters of each information platform among all information platforms in each period of an optimization period. The information placement parameters include the bids for placing the information;
[0043] When the optimization solving module uses a step-by-step optimization algorithm to solve the optimal information placement parameters that make the target function reach the placement demand parameters under the constraint conditions, it is specifically used for:
[0044] Set the bid, the number of iterations N, and the iteration step size corresponding to the initial solution of each information platform in all information platforms;
[0045] Initialize the iteration number iter = 1;
[0046] Initialize the calculation period t = 1;
[0047] Fix the bids of each information platform in the (t - 1)-th period and the (t + 1)-th period, and adjust the bid of the t-th period according to the iteration step size;
[0048] According to the placement demand parameters of each information platform in the t-th period, select the optimal bid of each information platform in the t-th period;
[0049] Judge whether t < T - 1 holds. If it holds, calculate t = t + 1 and jump to the step of fixing the bids of each information platform in the (t - 1)-th period and the (t + 1)-th period and adjusting the bid of the t-th period according to the iteration step size. If it does not hold, then:
[0050] Judge whether iter < N is satisfied. If it is satisfied, calculate iter = iter + 1 and jump to the step of initializing the calculation period t = 1. If it does not hold, then:
[0051] Calculate the placement demand parameters within the entire optimization period and take them as the optimal values;
[0052] Judge whether the accuracy of the optimal value meets the accuracy requirement. If so, the optimization ends, and the bids of each information platform in the optimal value state are the optimal information placement parameters. If not, take the bids of each information platform in the optimal value state as the initial solution, and then jump back to the step of initializing the iteration number iter = 1.
[0053] An embodiment of the present invention further provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the above-mentioned information placement optimization method is implemented.
[0054] An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the steps of the above-mentioned information placement optimization method are implemented.
[0055] On the basis of conforming to the common knowledge in the art, the above preferred conditions can be combined arbitrarily to obtain various preferred embodiments of the present invention.
[0056] The positive progressive effect of the embodiment of the present invention is that the embodiment of the present invention uses optimization theory to model the information delivery problem, provides a quantifiable objective function, takes the relationship formula between information delivery parameters and delivery result parameters as constraints, uses an optimization algorithm for optimization, and finally obtains the globally optimal information delivery parameters, so that the information delivery party can obtain the best delivery result under the constraints, improves the information delivery efficiency and the matching degree between the delivery parameters and the actual delivery results. BRIEF DESCRIPTION OF THE DRAWINGS
[0057] Figure 1 This is a flow chart of an information delivery optimization method according to Embodiment 1 of the present invention.
[0058] Figure 2 This is a specific flow chart of step 11 of an information delivery optimization method according to embodiment 1 of the present invention.
[0059] Figure 3 This is a specific flow chart of step 12 of an information delivery optimization method according to embodiment 1 of the present invention.
[0060] Figure 4 This is a flowchart of an App advertising information delivery optimization method according to Example 2 of the present invention.
[0061] Figure 5 This is a specific flow chart of step 23 of an App advertising information delivery optimization method according to embodiment 2 of the present invention.
[0062] Figure 6 This is a schematic diagram of an App advertising information delivery optimization system according to Example 3 of the present invention.
[0063] Figure 7 This is a schematic diagram of the structure of an electronic device according to Embodiment 4 of the present invention. DETAILED DESCRIPTION
[0064] The present invention is further described below by way of examples, but the present invention is not limited to the scope of the examples.
[0065] Example 1
[0066] Figure 1 A method for optimizing information delivery in this embodiment is shown. The method for optimizing information delivery includes the following steps:
[0067] Step 10: Collect historical delivery data, which includes historical information delivery parameters and actual delivery result parameters obtained by delivering information according to the historical information delivery parameters. The historical delivery data may have a time limit, for example, including information delivery parameters counted in a preset time period in the past and delivery result parameters actually achieved by the delivery information in the time period. The historical delivery data may be counted by the information delivery party itself, or provided by the information delivery information platform, or provided by a third-party service agency, which is not limited in this embodiment.
[0068] Step 11: Obtain delivery demand parameters, and establish an objective function based on the delivery demand parameters. The delivery demand parameters can be determined by the information delivery party according to the type of information and the delivery result to be achieved.
[0069] Step 12: Analyze the correlation between the historical information delivery parameters and the actual delivery result parameters based on the historical delivery data, and use the analysis results as optimization constraints.
[0070] Step 13: Under the constraints, an optimization algorithm is used to solve the optimal information delivery parameters that make the objective function reach the delivery demand parameters. The optimal information delivery parameters can generally be considered as information delivery parameters or a combination of parameters that achieve the best delivery results under the constraints.
[0071] In this embodiment, the actual delivery result parameter may include the actual value of at least one indicator used to measure the delivery result, and the delivery demand parameter may include the target value of the at least one indicator. The specific indicator selection may be determined by the information delivery party according to the type of information and the delivery result to be achieved. For example, for news and information information, the indicator for measuring the delivery result may be set to, but not limited to, the number of readings, clicks, forwardings, and collections of the information; for advertising information, the indicator for measuring the delivery result may be set to, but not limited to, the purchase volume, exposure volume, and purchase exposure volume of the product in the advertisement; for audio / video information, the indicator for measuring the delivery result may be set to, but not limited to, the number of clicks, the playback time of the audio / video, the number of forwardings, the number of collections, and the number of downloads.
[0072] like Figure 2 As shown, step 11 may specifically include the following steps:
[0073] Step 111: For each target value of the indicator, a sub-objective function is established respectively.
[0074] Step 112: Combine each sub-objective function to establish the overall objective function.
[0075] The above steps provide a quantifiable objective function for the optimization process of information delivery. Step 112 can directly merge the sub-objective functions to obtain the overall objective function, but considering that the magnitudes of the sub-objective functions may vary greatly and the importance of the indicators may vary, if the sub-objective functions are directly merged, a large error may occur. Therefore, step 112 can specifically use the weight method and the dimensionless method to combine each sub-objective function to establish the overall objective function.
[0076] like Figure 3 As shown, step 12 may specifically include the following steps:
[0077] Step 121: According to the mapping relationship between the historical information delivery parameter and the actual value of each of the indicators, respectively analyze the relationship function between the historical information delivery parameter and each of the indicators.
[0078] Step 122: Using the relationship function as a constraint condition for optimization.
[0079] When executing step 121, different algorithms can be used to model the relationship between the historical information delivery parameters and the actual values of each of the indicators, and then the relationship function can be obtained. Considering that this relationship characteristic may be very complex, in order to simplify the calculation process and reduce the difficulty of modeling, step 121 preferably uses regression analysis to analyze the relationship function between the historical information delivery parameters and each of the indicators, and further obtains the correlation between the information delivery parameters and the actual delivery results achieved. Regression analysis is a statistical analysis method that determines the quantitative relationship between two or more variables that are interdependent, and is widely used. Regression analysis is divided into univariate regression and multivariate regression analysis according to the number of variables involved; it can be divided into simple regression analysis and multiple regression analysis according to the number of dependent variables; it can be divided into linear regression analysis and nonlinear regression analysis according to the type of relationship between independent variables and dependent variables. If only one independent variable and one dependent variable are included in the regression analysis, and the relationship between the two can be approximated by a straight line, this regression analysis is called univariate linear regression analysis.
[0080] In this embodiment, the information delivery parameters may include the bid for delivering the information and the information delivery position associated with the bid on the information platform. The association between the bid and the information delivery position is usually determined by the information platform itself, and different information delivery positions may bring different delivery results. Since the relationship between the bid and the information delivery position may be fixed, what is actually analyzed in step 12 may be the correlation between the bid for delivering the information and the actual delivery result parameters achieved, and what is actually analyzed in step 121 may be the relationship function between the bid for delivering the information and each of the indicators.
[0081] In addition to the correlation between the information delivery parameters and the actual delivery result parameters achieved as optimization constraints, in actual applications, the constraints may also include but are not limited to: at least one of the platform investment budget cost limit, the information platform's delivery budget upper and lower limits, and the information platform's delivery bid upper and lower limits. Among them, the platform investment budget cost limit, the specific value can be determined by the information delivery party, which can represent the total budget cost of all information platforms within an optimization cycle, and the sum of related expenses of all information platforms brought about by the optimized optimal information delivery parameters should not exceed the total budget cost; the upper and lower limits of the information platform's delivery budget, the specific values can be determined by the information delivery party, which can represent the maximum and minimum values of the delivery budget cost of a certain information platform within an optimization cycle, and the sum of related expenses of the information platform brought about by the optimized optimal information delivery parameters should be between the maximum and minimum values of the delivery budget cost of the information platform; the upper and lower limits of the information platform's delivery bid, the specific values can be determined by the information platform, which can represent the maximum and minimum values allowed for the information platform to bid, and the bid of the information platform given by the optimized optimal information delivery parameters should be between the maximum and minimum values allowed for the information platform to bid.
[0082] In step 13, different optimization algorithms may be selected according to the correlation between the information delivery parameters and the actual delivery results achieved, the specific values of the information delivery parameters, and the specific indicators for measuring the delivery results. In this embodiment, the optimization algorithm used in step 13 may include but is not limited to any one of a stepwise optimization algorithm, a linear programming algorithm, and a dynamic programming algorithm.
[0083] Considering that in actual applications, the information delivery party may deliver information on more than one information platform in order to obtain better delivery results, in order to better coordinate and plan the delivery strategies of all information platforms, the information delivery optimization method of this embodiment may also:
[0084] In the case where the information is delivered to at least one information platform, the delivery demand parameters are set for each of all information platforms, and the optimal information delivery parameters include the information delivery parameters of each of all information platforms. Of course, the delivery demand parameters are set only for some of all information platforms, and the optimal information delivery parameters include the information delivery parameters of some of all information platforms, and the information delivery optimization method of this embodiment is also applicable.
[0085] Considering that the bidding price of the information platform in each period changes in real time in actual application, the information delivery optimization method of this embodiment can also:
[0086] In the case where an optimization cycle is divided into at least one time period, the delivery demand parameter is set for each time period in an optimization cycle, and the optimal information delivery parameter includes the information delivery parameter for each time period in an optimization cycle. Of course, the delivery demand parameter is set only for part of the time periods in an optimization cycle, and the optimal information delivery parameter includes the information delivery parameter for part of the time periods in an optimization cycle, and the information delivery optimization method of this embodiment is also applicable.
[0087] Of course, considering that the information delivery party may deliver information to more than one information platform in order to obtain better delivery results, in order to better coordinate the delivery strategies of all information platforms, and the bidding of each information platform in each time period changes in real time, the information delivery optimization method of this embodiment can also:
[0088] In the case where the information is delivered to at least one information platform and an optimization cycle is divided into at least one time period, the delivery demand parameters are set for each information platform in all information platforms in each time period or part of a time period in an optimization cycle, and the optimal information delivery parameters include the information delivery parameters of each information platform in all information platforms in each time period or part of a time period in an optimization cycle. Of course, the information delivery optimization method of this embodiment is also applicable if the delivery demand parameters are only set for some information platforms in each time period or part of a time period in an optimization cycle, and the optimal information delivery parameters include the information delivery parameters of some information platforms in each time period or part of a time period in an optimization cycle.
[0089] Example 2
[0090] The information delivery optimization method in Example 1 can be applied to a variety of information delivery scenarios, such as advertising information delivery scenarios: the advertising information delivery party delivers advertising information to some information platforms to achieve the purpose of promoting and advertising products, and the indicators used to measure the delivery results may include but are not limited to: the purchase volume of the advertising product, the exposure volume of the advertising product, and at least one of the purchase exposure volume of the advertising product. Of course, the application scenarios are not limited to this, and this embodiment will not cite them one by one.
[0091] In order to better explain the present invention, this embodiment specifically provides a specific method for optimizing App advertising information delivery by using the information delivery optimization method of the present invention.
[0092] App brands need to place App ads on information platforms to promote Apps, expand brand publicity, and gain more App users. The information platforms that receive the ads can include App markets, as well as other widely used Internet websites or applications. The original intention of the App market was to allow smartphone users to complete more work and entertainment on their phones. Users can download various software in the App market and fully experience the fun brought by the mobile Internet. The App market uses bidding ranking, that is, the comprehensive index of the App is calculated according to the algorithm (unit * weight value), and then the ranking is determined according to the index. Whenever a user browses App advertising information in the App market, the App brand needs to pay the App market. This payment value is called the cost per exposure, or CPE (Cost Per Exposure). When the App is downloaded by clicking on the App advertising information, the App brand also needs to pay the App market. This payment value is called the cost per download, or CPD (Cost Per Download). Because it is real-time bidding, the App advertising information at each time and location will change, and the App brand needs to modify the bidding price in time to ensure the delivery result. There are multiple operating systems or brand application markets. App brands need to launch products in multiple application markets at the same time and coordinate the bidding of all application markets.
[0093] Figure 4 An optimization method for App advertising information delivery of the present embodiment is shown. It comprehensively considers the download volume, exposure volume, and download exposure volume of the App as indicators for measuring the delivery results of App advertising information, proposes a quantifiable objective function, and uses the relationship formula between the information delivery parameters (mainly including the bid for delivering App advertising information) and related indicators obtained by the regression analysis method as constraints. A step-by-step optimization algorithm is introduced. Within an optimization cycle, a global optimization calculation is performed on the bid parameter values of all information platforms, that is, optimization is performed from two dimensions: all information platforms and an entire optimization cycle. Finally, a bid reference value for each information platform in each time period is given, so that the App brand's budget can obtain the maximum benefit under the existing constraints. The specific steps are as follows:
[0094] Step 20: Collect historical delivery data.
[0095] Step 21: Establishing the objective function. This embodiment provides three optimization objectives: ① maximizing the download volume D; ② maximizing the exposure volume E; ③ download exposure rate maximum.
[0096] Goal ① has the largest download volume, and the mathematical expression is:
[0097]
[0098] Where D is the download volume of all information platforms in one optimization cycle; t is the optimization period sequence number; D i,t is the download volume of the tth period of the i-th information platform; T is the total number of periods in an optimization cycle, and Γ is the set of information platforms.
[0099] Target ② has the largest exposure, and the mathematical expression is:
[0100]
[0101] Where, E is the exposure of all information platforms in one optimization cycle; t is the optimization period number; E i,t is the exposure of the ith information platform in the tth period.
[0102] Target ③ has the largest download exposure rate, and the mathematical expression is:
[0103]
[0104] Where ξ is the download exposure rate of the App during the entire optimization cycle.
[0105] The magnitudes of the objective functions D, E, and ξ are very different, so the three objective functions cannot be directly combined. The objectives must be dimensionless, and there is a competitive relationship between the three objectives, which makes the problem a multi-objective optimization problem. The weight method and the dimensionless method are combined to establish the total objective function of the bid parameter value, as shown in the following formula:
[0106] maxF=α D (D / D max )+α E (E / E max )+α ξ ξ
[0107] Where D max is the theoretical maximum number of downloads of the information platform app, which is equal to the budget / the unit price of each download; E max is the theoretical maximum exposure of the information platform App, which is equal to budget / price per exposure; α D , α E and α ξ The weight coefficients for the three optimization goals are between 0 and 1, and their values are determined by the App brand.
[0108] Step 22: Calculate constraints. This embodiment provides five constraints: ① platform investment budget limit; ② bid and download relationship constraint; ③ bid and exposure relationship constraint; ④ budget upper and lower limits of each information platform; ⑤ bid upper and lower limits of each information platform.
[0109] Regarding ① Platform investment budget limitation
[0110]
[0111]
[0112]
[0113] Where M is the total budget cost of all market information platforms during the entire optimization cycle, which is a fixed value; M i is the budget cost of the i-th information platform; P i,t is the bid of the tth period of the i-th information platform; C i,t is the amount of money spent on the i-th information platform in the t-th period, which is related to the fees that need to be paid to the information platform for downloading and exposure; P i,t to C i,t The functional relationship of is obtained by regression analysis based on the historical bidding and spending relationship of the ith information platform in the tth period.
[0114] Assume that the linear formula between bid and cost is C i,t =α i,t P i,t +β i,t , and use the least squares formula to get the fitting parameters:
[0115] C i,t =α i,t P i,t +β i,t
[0116]
[0117]
[0118] In the formula, α i,t , β i,t Cost C for the tth period of the i-th information platform i,t and bid P i,t The fitting parameter between i,j,t is the bid price of the tth period on the i-th information platform j days ago, C i,j,t is the expenditure of the tth period j days ago on the ith information platform, and n is the maximum number of days of historical data.
[0119] Regarding ② the relationship between bid and download volume
[0120]
[0121] In the formula, P i,t To D i,tThe functional relationship is obtained by fitting the historical bid and download volume relationship of the ith information platform in the tth period using the univariate linear regression analysis method. Please refer to the fitting formula between bid and cost in ①.
[0122] Regarding ③ the relationship between bid and exposure
[0123]
[0124] In the formula, P i,t To E i,t The functional relationship is obtained by fitting the historical bid and exposure relationship of the ith information platform in the tth period using the univariate linear regression analysis method. The fitting formula between bid and cost in formula ① can be used as a reference.
[0125] ④ The upper and lower limits of the budget for each information platform
[0126]
[0127] In the formula, is the lower limit of the budget of the i-th information platform; is the budget upper limit of the i-th information platform.
[0128] ⑤ The upper and lower limits of bids for each information platform
[0129]
[0130] In the formula, is the bid lower limit of the i-th information platform; is the upper limit of the bid of the i-th information platform.
[0131] Step 23: Under the constraints, use the POA algorithm to optimize and solve.
[0132] Considering that the bidding model established above is a multi-dimensional, multi-stage optimization problem, the more mature POA algorithm is used to solve the model. The POA algorithm does not need to discretize the state variables and decision variables in the model, and can be solved directly according to continuous variables, so it will not cause the "curse of dimensionality" problem caused by discretized variables.
[0133] The POA algorithm is an improved dynamic programming algorithm. During calculation, only the decision variables of two stages in the multi-stage decision are adjusted each time, thereby converting the multi-stage problem into multiple two-stage problems. Except for the initial solution of the first optimization, which needs to be set manually, the previous optimization result is used as the initial solution for each optimization. This is repeated until the result meets the requirements. Because the POA algorithm only needs to consider the optimal problem of two stages in each cycle, it greatly reduces the workload and is suitable for solving this model.
[0134] As Figure 5 shown, the specific solution process is as follows:
[0135] Step 231: Set the bids, the number of iterations N, and the iteration step size corresponding to the initial solutions of all information platforms (assuming there are m information platforms).
[0136] Step 232: Initialize the iteration number iter = 1.
[0137] Step 233: Initialize the calculation period t = 1.
[0138] Step 234: Fix the bids of each information platform in the (t - 1)-th period and the (t + 1)-th period, and adjust the bid of the t-th period according to the iteration step size.
[0139] Step 235: According to the placement demand parameters of each information platform in the t-th period, select the optimal bids {P 1,t , P 2,t … P m,t} of each information platform in the t-th period.
[0140] Step 236: Judge whether t < T - 1 holds. If it holds, calculate t = t + 1 and jump to Step 234. If not, jump to Step 237.
[0141] Step 237: Judge whether iter < N is satisfied. If it is satisfied, calculate iter = iter + 1 and jump to Step 233. If not, jump to Step 238.
[0142] Step 238: Calculate the placement demand parameters within the entire optimization period and take them as the optimal values.
[0143] Step 239: Judge whether the accuracy of the optimal value meets the accuracy requirement (that is, judge whether the difference between the optimal values calculated in the last two iterations is less than the preset accuracy). If so, the optimization ends, and the bids of each information platform in the optimal value state are the optimal information placement parameters. If not, take the bids of each information platform in the optimal value state as the initial solutions, and then jump to Step 232.
[0144] The optimization method of this embodiment uses optimization theory to model the problem of advertising information placement bids, gives the objective function and constraint conditions, and uses an optimization algorithm for optimization. Finally, the process of obtaining the globally optimal bid has the following advantages:
[0145] (1) A quantifiable objective function is proposed. The dimensionless processing method is used to process the three indicators of download volume, exposure volume and download exposure rate, which have inconsistent orders of magnitude. A weight coefficient ranging from 0 to 1 is given for App brands to select. The three objectives that affect the bid are comprehensively considered so that the bid meets the requirements of the App brand.
[0146] (2) In order to analyze the correlation between historical bids, exposure, downloads, and spending, the least squares method in univariate linear regression analysis is used to perform correlation analysis on historical bids and exposure, historical bids and downloads, and historical bids and spending. The correlation in the historical data is quantified using a functional relationship and used as a constraint for optimization, so that the optimization bidding process can fully refer to the historical data.
[0147] (3) The efficiency of information delivery and the matching degree between delivery parameters and actual delivery results are improved. Global optimization is performed in two dimensions: all information platforms and one optimization cycle. An optimization objective function with weight coefficients and reasonable constraints are given. The optimization algorithm is used to solve the problem to ensure the optimal bid reference value.
[0148] Example 3
[0149] Figure 6 The information delivery optimization system of this embodiment is shown. The information delivery optimization system comprises: a data collection module 30 , an objective function module 31 , a constraint condition module 32 and an optimization solution module 33 .
[0150] The data collection module 30 is used to collect historical delivery data, and the historical delivery data includes historical information delivery parameters and actual delivery result parameters obtained by delivering information according to the historical information delivery parameters. The objective function module 31 is used to obtain delivery demand parameters and establish an objective function according to the delivery demand parameters. The constraint condition module 32 is used to analyze the correlation between the historical information delivery parameters and the actual delivery result parameters according to the historical delivery data, and use the analysis results as optimization constraints. The optimization solution module 33 is used to solve the optimal information delivery parameters that make the objective function reach the delivery demand parameters under the constraints using an optimization algorithm. Among them, the historical delivery data can have a time limit, for example, including information delivery parameters counted in a preset time period in the past and delivery result parameters actually achieved by the delivery information in the time period. The historical delivery data can be counted by the information delivery party itself, or provided by the information platform for delivery information, or provided by a third-party service agency, which is not limited in this embodiment. The delivery demand parameters can be determined by the information delivery party according to the type of information and the delivery result to be achieved. The optimal information delivery parameters can usually be considered as information delivery parameters or a combination of parameters that achieve the best delivery result under the constraints.
[0151] In this embodiment, the actual delivery result parameter may include the actual value of at least one indicator used to measure the delivery result, and the delivery demand parameter may include the target value of the at least one indicator. The specific indicator selection can be determined by the information delivery party based on the type of information and the delivery result to be achieved.
[0152] In this embodiment, the objective function module 31 can be used to establish sub-objective functions for the target value of each indicator, and combine each sub-objective function to establish a total objective function. If the sub-objective functions are directly combined, a large error may occur. The objective function module 31 uses a weight method and a dimensionless method to combine each sub-objective function to establish a total objective function.
[0153] The constraint condition module 32 can be specifically used to analyze the relationship function between the historical information delivery parameter and each of the indicators according to the mapping relationship between the historical information delivery parameter and the actual value of each of the indicators, and use the relationship function as the optimization constraint condition. In order to simplify the calculation process and reduce the difficulty of modeling, the constraint condition module 32 preferably uses regression analysis to analyze the relationship function between the historical information delivery parameter and each of the indicators, and further obtains the correlation between the information delivery parameter and the actual delivery result achieved.
[0154] In this embodiment, the information delivery parameters may include the bid for delivering the information and the information delivery position associated with the bid on the information platform. The association between the bid and the information delivery position is usually determined by the information platform itself, and different information delivery positions may bring different delivery results. Since the relationship between the bid and the information delivery position may be fixed, the objective function module 31 may actually analyze the correlation between the bid for delivering the information and the actual delivery result parameters achieved, such as the relationship function between the bid for delivering the information and each of the indicators.
[0155] In addition to the correlation between the information delivery parameters and the actual delivery result parameters achieved as optimization constraints, in actual applications, the constraints may also include but are not limited to: at least one of the platform investment budget cost limit, the information platform's delivery budget upper and lower limits, and the information platform's delivery bid upper and lower limits.
[0156] The optimization solution module 33 can select different optimization algorithms according to the correlation between the information delivery parameters and the actual delivery results achieved, the specific values of the information delivery parameters, and the specific indicators for measuring the delivery results. In this embodiment, the optimization algorithm used by the optimization solution module 33 can include but is not limited to any one of a stepwise optimization algorithm, a linear programming algorithm, and a dynamic programming algorithm.
[0157] Considering that in actual applications, the information delivery party may deliver information on more than one information platform in order to obtain better delivery results, in order to better coordinate and plan the delivery strategies of all information platforms, the information delivery optimization system of this embodiment can also:
[0158] In the case where the information is delivered to at least one information platform, the delivery demand parameters are set for each of all information platforms, and the optimal information delivery parameters include the information delivery parameters of each of all information platforms. Of course, the delivery demand parameters are set only for some of all information platforms, and the optimal information delivery parameters include the information delivery parameters of some of all information platforms, and the information delivery optimization system of this embodiment is also applicable.
[0159] Considering that the bidding price of the information platform in each period changes in real time in actual application, the information delivery optimization system of this embodiment can also:
[0160] In the case where an optimization cycle is divided into at least one time period, the delivery demand parameter is set for each time period in an optimization cycle, and the optimal information delivery parameter includes the information delivery parameter for each time period in an optimization cycle. Of course, the delivery demand parameter is set only for part of the time periods in an optimization cycle, and the optimal information delivery parameter includes the information delivery parameter for part of the time periods in an optimization cycle, and the information delivery optimization system of this embodiment is also applicable.
[0161] Of course, considering that the information delivery party may deliver information to more than one information platform in order to obtain better delivery results, in order to better coordinate the delivery strategies of all information platforms, and the bidding of each information platform in each period changes in real time, the information delivery optimization system of this embodiment can also:
[0162] In the case where the information is delivered to at least one information platform and an optimization cycle is divided into at least one time period, the delivery demand parameters are set for each information platform in all information platforms in each time period or part of a time period in an optimization cycle, and the optimal information delivery parameters include the information delivery parameters of each information platform in all information platforms in each time period or part of a time period in an optimization cycle. Of course, the information delivery optimization system of this embodiment is also applicable if the delivery demand parameters are only set for some information platforms in each time period or part of a time period in an optimization cycle, and the optimal information delivery parameters include the information delivery parameters of some information platforms in each time period or part of a time period in an optimization cycle.
[0163] For example, in the case where the information is delivered to at least one information platform and an optimization period is divided into at least one time period, the delivery demand parameter can be set for each information platform among all information platforms in each time period within an optimization period, the optimal information delivery parameter can include the information delivery parameter of each information platform among all information platforms in each time period within an optimization period, and the information delivery parameter includes the bid for delivering the information.
[0164] When the optimization solving module 33 solves for the optimal information delivery parameter that makes the objective function reach the delivery demand parameter under the constraint conditions, it is specifically used for:
[0165] Set the bid, the number of iterations N, and the iteration step size corresponding to the initial solution of each information platform among all information platforms;
[0166] Initialize the number of iterations iter = 1;
[0167] Initialize the calculation time period t = 1;
[0168] Fix the bids of each information platform in the (t - 1)-th time period and the (t + 1)-th time period and adjust the bid of the t-th time period according to the iteration step size;
[0169] According to the delivery demand parameter of each information platform in the t-th time period, select the optimal bid of each information platform in the t-th time period;
[0170] Judge whether t < T - 1 holds. If it holds, calculate t = t + 1 and jump to the step of fixing the bids of each information platform in the (t - 1)-th time period and the (t + 1)-th time period and adjusting the bid of the t-th time period according to the iteration step size. If it does not hold, then:
[0171] Judge whether iter < N is satisfied. If it is satisfied, calculate iter = iter + 1 and jump to the step of initializing the calculation time period t = 1. If it is not satisfied, then:
[0172] Calculate the delivery demand parameter within the entire optimization period and take it as the optimal value;
[0173] Judge whether the accuracy of the optimal value meets the accuracy requirement. If so, the optimization ends, and the bid of each information platform in the optimal value state is the optimal information delivery parameter. If not, then take the bid of each information platform in the optimal value state as the initial solution, and then jump back to the step of initializing the number of iterations iter = 1.
[0174] The information delivery optimization system in this embodiment can be applied to a variety of information delivery scenarios, such as advertising information delivery scenarios: the advertising information delivery party delivers advertising information to some information platforms to achieve the purpose of promoting and advertising products, and the indicators used to measure the delivery results may include but are not limited to: the purchase volume of the advertising product, the exposure volume of the advertising product, and at least one of the purchase exposure volume of the advertising product. Of course, the application scenarios are not limited to this, and this embodiment will not cite them one by one.
[0175] Example 4
[0176] Figure 7 This is a schematic diagram of the structure of an electronic device provided in Example 4 of the present invention. The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the information delivery optimization method of Example 1 is implemented when the processor executes the program. Figure 7 The electronic device 40 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.
[0177] like Figure 7 As shown, the electronic device 40 may be in the form of a general-purpose computing device, for example, it may be a server device. The components of the electronic device 40 may include, but are not limited to: at least one processor 41, at least one memory 42, and a bus 43 connecting different system components (including the memory 42 and the processor 41).
[0178] The bus 43 includes a data bus, an address bus, and a control bus.
[0179] The memory 42 may include a volatile memory, such as a random access memory (RAM) 421 and / or a cache memory 422 , and may further include a read-only memory (ROM) 423 .
[0180] The memory 42 may also include a program / utility 425 having a set (at least one) of program modules 424, such program modules 424 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.
[0181] The processor 41 executes various functional applications and data processing by running the computer programs stored in the memory 42, such as the information delivery optimization method provided in Example 1 of the present invention.
[0182] The electronic device 40 may also communicate with one or more external devices 44 (e.g., keyboards, pointing devices, etc.). Such communication may be performed via an input / output (I / O) interface 45. Furthermore, the model generating device 40 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 46. Figure 7 As shown, the network adapter 46 communicates with other modules of the model-generated device 40 via the bus 43. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the model-generated device 40, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID (disk array) systems, tape drives, and data backup storage systems, etc.
[0183] It should be noted that although several units / modules or sub-units / modules of the electronic device are mentioned in the above detailed description, this division is merely exemplary and not mandatory. In fact, according to an embodiment of the present invention, the features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided into multiple units / modules to be embodied.
[0184] Example 5
[0185] This embodiment provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps of the information delivery optimization method provided in Embodiment 1 are implemented.
[0186] The readable storage medium may include but is not limited to: a portable disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical storage device, a magnetic storage device or any suitable combination of the above.
[0187] In a possible implementation manner, the present invention can also be implemented in the form of a program product, which includes a program code. When the program product is run on a terminal device, the program code is used to enable the terminal device to execute the steps in the information delivery optimization method described in Example 1.
[0188] The program code for executing the present invention may be written in any combination of one or more programming languages, and may be executed entirely on a user device, partially on a user device, as an independent software package, partially on a user device and partially on a remote device, or entirely on a remote device.
[0189] Although the specific embodiments of the present invention are described above, those skilled in the art should understand that these are only examples, and the protection scope of the present invention is defined by the appended claims. Those skilled in the art may make various changes or modifications to these embodiments without departing from the principles and essence of the present invention, but these changes and modifications all fall within the protection scope of the present invention.
Claims
1. A method for optimizing information delivery, characterized in that: Including: Collecting historical delivery data, where the historical delivery data includes historical information delivery parameters and actual delivery result parameters obtained by delivering information according to the historical information delivery parameters; Obtaining delivery demand parameters and establishing an objective function based on the delivery demand parameters; Analyzing the correlation between the historical information delivery parameters and the actual delivery result parameters according to the historical delivery data, and using the analysis result as an optimization constraint condition; The information is delivered to at least one information platform, and an optimization period is divided into at least one time period. The delivery demand parameters are set for each information platform in each time period within an optimization period. The optimal information delivery parameters include the information delivery parameters of each information platform in each time period within an optimization period. The information delivery parameters include the bid price for delivering the information; Under the constraint condition, using a step-by-step optimization algorithm to solve the optimal information delivery parameters that make the objective function reach the delivery demand parameters, including: Setting the bid price, the number of iterations N, and the iteration step size corresponding to the initial solution of each information platform; Initializing the iteration number iter = 1; Initializing the calculation time period t = 1; Fixing the bid prices of the (t - 1)-th and (t + 1)-th time periods of each information platform and adjusting the bid price of the t-th time period according to the iteration step size; Selecting the optimal bid price of each information platform in the t-th time period according to the delivery demand parameters of each information platform in the t-th time period; Judging whether t < T - 1 holds. If it holds, calculate t = t + 1 and jump to the step of fixing the bid prices of the (t - 1)-th and (t + 1)-th time periods of each information platform and adjusting the bid price of the t-th time period. If it does not hold, then: Judging whether iter < N is satisfied. If it is satisfied, calculate iter = iter + 1 and jump to the step of initializing the calculation time period t = 1. If it is not satisfied, then: Calculating the delivery demand parameters within the entire optimization period and taking them as the optimal values; Judging whether the accuracy of the optimal value meets the accuracy requirement. If so, the optimization ends, and the bid price of each information platform in the optimal value state is the optimal information delivery parameter. If not, then taking the bid price of each information platform in the optimal value state as the initial solution, and then jumping back to the step of initializing the iteration number iter = 1; 2. The information delivery optimization method according to claim 1, characterized in that: The actual delivery result parameters include the actual values of at least one indicator for measuring the delivery result, and the delivery demand parameters include the target values of the at least one indicator; The step of establishing an objective function according to the delivery demand parameters includes: Respectively establishing sub-objective functions for the target values of each of the indicators; Combining each sub-objective function to establish a total objective function; The step of analyzing the correlation between the historical information delivery parameters and the actual delivery result parameters according to the historical delivery data and using the analysis result as an optimization constraint condition includes: Respectively analyzing the relationship functions between the historical information delivery parameters and each of the indicators according to the mapping relationship between the historical information delivery parameters and the actual values of each of the indicators; Taking the relationship function as an optimization constraint condition.
3. The information delivery optimization method according to claim 2, characterized in that: Use the weight method and dimensionless method to combine each sub-objective function and establish the overall objective function; And / or, using regression analysis method to analyze the relationship function between the historical information delivery parameter and each of the indicators.
4. The information delivery optimization method according to claim 1, characterized in that: The information is delivered to at least one information platform, the delivery demand parameters are set for each information platform or part of all information platforms, and the optimal information delivery parameters include information delivery parameters for each information platform or part of all information platforms; Or, an optimization cycle is divided into at least one time period, the delivery demand parameter is set for each time period or part of the time period in an optimization cycle, and the optimal information delivery parameter includes the information delivery parameter for each time period or part of the time period in an optimization cycle; Or, the information is delivered to at least one information platform, an optimization cycle is divided into at least one time period, the delivery demand parameters are set for each information platform or part of the information platforms among all the information platforms in each time period or part of the time period in an optimization cycle, and the optimal information delivery parameters include the information delivery parameters of each information platform or part of the information platforms among all the information platforms in each time period or part of the time period in an optimization cycle.
5. The information delivery optimization method according to claim 1 or 4, characterized in that: The information delivery parameters include a bid for delivering the information and an information delivery position associated with the bid on the information platform; And / or, the constraint conditions also include: at least one of: a limit on the platform's investment budget, an upper and lower limit on the information platform's delivery budget, and an upper and lower limit on the information platform's delivery bid.
6. An information delivery optimization system, characterized in that: include: A data collection module is used to collect historical delivery data, wherein the historical delivery data includes historical information delivery parameters and actual delivery result parameters obtained by delivering information according to the historical information delivery parameters; An objective function module is used to obtain delivery demand parameters and establish an objective function according to the delivery demand parameters; A constraint condition module, used to analyze the correlation between the historical information delivery parameters and the actual delivery result parameters according to the historical delivery data, and use the analysis results as optimization constraint conditions; The information is delivered to at least one information platform, and an optimization cycle is divided into at least one time period. The delivery demand parameter is set for each information platform in all information platforms in each time period in an optimization cycle. The optimal information delivery parameter includes the information delivery parameter of each information platform in all information platforms in each time period in an optimization cycle, and the information delivery parameter includes the bid for delivering the information; The optimization solution module is used to solve the optimal information delivery parameters that make the objective function reach the delivery demand parameters under the constraints by using a step-by-step optimization algorithm, and is specifically used to: Set the bid, number of iterations N and iteration step size corresponding to the initial solution of each information platform in all information platforms; Initialize the number of iterations iter = 1; Initialize calculation period t=1; Fix the bids of each information platform in the t-1th period and the t+1th period and adjust the bids of the tth period according to the iteration step; According to the delivery demand parameters of each information platform in the tth period, select the optimal bid of each information platform in the tth period; Determine whether t < T - 1 holds. If it holds, calculate t = t + 1 and jump to the step of fixing the bids for the (t - 1)-th period and the (t + 1)-th period of each information platform and adjusting the bid for the t-th period by the iteration step length. If it does not hold, then: Determine whether iter < N is satisfied. If it is satisfied, calculate iter = iter + 1 and jump to the step of initializing the calculation period t = 1. If it does not hold, then: Calculate the delivery demand parameters within the entire optimization period and take them as the optimal values; Determine whether the accuracy of the optimal value meets the accuracy requirement. If so, the optimization ends, and the bids of each information platform in the optimal value state are the optimal information delivery parameters. If not, take the bids of each information platform in the optimal value state as the initial solution, and then jump back to initialize the iteration count iter = 1.
7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, it implements the information delivery optimization method described in any one of claims 1 to 5.
8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by the processor, it implements the steps of the information delivery optimization method described in any one of claims 1 to 5.
