WO2022141456A1 - 一种广告投放分配方法、装置、电子设备及存储介质 - Google Patents

一种广告投放分配方法、装置、电子设备及存储介质 Download PDF

Info

Publication number
WO2022141456A1
WO2022141456A1 PCT/CN2020/142288 CN2020142288W WO2022141456A1 WO 2022141456 A1 WO2022141456 A1 WO 2022141456A1 CN 2020142288 W CN2020142288 W CN 2020142288W WO 2022141456 A1 WO2022141456 A1 WO 2022141456A1
Authority
WO
WIPO (PCT)
Prior art keywords
user
sequence
added sequence
added
preset
Prior art date
Application number
PCT/CN2020/142288
Other languages
English (en)
French (fr)
Inventor
麦浩前
Original Assignee
百果园技术(新加坡)有限公司
麦浩前
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 百果园技术(新加坡)有限公司, 麦浩前 filed Critical 百果园技术(新加坡)有限公司
Priority to PCT/CN2020/142288 priority Critical patent/WO2022141456A1/zh
Priority to CN202080003840.7A priority patent/CN115066702A/zh
Publication of WO2022141456A1 publication Critical patent/WO2022141456A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00Commerce
    • G06Q30/02Marketing; Price estimation or determination; Fundraising
    • G06Q30/0241Advertisements

Definitions

  • the invention relates to the technical field of product promotion, and in particular, to a method, device, electronic device and storage medium for advertising distribution.
  • the user growth of the product in addition to the natural growth brought about by the brand effect, cannot avoid advertising in the market to attract new ones.
  • the channels involved in advertising include different countries, different platforms, and different advertisers.
  • the cost of product promotion is limited, and major APPs hope to achieve rapid user growth through advertising. In the face of the fierce advertising market, how to allocate advertisements reasonably is the current problem.
  • Most of the existing advertising allocation methods are based on heuristic greedy algorithms. Specifically, set restrictions, including budget, target number of daily active users (DAU), and target revenue. After each adjustment of the advertising distribution plan, obtain information such as costs, benefits, and new users of recent advertising channels. Based on the objective function (return on investment ROI function), the maximum value of the objective function and the minimum value of the objective function are selected under the guaranteed constraints. Increase the delivery volume of the channel corresponding to the maximum value of the objective function, decrease the delivery volume of the channel corresponding to the minimum value of the objective function, calculate the maximum value of the objective function and the minimum value of the objective function again, and repeat the above steps until the objective function converges.
  • the problem with the existing advertising distribution scheme is that it is easy to fall into a short-term local optimum.
  • Embodiments of the present invention provide an advertising distribution method, apparatus, electronic device and storage medium, which are used to solve the problem that the existing advertising distribution scheme tends to fall into a short-term local optimum, resulting in unreasonable advertising distribution.
  • An embodiment of the present invention provides an advertisement placement distribution method, the method includes:
  • the predetermined corresponding relationship between the number of new users and the unit price of users under each advertising delivery channel, and the corresponding relationship between the number of new users and the income of a single user determine the return on investment of the new sequence of users, and judge whether it meets the predetermined requirements.
  • Set the convergence conditions if not, take each third user newly added sequence as the second user newly added sequence; Target users add sequences for ad serving assignment.
  • an advertising distribution device the device includes:
  • an acquisition module configured to acquire a first user addition sequence within a preset first time length, wherein the elements in the first user addition sequence are the number of user additions corresponding to each advertisement placement channel;
  • the first determination module is configured to copy the first user addition sequence to obtain a preset number of second user addition sequences, and for each second user addition sequence, add elements in the second user addition sequence Perform random changes to obtain a third user newly added sequence corresponding to the second user newly added sequence;
  • the second determination module is configured to determine the investment in the sequence of new user additions according to the predetermined corresponding relationship between the number of new users and the unit price of the user, and the corresponding relationship between the number of new users and the individual income of the user under each advertisement placement channel. rate of return, to determine whether the preset convergence conditions are met, if not, take each third user-added sequence as the second user-added sequence; if so, take the user-added sequence corresponding to the largest return on investment as the target user Add a sequence; perform advertisement placement allocation according to the target user's newly added sequence.
  • an embodiment of the present invention provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus;
  • the processor is configured to implement any of the method steps described above when executing the program stored in the memory.
  • an embodiment of the present invention provides a computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, any one of the method steps described above is implemented .
  • Embodiments of the present invention provide an advertising distribution method, apparatus, electronic device, and storage medium.
  • the method includes: acquiring a first user addition sequence within a preset first time length, wherein the first user newly added sequence
  • the elements in the increment sequence are the number of new users corresponding to each advertisement placement channel;
  • the first user increment sequence is copied to obtain a preset number of second user increment sequences, and for each second user increment sequence, Randomly changing elements in the second user-added sequence to obtain a third user-added sequence corresponding to the second user-added sequence;
  • the correspondence between the unit price, the number of new users and the individual income of the user determine the return on investment of the user-added sequence, and judge whether the preset convergence conditions are met.
  • each third user-added sequence is taken as the The second user-added sequence; if yes, the user-added sequence corresponding to the maximum return on investment is used as the target user-added sequence; the advertisement placement is allocated according to the target user-added sequence.
  • FIG. 1 is a schematic diagram of an advertisement placement allocation process provided in Embodiment 1 of the present invention.
  • FIG. 3 is a schematic diagram of the mating variation of user-added sequences provided in Embodiment 6 of the present invention.
  • FIG. 4 is a schematic structural diagram of an advertisement placement and distribution apparatus provided in Embodiment 7 of the present invention.
  • FIG. 5 is a schematic structural diagram of an electronic device according to Embodiment 8 of the present invention.
  • FIG. 1 is a schematic diagram of an advertisement placement allocation process provided by an embodiment of the present invention, and the process includes the following steps:
  • S101 Acquire a first user addition sequence within a preset first time length, wherein an element in the first user addition sequence is the number of user additions corresponding to each advertisement placement channel.
  • S102 Copy the first user-added sequence to obtain a preset number of second user-added sequences, and for each second user-added sequence, randomly change elements in the second user-added sequence to obtain
  • the second user addition sequence corresponds to the third user addition sequence.
  • S103 Determine the return on investment of the user-added sequence according to the predetermined correspondence between the number of new users and the unit price of the user, and the corresponding relationship between the number of new users and the individual income of the user under each advertisement placement channel, and determine whether If the preset convergence conditions are met, if not, use each third user-added sequence as the second user-added sequence; if yes, use the user-added sequence corresponding to the maximum return on investment as the target user-added sequence; according to The target user adds a sequence for advertisement placement allocation.
  • the advertising distribution method provided by the embodiment of the present invention is applied to an electronic device, and the electronic device may be a PC, a tablet computer, or the like.
  • the statistics are generally performed in units of days, that is, the number of new users added each day is counted. And the number of new users added every day is counted for each advertising channel, and the advertising channels can be subdivided into countries & advertisers & platforms.
  • the electronic device obtains, for each advertising delivery channel, the number of newly added users within a preset first period of time. Wherein, the preset first time length may be the last day, the last week, or the like. If it is the last week, get the number of new users added every day in the last week for each advertising channel, and then calculate the average of the number of new users added every day in the last week as the new user additions corresponding to the advertising channel quantity.
  • the number of newly added users corresponding to each advertising delivery channel is used as an element to constitute the first sequence of newly added users.
  • the electronic device After acquiring the first user-added sequence, the electronic device copies the first user-added sequence to obtain a preset number of second user-added sequences, and the preset number may be 100, 200, or the like.
  • the preset number may be 100, 200, or the like.
  • randomly change the elements in the second user-added sequence for example, randomly increase a certain value for each element in the second user-added sequence, or randomly decrease a certain value value, or remain unchanged, to obtain the third user-added sequence corresponding to the second user-added sequence.
  • the obtained return on investment of each user-added sequence can be determined.
  • words such as “first” and “second” in the first user-added sequence and the second user-added sequence in the embodiments of the present invention are only used for the purpose of distinguishing and describing, and cannot be understood To indicate or imply relative importance, nor should it be construed to indicate or imply order.
  • the first user-added sequence, the second user-added sequence, and the third user-added sequence are all user-added sequences.
  • the electronic device determines the return on investment of the new sequence of users according to the pre-determined correspondence between the number of new users and the unit price of users under each advertising channel, and the corresponding relationship between the number of new users and the individual income of the user. Specifically, for the newly added number of users corresponding to each advertisement delivery channel in each user addition sequence, the advertisement delivery can be determined according to the predetermined correspondence between the newly added user quantity and the user unit price under the advertisement delivery channel.
  • the cost of the channel sum the costs of each advertisement placement channel in the user-added sequence to obtain the total cost of the user-added sequence.
  • the advertising delivery channel can be determined according to the predetermined correspondence between the number of newly added users under the advertising delivery channel and the individual revenue of the user.
  • the revenue of each advertising channel in the user-added sequence is summed to obtain the total revenue of the user-added sequence. Then calculate the ratio of total revenue to total cost to get the return on investment of the user's newly added sequence.
  • the electronic device is preset with a convergence condition, and the preset convergence condition may be that when the determined maximum return on investment is greater than a preset return threshold, the convergence is determined.
  • the judging whether a preset convergence condition is met includes: judging whether the number of iterations of the second user-added sequence reaches a preset number threshold, or judging whether the determined maximum number of iterations in two adjacent iterations. Whether the difference in ROI is less than a preset convergence threshold. In this embodiment of the present invention, it is judged whether a preset convergence condition is met, and if not, each third user newly added sequence is used as a second user newly added sequence, and the process for each second user newly added sequence is continued.
  • Elements in the second user-added sequence are randomly changed to obtain a third user-added sequence corresponding to the second user-added sequence, until a preset convergence condition is satisfied. After judging that the preset convergence condition is satisfied, the user-added sequence corresponding to the maximum return on investment is used as the target user-added sequence.
  • Ad serving is allocated according to the new sequence of target users. Specifically, according to the predetermined correspondence between the number of new users and the unit price of users under each advertising channel, and the sequence of new target users, the cost of each advertising channel can be determined. The cost of the channel to advertise to each advertising channel.
  • the elements in the second user-added sequence are performed. Randomly changing, the obtained third user newly added sequence corresponding to the second user added sequence includes:
  • the elements of the corresponding channel in the fourth user newly added sequence and the corresponding second user newly added sequence are exchanged according to a preset random probability, and the fourth user newly added sequence is obtained.
  • Each fifth user-added sequence is regarded as a third user-added sequence.
  • each second user addition sequence random changes are performed on elements in the second user addition sequence to obtain a fourth user addition sequence corresponding to the second user addition sequence. Then, for each element in each fourth user-added sequence, according to a preset random probability, it is determined whether to exchange the element with the corresponding element in the corresponding second user-added sequence.
  • the elements of the same advertisement placement channel are the corresponding elements.
  • the preset random probability may be, for example, 40%, 50%, or the like.
  • the preset random probability is 0.4, for each element in the new sequence of each fourth user, there is a 40% probability of exchanging the element with the corresponding element in the corresponding second user's new sequence, and 60% The probability of not exchanging the element with the corresponding element in the corresponding second user-added sequence.
  • the elements of the corresponding channel in the fourth user newly added sequence and the corresponding second user newly added sequence are exchanged according to a preset random probability, and the fourth user newly added sequence is obtained.
  • the corresponding fifth user-added sequence is added, and then each fifth user-added sequence is used as a third user-added sequence.
  • elements in the second user newly added sequence are randomly changed to obtain a fourth user newly added sequence corresponding to the second user newly added sequence; For each fourth user newly added sequence, the elements of the corresponding channel in the fourth user newly added sequence and the corresponding second user newly added sequence are exchanged according to a preset random probability, and the fourth user newly added sequence is obtained.
  • the corresponding fifth user-added sequence; each fifth user-added sequence is used as a third user-added sequence. This further increases the randomness of the user's newly added sequence, which better avoids falling into the short-term local optimum problem, and makes the final advertising distribution more reasonable.
  • the idea of "survival of the fittest and survival of the fittest" in genetics is introduced.
  • the elements of the corresponding channel in the fourth user added sequence and the corresponding second user added sequence are based on a preset random probability.
  • each fourth user newly added sequence is selected, a preset number of selections are performed in each of the fourth user newly added sequences, and the fourth user newly added sequence selected each time is used for subsequent targeting.
  • Each fourth user addition sequence is a step of exchanging elements of the corresponding channel in the fourth user addition sequence and the corresponding second user addition sequence according to a preset random probability.
  • the electronic device determines the corresponding relationship between the number of new users and the unit price of users under each advertisement placement channel. , The corresponding relationship between the number of new users and the income of individual users, to determine the return on investment of each fourth user-added sequence.
  • the probability that each fourth user newly added sequence is selected is determined according to the return on investment of each fourth user newly added sequence, wherein, the greater the investment return rate, the greater the selected probability of the fourth user newly added sequence.
  • each fourth user-added sequence is selected.
  • the preset number of times and the preset number may be the same or different. It should be noted that each selection is not taken out from each fourth user-added sequence, but copied from each fourth user-added sequence. That is, the fourth user-added sequence selected each time still has a chance to be selected next time. Because the fourth user-added sequence with a greater return on investment has a higher probability of being selected, the fourth user-added sequence with a greater return on investment is more likely to be selected, and then the fourth user-added sequence selected each time is used.
  • the method further includes:
  • For each fifth user-added sequence randomly change the elements in the fifth user-added sequence to obtain a sixth user-added sequence corresponding to the fifth user-added sequence; use the sixth user-added sequence The sequence is updated to the fifth user-added sequence.
  • the elements in each fifth user-added sequence are added.
  • a random change is performed to obtain a sixth user-added sequence corresponding to the fifth user-added sequence.
  • the process of randomly changing the elements in the fifth user-added sequence is similar to the above-mentioned process of randomly changing the elements in the second user-added sequence, and details are not repeated here.
  • each fifth user-added sequence randomly change the elements in the fifth user-added sequence, and after obtaining the sixth user-added sequence corresponding to the fifth user-added sequence, use the sixth user-added sequence
  • the addition sequence updates the fifth user addition sequence. Then, a subsequent step of using each fifth user-added sequence as a third user-added sequence is performed.
  • the elements in the fifth user newly added sequence are randomly changed, so the randomness of the user newly added sequence is further increased, thereby further avoiding falling into
  • the problem of local optimization in the short term makes the final advertising distribution more reasonable.
  • the method further includes:
  • the cost of the user-added sequence is determined according to the predetermined correspondence between the number of new users and the unit price of the user; ;
  • the cost of the user-added sequence is not greater than the preset cost threshold, and the number of newly-added users of the user-added sequence is not less than the preset user.
  • the number threshold is exceeded, the user-added sequence is retained; otherwise, the user-added sequence is deleted.
  • the electronic device After each time the electronic device determines the user-added sequence through random changes or element exchange, it first determines whether the determined user-added sequence meets the requirements according to the preset restriction conditions, and if not, deletes the user-added sequence. , if satisfied, keep the user-added sequence.
  • the preset restrictions include that the revenue of the user-added sequence is not less than the preset revenue threshold, the cost of the user-added sequence is not greater than the preset cost threshold, and the number of new users of the user-added sequence is not less than the preset number of users threshold. If the above three judgments are satisfied at the same time, it is considered that the user-added sequence meets the preset restriction conditions; otherwise, it is considered that the user-added sequence does not meet the preset restriction conditions.
  • the cost of the user addition sequence is determined according to the predetermined correspondence between the user addition quantity and the user unit price.
  • the income of the new sequence of users is determined.
  • the sum of the number of new users in the user-added sequence is the number of new users in the user-added sequence.
  • the user-added sequence corresponding to the upper level of the deleted user-added sequence can be determined.
  • the added sequence changes to get the deleted user-added sequence.
  • the user-added sequence corresponding to the upper level is randomly changed or element exchanged to obtain the user-added sequence, until the obtained user-added sequence satisfies the preset restriction conditions. This not only retains the user-added sequences that meet the preset restriction conditions, but also ensures the number of user-added sequences.
  • each user-added sequence After each user-added sequence is determined, it is determined whether the user-added sequence satisfies a preset restriction condition. If not satisfied, delete the user-added sequence, if satisfied, keep the user-added sequence. This avoids the problem of slow convergence due to the existence of user-added sequences that do not meet the preset constraints during the iteration process, and may cause the final target user-added sequence to be not the optimal user-added sequence. .
  • the ad distribution plan has been further optimized.
  • the process of predetermining the corresponding relationship between the number of new users and the unit price of users under each advertisement placement channel includes:
  • each advertisement placement channel obtains the daily number of new users and the unit price of users under the advertisement placement channel; perform curve fitting on the daily number of new users and the unit price of users , to determine the corresponding relationship between the number of new users and the unit price of users under the advertising channel;
  • the process of predetermining the corresponding relationship between the number of new users and the individual income of the user under each advertisement placement channel includes:
  • each advertisement placement channel obtains the daily number of new users and the individual revenue of users under the advertisement placement channel; Perform curve fitting to determine the corresponding relationship between the number of new users under the advertising channel and the individual revenue of users.
  • the electronic device pre-determines the corresponding relationship between the number of new users under each advertising delivery channel and the unit price of the user, within a preset second period of time, for each advertising delivery channel, obtains the daily number of new users under the advertising delivery channel. and user price.
  • the preset second time length may be the last year, the last 6 months, or the like.
  • the daily new number of users under the advertising delivery channel can be obtained.
  • the daily user unit price can be determined . Then, curve fitting is performed on the number of new users per day and the unit price of users, and the corresponding relationship between the number of new users and the unit price of users under the advertising channel is determined.
  • the electronic device pre-determines the corresponding relationship between the number of new users under each advertising delivery channel and the individual revenue of the user, within a preset second period of time, for each advertising delivery channel, obtain the daily new user information of the advertising delivery channel. Increase the number and individual revenue of users. Specifically, for each advertising delivery channel, the daily new number of users under the advertising delivery channel can be obtained. In addition, by obtaining the daily revenue obtained by the advertising delivery channel and the daily newly added number of users, the daily user order can be determined. body income. Then, curve fitting is performed on the number of new users per day and the individual income of the user, and the corresponding relationship between the new number of users and the individual income of the user under the advertising channel is determined.
  • the corresponding relationship between the newly added number of users under each advertising delivery channel and the user unit price, and the corresponding relationship between the newly added number of users under each advertising delivery channel and the individual revenue of the user may be a one-dimensional quadratic equation corresponding relationship.
  • Each parameter in the quadratic equation of one variable is obtained by curve fitting, and the corresponding relationship is obtained.
  • FIG. 2 is a flowchart of advertisement placement allocation provided by an embodiment of the present invention.
  • a user presets restrictions, which include a preset revenue threshold, a preset cost threshold, and a preset user quantity threshold DAU.
  • a preset revenue threshold for each advertising delivery channel, obtain the daily new number of users, unit price of users, and individual revenue of users under the advertising delivery channel, and determine the new number of users under the advertising delivery channel through curve fitting.
  • DAU Daily Active User, means the number of daily active users, and in the embodiment of the present invention, represents the number of new users per day.
  • CPI The unit price of each new user who buys on average.
  • LTV The average benefit of each new user who buys the product, such as consumption and recharge.
  • ROI Return on investment, the ratio of the value that should be returned through an investment, that is, the ratio of benefits to costs.
  • Buying volume means the increase in the number of users who use the product for promotion through advertising.
  • the number of newly added users corresponding to each advertisement delivery channel within the preset first time length is regarded as a first newly added user sequence, and the first newly added user sequence is copied to obtain a preset number of second newly added users.
  • User-added sequences so that different second user-added sequences change randomly. Random change is mainly divided into selection and reproduction, where reproduction includes mating and mutation. The selection is related to the ROI of the user-added sequence, and the higher the ROI is, the greater the probability of the user-added sequence being selected. Gradually, user-added sequences with higher ROI will be inherited in the iteration, while user-added sequences with lower ROI will disappear in the iterative process.
  • the selected user-added sequence will enter the mating process.
  • the mating process is that the selected user-added sequence and the previous user-added sequence can exchange their respective elements with a certain probability, thereby generating a new user-added sequence to replace the original "old" user-added sequence.
  • the elements in the new user-added sequence will also mutate with a certain probability, and the final new user-added sequence will be generated through mutation.
  • Fig. 3 is a schematic diagram of the mating variation of user-added sequences provided by an embodiment of the present invention. As shown in Fig. 3, the newly-added sequence of user A is [A1 A2 A3 A4 ...
  • the newly-added sequence of user B is [B1 B2 B3 B4 ... ... Bn]
  • the new sequence of user A and the new sequence of user B are mated to obtain the new sequence of sub-user [A1 B2 A3 B4 ... Bn].
  • the sub-user added sequence is mutated to obtain the final user added sequence [A1 B2+rand() A3-rand() B4 ... Bn]. This process is similar to the reproductive process of biology. The offspring continuously acquire excellent genes from their parents, and then achieve the effect of adjustment through changes themselves.
  • each user-added sequence needs to meet the following constraints to ensure that the direction of optimization must meet the requirements.
  • Dau_predict(Ins i ) is the number of new users under each advertising channel in the new user sequence. Among them, the change of the user's newly added quantity Ins i will correspondingly affect the user's unit price Cpi i and the user's single income Ltv i , thereby affecting the cost and benefit of the user's new sequence.
  • Cpi_predict(Ins i ) and Ltv_predict(Ins i ) are based on each The corresponding relationship between the number of new users and the unit price of users under each advertising channel, and the corresponding relationship between the number of new users and the individual revenue of users under each advertising channel. The unit price and individual revenue of users are determined.
  • FIG. 4 is a schematic structural diagram of an advertisement placement and distribution device provided by an embodiment of the present invention, and the device includes:
  • the obtaining module 41 is configured to obtain a first user addition sequence within a preset first time length, wherein the elements in the first user addition sequence are the number of user additions corresponding to each advertisement placement channel;
  • the first determination module 42 is configured to copy the first user-added sequence to obtain a preset number of second user-added sequences, and for each second-user-added sequence, select the second user-added sequence for the second user-added sequence.
  • the elements are randomly changed to obtain the third user newly added sequence corresponding to the second user newly added sequence;
  • the second determination module 43 is configured to determine the sequence of user additions according to the predetermined corresponding relationship between the number of new users and the unit price of the user, and the corresponding relationship between the number of new users and the individual income of the user under each advertisement placement channel. Return on investment, determine whether the preset convergence conditions are met, if not, take each third user newly added sequence as the second user added sequence; if so, take the user added sequence corresponding to the largest ROI as the target User-added sequence; advertisement placement distribution is performed according to the target user-added sequence.
  • the first determining module 42 is specifically configured to randomly change the elements in the second user-added sequence for each second user-added sequence to obtain the fourth user-added sequence corresponding to the second user-added sequence. increase sequence; for each fourth user newly added sequence, exchange the elements of the corresponding channel in the fourth user newly added sequence and the corresponding second user newly added sequence according to a preset random probability, and obtain a new sequence with the fourth user.
  • the fifth user-added sequence corresponding to the newly-added sequence; each fifth user-added sequence is regarded as a third user-added sequence.
  • the device also includes:
  • the selection module 44 is configured to determine the probability that each fourth user newly added sequence is selected according to the return on investment of each fourth user newly added sequence, wherein the fourth user newly added sequence with a greater return on investment is selected. The greater the probability of selection; according to the probability that each fourth user newly added sequence is selected, a preset number of selections are performed in each of the fourth user newly added sequences, and the selected fourth user new sequence is used each time. Incremental sequence, triggering the first determining module 42 .
  • the device also includes:
  • the updating module 45 is configured to randomly change the elements in the fifth user-added sequence for each fifth user-added sequence to obtain a sixth user-added sequence corresponding to the fifth user-added sequence; adopting The sixth user addition sequence updates the fifth user addition sequence.
  • the device also includes:
  • the deletion module 46 is configured to determine the cost of the user-added sequence according to the predetermined correspondence between the number of new users and the unit price of the user; The revenue of the user-added sequence; when the revenue of the user-added sequence is not less than the preset revenue threshold, the cost of the user-added sequence is not greater than the preset cost threshold, the user-added sequence of the user-added user When the number is not less than the preset user number threshold, the user-added sequence is retained; otherwise, the user-added sequence is deleted.
  • the second determination module 43 is specifically configured to determine whether the number of iterations of the second user-added sequence reaches a preset number threshold, or to determine the difference between the determined maximum ROI in two adjacent iterations. Whether the value is less than the preset convergence threshold.
  • the device also includes:
  • the third determining module 47 is configured to, within a preset second time length, for each advertisement placement channel, obtain the daily number of new users and the unit price of users under the advertisement placement channel; Curve fitting is performed between the quantity and the unit price of the user, and the corresponding relationship between the number of new users and the unit price of the user under the advertising channel is determined;
  • the third determination module 47 is further configured to, within a preset second time length, for each advertisement placement channel, obtain the daily number of new users and individual user income under the advertisement placement channel; Curve fitting is performed between the number of new users per day and the income of individual users, and the corresponding relationship between the number of new users and the income of individual users under the advertising channel is determined.
  • the embodiments of the present invention further provide an electronic device, as shown in FIG. 5 , including: a processor 301 , a communication interface 302 , a memory 303 and a communication bus 304 , wherein the processor 301 , the communication interface 302, and the memory 303 through the communication bus 304 to complete the mutual communication;
  • a computer program is stored in the memory 303.
  • the program is executed by the processor 301, the following steps are performed: first, the first user-added sequence within a preset first time length is obtained, and a preset number of second user sequences are obtained by copying.
  • the elements in each second user addition sequence are randomly changed to obtain a third user addition sequence corresponding to the second user addition sequence.
  • Each third user-added sequence is taken as the second user-added sequence, and the above random change operation is repeated until the preset convergence condition is satisfied.
  • the user-added sequence corresponding to the maximum return on investment is used as the target user-added sequence; advertisement placement is allocated according to the target user-added sequence.
  • An electronic device provided by the embodiment of the present invention can be used to execute the advertisement distribution method provided by any of the above embodiments, and has corresponding functions and beneficial effects.
  • embodiments of the present invention further provide a computer storage-readable storage medium, where a computer program executable by an electronic device is stored in the computer-readable storage medium.
  • the electronic device When running on the electronic device, the electronic device is caused to perform the following steps: firstly obtain the first user-added sequence within a preset first time length, obtain a preset number of second user-added sequences through duplication, and perform the following steps for each first user-added sequence. Elements in the second user-added sequence are randomly changed to obtain a third user-added sequence corresponding to the second user-added sequence. Each third user-added sequence is taken as the second user-added sequence, and the above random change operation is repeated until the preset convergence condition is satisfied. Then, the user-added sequence corresponding to the maximum return on investment is used as the target user-added sequence; advertisement placement is allocated according to the target user-added sequence.
  • the computer storage readable storage medium further provided by the embodiments of the present invention stores therein a computer program executable by an electronic device.
  • the program runs on the electronic device, the program can be used to execute the advertisement distribution provided in any of the foregoing embodiments.
  • the method has corresponding functions and beneficial effects.

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Strategic Management (AREA)
  • Finance (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

一种广告投放分配方法、装置、电子设备及存储介质,该方法中,获取到第一用户新增序列之后,复制得到预设数量的第二用户新增序列,然后对每个第二用户新增序列中的元素进行随机变化,将变化后得到的第三用户新增序列作为第二用户新增序列,继续进行上述过程,直至满足预设的收敛条件。然后将整个过程中最大的投资回报率对应的用户新增序列作为目标用户新增序列,根据目标用户新增序列进行广告投放分配。该方法中首先复制得到多个第二用户新增序列,其次对每个第二用户新增序列中的元素进行随机变化,通过上述操作避免陷入短期内局部最优的问题,使得最终的广告投放分配更加合理。

Description

一种广告投放分配方法、装置、电子设备及存储介质 技术领域
本发明涉及产品推广技术领域,尤其涉及一种广告投放分配方法、装置、电子设备及存储介质。
背景技术
产品的用户增长,除了品牌效应带来的自然增长,避免不了去市场上投放广告进行拉新。广告投放涉及的渠道包括不同国家、不同平台、不同广告商等。产品推广的成本是有限的,各大APP都希望通过广告投放来达到用户快速增长。面对激烈的广告市场,如何合理分配广告投放是目前面对的难题。
现有的广告投放分配方法大多是基于启发式的贪心算法。具体来说,设定限制条件,包括预算、目标日活跃用户数量DAU、目标营收,每次调整广告投放分配方案之后,获取近期广告投放各渠道的成本、收益、新增用户等信息。基于目标函数(投资回报率ROI函数),在保证限制条件情况下,选择目标函数最大值和目标函数最小值。增大目标函数最大值对应的渠道的投放量,减小目标函数最小值对应的渠道的投放量,再次进行目标函数最大值和目标函数最小值的计算,重复上述步骤,直到目标函数收敛为止。现有的广告投放分配方案存在的问题是,容易陷入短期内局部最优。
发明内容
本发明实施例提供了一种广告投放分配方法、装置、电子设备及存储介质,用以解决现有的广告投放分配方案容易陷入短期内局部最优,导致广告投放分配不合理的问题。
本发明实施例提供了一种广告投放分配方法,所述方法包括:
获取预设第一时间长度内的第一用户新增序列,其中,所述第一用户新增序列中的元素为每个广告投放渠道对应的用户新增数量;
将所述第一用户新增序列复制得到预设数量的第二用户新增序列,针对每个第二用户新增序列,对该第二用户新增序列中的元素进行随机变化,得到该第二用户新增序列对应的第三用户新增序列;
根据预先确定的所述每个广告投放渠道下用户新增数量与用户单价的对应关系、用户新增数量与用户单体收益的对应关系,确定用户新增序列的投资回报率,判断是否满足预设的收敛条件,如果否,将每个第三用户新增序列作为第二用户新增序列;如果是,将最大的投资回报率对应的用户新增序列作为目标用户新增序列;根据所述目标用户新增序列进行广告投放分配。
另一方面,本发明实施例提供了一种广告投放分配装置,所述装置包括:
获取模块,用于获取预设第一时间长度内的第一用户新增序列,其中,所述第一用户新增序列中的元素为每个广告投放渠道对应的用户新增数量;
第一确定模块,用于将所述第一用户新增序列复制得到预设数量的第二用户新增序列,针对每个第二用户新增序列,对该第二用户新增序列中的元素进行随机变化,得到该第二用户新增序列对应的第三用户新增序列;
第二确定模块,用于根据预先确定的所述每个广告投放渠道下用户新增数量与用户单价的对应关系、用户新增数量与用户单体收益的对应关系,确定用户新增序列的投资回报率,判断是否满足预设的收敛条件,如果否,将每个第三用户新增序列作为第二用户新增序列;如果是,将最大的投资回报率对应的用户新增序列作为目标用户新增序列;根据所述目标用户新增序列进行广告投放分配。
另一方面,本发明实施例提供了一种电子设备,包括处理器、通信接口、存储器和通信总线,其中,处理器,通信接口,存储器通过通信总线完成相互间的通信;
存储器,用于存放计算机程序;
处理器,用于执行存储器上所存放的程序时,实现上述任一项所述的方法步骤。
另一方面,本发明实施例提供了一种计算机可读存储介质,所述计算机可读存储介质内存储有计算机程序,所述计算机程序被处理器执行时实现上述任一项所述的方法步骤。
本发明实施例提供了一种广告投放分配方法、装置、电子设备及存储介质,所述方法包括:获取预设第一时间长度内的第一用户新增序列,其中,所述第一用户新增序列中的元素为每个广告投放渠道对应的用户新增数量;将所述第一用户新增序列复制得到预设数量的第二用户新增序列,针对每个第二用户新增序列,对该第二用户新增序列中的元素进行随机变化,得到该第二用户新增序列对应的第三用户新增序列;根据预先确定的所述每个广告投放渠道下用户新增数量与用户单价的对应关系、用户新增数量与用户单体收益的对应关系,确定用户新增序列的投资回报率,判断是否满足预设的收敛条件,如果否,将每个第三用户新增序列作为第二用户新增序列;如果是,将最大的投资回报率对应的用户新增序列作为目标用户新增序列;根据所述目标用户新增序列进行广告投放分配。
由于在本发明实施例中,获取到第一用户新增序列之后,复制得到预设数量的第二用户新增序列,然后对每个第二用户新增序列中的元素进行随机变化,将变化后得到的第三用户新增序列作为第二用户新增序列,继续进行上述过程,直至满足预设的收敛条件。然后将整个过程中最大的投资回报率对应的用户新增序列作为目标用户新增序列,根据目标用户新增序列进行广告投放分配。本发明实施例中首先复制得到多个第二用户新增序列,其次对每个第二用户新增序列中的元素进行随机变化,通过上述操作避免陷入短期内局部最优的问题,使得最终的广告投放分配更加合理。
附图说明
为了更清楚地说明本发明实施例中的技术方案,下面将对实施例描述中所需要使用的附图作简要介绍,显而易见地,下面描述中的附图仅仅是本发明的一些实施例,对于本领域的普通技术人员来讲,在不付出创造性劳动的前提下,还可以根据这些附图获得其他的附图。
图1为本发明实施例1提供的广告投放分配过程示意图;
图2为本发明实施例6提供的广告投放分配流程图;
图3为本发明实施例6提供用户新增序列交配变异示意图;
图4为本发明实施例7提供的广告投放分配装置结构示意图;
图5为本发明实施例8提供的电子设备结构示意图。
具体实施方式
下面将结合附图对本发明作进一步地详细描述,显然,所描述的实施例仅仅是本发明一部分实施例,而不是全部的实施例。基于本发明中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其它实施例,都属于本发明保护的范围。
实施例1:
图1为本发明实施例提供的广告投放分配过程示意图,该过程包括以下步骤:
S101:获取预设第一时间长度内的第一用户新增序列,其中,所述第一用户新增序列中的元素为每个广告投放渠道对应的用户新增数量。
S102:将所述第一用户新增序列复制得到预设数量的第二用户新增序列,针对每个第二用户新增序列,对该第二用户新增序列中的元素进行随机变化,得到该第二用户新增序列对应的第三用户新增序列。
S103:根据预先确定的所述每个广告投放渠道下用户新增数量与用户单价的对应关系、用户新增数量与用户单体收益的对应关系,确定用户新增序列的投资回报率,判断是否满足预设的收敛条件,如果否,将每个第 三用户新增序列作为第二用户新增序列;如果是,将最大的投资回报率对应的用户新增序列作为目标用户新增序列;根据所述目标用户新增序列进行广告投放分配。
本发明实施例提供的广告投放分配方法应用于电子设备,该电子设备可以是PC、平板电脑等设备。
在统计用户新增数量时,一般是以天为单位进行统计的,也就是统计每天的用户新增数量。并且统计每天的用户新增数量是针对每个广告投放渠道进行统计的,广告投放渠道可以细分至国家&广告商&平台。电子设备针对每个广告投放渠道,获取预设第一时间长度内的用户新增数量。其中,预设第一时间长度可以是最近一天,也可以是最近一周等。如果是最近一周,则获取针对每个广告投放渠道,获取最近一周内每天的用户新增数量,然后计算最近一周内每天的用户新增数量的平均值,作为该广告投放渠道对应的用户新增数量。每个广告投放渠道对应的用户新增数量作为元素构成第一用户新增序列。
电子设备获取到第一用户新增序列之后,将第一用户新增序列复制得到预设数量的第二用户新增序列,预设数量可以是100、200等。针对每个第二用户新增序列,对该第二用户新增序列中的元素进行随机变化,例如对该第二用户新增序列中的每个元素随机增加某个数值,或者随机减少某个数值,或者保持不变,得到该第二用户新增序列对应的第三用户新增序列。
在广告投放分配过程中,可以确定出得到的每个用户新增序列的投资回报率。需要说明的是,本发明实施例中的第一用户新增序列、第二用户新增序列等中的“第一”、“第二”等词汇,仅用于区分描述的目的,而不能理解为指示或暗示相对重要性,也不能理解为指示或暗示顺序。第一用户新增序列、第二用户新增序列、第三用户新增序列等都是用户新增序列。电子设备根据预先确定的每个广告投放渠道下用户新增数量与用户单价的对应关系、用户新增数量与用户单体收益的对应关系,确定用户新增 序列的投资回报率。具体的,针对每个用户新增序列中的每个广告投放渠道对应的用户新增数量,根据预先确定的该广告投放渠道下用户新增数量与用户单价的对应关系,可以确定出该广告投放渠道的成本,将该用户新增序列中每个广告投放渠道的成本求和,得到该用户新增序列的总成本。针对每个用户新增序列中的每个广告投放渠道对应的用户新增数量,根据预先确定的该广告投放渠道下用户新增数量与用户单体收益的对应关系,可以确定出该广告投放渠道的收益,将该用户新增序列中每个广告投放渠道的收益求和,得到该用户新增序列的总收益。然后计算总收益与总成本的比值,得到该用户新增序列的投资回报率。
电子设备预设有收敛条件,预设的收敛条件可以是当确定出的最大的投资回报率大于预设的回报率阈值,确定收敛。较佳的,所述判断是否满足预设的收敛条件包括:判断所述第二用户新增序列的迭代次数是否达到预设的次数阈值,或判断相邻两次迭代中,确定出的最大的投资回报率的差值是否小于预设的收敛阈值。在本发明实施例中,判断是否满足预设的收敛条件,如果否,将每个第三用户新增序列作为第二用户新增序列,并继续进行针对每个第二用户新增序列,对该第二用户新增序列中的元素进行随机变化,得到该第二用户新增序列对应的第三用户新增序列的步骤,直至满足预设的收敛条件。判断满足预设的收敛条件之后,将最大的投资回报率对应的用户新增序列作为目标用户新增序列。
根据目标用户新增序列进行广告投放分配。具体的,根据预先确定的每个广告投放渠道下用户新增数量与用户单价的对应关系,以及目标用户新增序列,可以确定出每个广告投放渠道的成本,按照确定出的每个广告投放渠道的成本向每个广告投放渠道进行广告投放。
由于在本发明实施例中,获取到第一用户新增序列之后,复制得到预设数量的第二用户新增序列,然后对每个第二用户新增序列中的元素进行随机变化,将变化后得到的第三用户新增序列作为第二用户新增序列,继续进行上述过程,直至满足预设的收敛条件。然后将整个过程中最大的投 资回报率对应的用户新增序列作为目标用户新增序列,根据目标用户新增序列进行广告投放分配。本发明实施例中首先复制得到多个第二用户新增序列,其次对每个第二用户新增序列中的元素进行随机变化,通过上述操作避免陷入短期内局部最优的问题,使得最终的广告投放分配更加合理。
实施例2:
为了进一步增加用户新增序列的随机性,在上述实施例的基础上,在本发明实施例中,所述针对每个第二用户新增序列,对该第二用户新增序列中的元素进行随机变化,得到该第二用户新增序列对应的第三用户新增序列包括:
针对每个第二用户新增序列,对该第二用户新增序列中的元素进行随机变化,得到该第二用户新增序列对应的第四用户新增序列;
针对每个第四用户新增序列,将该第四用户新增序列和对应的第二用户新增序列中对应渠道的元素按照预设的随机概率进行交换,得到与该第四用户新增序列对应的第五用户新增序列;
将每个第五用户新增序列作为第三用户新增序列。
本发明实施例中,针对每个第二用户新增序列,对该第二用户新增序列中的元素进行随机变化,得到该第二用户新增序列对应的第四用户新增序列。然后针对每个第四用户新增序列中的每个元素,按照预设的随机概率,确定是否将该元素与对应的第二用户新增序中对应的元素进行交换。其中,第二用户新增序列和对应的第二用户新增序列中,同一个广告投放渠道的元素即为对应的元素。预设的随机概率例如可以是40%、50%等。如果预设的随机概率是0.4,针对每个第四用户新增序列中的每个元素,有40%的概率将该元素与对应的第二用户新增序中对应的元素进行交换,60%的概率不将该元素与对应的第二用户新增序中对应的元素进行交换。针对每个第四用户新增序列,将该第四用户新增序列和对应的第二用户新增序列中对应渠道的元素按照预设的随机概率进行交换,得到与该第四用户新增序列对应的第五用户新增序列,然后将每个第五用户新增序列作为 第三用户新增序列。
由于在本发明实施例中,针对每个第二用户新增序列,对该第二用户新增序列中的元素进行随机变化,得到该第二用户新增序列对应的第四用户新增序列;针对每个第四用户新增序列,将该第四用户新增序列和对应的第二用户新增序列中对应渠道的元素按照预设的随机概率进行交换,得到与该第四用户新增序列对应的第五用户新增序列;将每个第五用户新增序列作为第三用户新增序列。从而进一步增加用户新增序列的随机性,也就更好的避免了陷入短期内局部最优的问题,使得最终的广告投放分配更加合理。
实施例3:
为了进一步使得最终的广告投放分配更加合理,在上述各实施例的基础上,在本发明实施例中,引入了遗传学中“优胜劣汰,适者生存”的思想,所述得到该第二用户新增序列对应的第四用户新增序列之后,针对每个第四用户新增序列,将该第四用户新增序列和对应的第二用户新增序列中对应渠道的元素按照预设的随机概率进行交换之前,所述方法还包括:
根据每个第四用户新增序列的投资回报率确定所述每个第四用户新增序列被选择的概率,其中,投资回报率越大的第四用户新增序列被选择的概率越大;
根据所述每个第四用户新增序列被选择的概率,在所述每个第四用户新增序列中进行预设次数的选择,采用每次选中的第四用户新增序列,进行后续针对每个第四用户新增序列,将该第四用户新增序列和对应的第二用户新增序列中对应渠道的元素按照预设的随机概率进行交换的步骤。
本发明实施例中,电子设备确定出每个第二用户新增序列对应的第四用户新增序列之后,根据预先确定的所述每个广告投放渠道下用户新增数量与用户单价的对应关系、用户新增数量与用户单体收益的对应关系,确定每个第四用户新增序列的投资回报率。根据每个第四用户新增序列的投资回报率确定每个第四用户新增序列被选择的概率,其中,投资回报率越 大的第四用户新增序列被选择的概率越大。
根据每个第四用户新增序列被选择的概率,在每个第四用户新增序列中进行预设次数的选择。其中,预设次数和预设数量可以相同也可以不同。需要说明的是,每次选择不是从每个第四用户新增序列中取出,而是从每个第四用户新增序列中复制。也就是每次选择出的第四用户新增序列,在下一次仍然有被选择的机会。因为投资回报率越大的第四用户新增序列被选择的概率越大,所以投资回报率越大的第四用户新增序列越容易被选中,然后采用每次选中的第四用户新增序列,进行后续针对每个第四用户新增序列,将该第四用户新增序列和对应的第二用户新增序列中对应渠道的元素按照预设的随机概率进行交换的步骤。投资回报率越大的第四用户新增序列被选择的概率越大,这也就是遗传学中“优胜劣汰,适者生存”思想的应用,也进一步使得最终的广告投放分配更加合理。
实施例4:
为了进一步增加用户新增序列的随机性,在上述实施例的基础上,在本发明实施例中,所述得到与该第四用户新增序列对应的第五用户新增序列之后,将每个第五用户新增序列作为第三用户新增序列之前,所述方法还包括:
针对每个第五用户新增序列,对该第五用户新增序列中的元素进行随机变化,得到与该第五用户新增序列对应的第六用户新增序列;采用该第六用户新增序列对该第五用户新增序列进行更新。
在本发明实施例中,为了进一步增加用户新增序列的随机性,得到与该第四用户新增序列对应的第五用户新增序列之后,再将每个第五用户新增序列中的元素进行随机变化,得到与该第五用户新增序列对应的第六用户新增序列。对第五用户新增序列中的元素进行随机变化的过程与上述对第二用户新增序列中的元素进行随机变化的过程类似,在此不再进行赘述。针对每个第五用户新增序列,对该第五用户新增序列中的元素进行随机变化,得到与该第五用户新增序列对应的第六用户新增序列之后,采用该第 六用户新增序列对该第五用户新增序列进行更新。然后再进行后续将每个第五用户新增序列作为第三用户新增序列的步骤。
由于在本发明实施例中,得到每个第五用户新增序列之后,对第五用户新增序列中的元素进行随机变化,因此进一步增加了用户新增序列的随机性,从而进一步避免了陷入短期内局部最优的问题,使得最终的广告投放分配更加合理。
实施例5:
在上述各实施例的基础上,在本发明实施例中,每当确定出用户新增序列之后,所述方法还包括:
根据预先确定的用户新增数量与用户单价的对应关系确定所述用户新增序列的成本,根据预先确定的用户新增数量与用户单体收益的对应关系,确定所述用户新增序列的收益;
当所述用户新增序列的收益不小于预设的收益阈值,所述用户新增序列的成本不大于预设的成本阈值,所述用户新增序列的新增用户数量不小于预设的用户数量阈值时,保留所述用户新增序列,否则,删除所述用户新增序列。
电子设备每次通过随机变化或者元素交换的形式确定出用户新增序列之后,首先根据预先设置的限制条件,判断确定出的用户新增序列是否满足要求,如果不满足,则删除用户新增序列,如果满足,则保留用户新增序列。
预先设置的限制条件包括用户新增序列的收益不小于预设的收益阈值、用户新增序列的成本不大于预设的成本阈值以及用户新增序列的新增用户数量不小于预设的用户数量阈值。上述三个判断同时满足才认为用户新增序列满足预先设置的限制条件,否则认为用户新增序列不满足预先设置的限制条件。
具体的,每当确定出用户新增序列之后,根据预先确定的用户新增数量与用户单价的对应关系确定用户新增序列的成本。根据预先确定的用户 新增数量与用户单体收益的对应关系,确定用户新增序列的收益。用户新增序列中各个用户新增数量的和即为用户新增序列的新增用户数量。计算出上述值之后,再判断用户新增序列是否满足预先设置的限制条件。如果不满足,则删除用户新增序列,如果满足,则保留用户新增序列。
需要说明的是,如果删除了用户新增序列,为了保证用户新增序列的数量要求,可以确定出删除的用户新增序列的上一级对应的用户新增序列,该上一级对应的用户新增序列变化得到了该删除的用户新增序列。然后再将该上一级对应的用户新增序列进行随机变化或者元素交换的形式得到用户新增序列,直至得到的用户新增序列满足预先设置的限制条件。这样既保留了满足预先设置的限制条件的用户新增序列,也保证了用户新增序列的数量。
由于在本发明实施例中,每当确定出用户新增序列之后,判断用户新增序列是否满足预先设置的限制条件。如果不满足,则删除用户新增序列,如果满足,则保留用户新增序列。从而避免了在迭代过程中因为不满足预先设置的限制条件的用户新增序列的存在,导致收敛较慢,并且有可能造成最终得到的目标用户新增序列并非是最优用户新增序列的问题。进一步优化了广告投放分配方案。
实施例6:
在上述各实施例的基础上,在本发明实施例中,预先确定所述每个广告投放渠道下用户新增数量与用户单价的对应关系的过程包括:
在预设第二时间长度内,针对所述每个广告投放渠道,获取该广告投放渠道下每天的用户新增数量和用户单价;将所述每天的用户新增数量和用户单价进行曲线拟合,确定该广告投放渠道下用户新增数量与用户单价的对应关系;
预先确定所述每个广告投放渠道下用户新增数量与用户单体收益的对应关系的过程包括:
在预设第二时间长度内,针对所述每个广告投放渠道,获取该广告投 放渠道下每天的用户新增数量和用户单体收益;将所述每天的用户新增数量和用户单体收益进行曲线拟合,确定该广告投放渠道下用户新增数量与用户单体收益的对应关系。
电子设备预先确定每个广告投放渠道下用户新增数量与用户单价的对应关系时,在预设第二时间长度内,针对每个广告投放渠道,获取该广告投放渠道下每天的用户新增数量和用户单价。预设第二时间长度可以是最近一年、最近6个月等。具体的,针对每个广告投放渠道,可以获取该广告投放渠道下每天的用户新增数量,另外,通过获取每天该广告投放渠道投资的成本和每天的用户新增数量,可以确定每天的用户单价。然后将每天的用户新增数量和用户单价进行曲线拟合,确定该广告投放渠道下用户新增数量与用户单价的对应关系。
电子设备预先确定每个广告投放渠道下用户新增数量与用户单体收益的对应关系时,在预设第二时间长度内,针对每个广告投放渠道,获取该广告投放渠道下每天的用户新增数量和用户单体收益。具体的,针对每个广告投放渠道,可以获取该广告投放渠道下每天的用户新增数量,另外,通过获取每天该广告投放渠道获得的收益和每天的用户新增数量,可以确定每天的用户单体收益。然后将每天的用户新增数量和用户单体收益进行曲线拟合,确定该广告投放渠道下用户新增数量与用户单体收益的对应关系。
其中,每个广告投放渠道下用户新增数量与用户单价的对应关系以及每个广告投放渠道下用户新增数量与用户单体收益的对应关系可以是一种一元二次方程的对应关系。通过曲线拟合得到一元二次方程中的每个参数,也就得到的对应关系。
图2为本发明实施例提供的广告投放分配流程图,首先用户预先设置的限制条件,限制条件包括预设的收益阈值、预设的成本阈值和预设的用户数量阈值DAU。在预设第二时间长度内,针对每个广告投放渠道,获取该广告投放渠道下每天的用户新增数量、用户单价和用户单体收益,通过 曲线拟合,确定该广告投放渠道下用户新增数量与用户单价CPI的对应关系、用户新增数量与用户单体收益LTV的对应关系。获取预设第一时间长度内的第一用户新增序列,将第一用户新增序列复制得到预设数量的第二用户新增序列,对每个第二用户新增序列中的元素进行随机变化,得到对应的第三用户新增序列。判断是否满足预设的收敛条件,如果否,将每个第三用户新增序列作为第二用户新增序列;如果是,将最大的投资回报率ROI对应的用户新增序列作为目标用户新增序列;根据所述目标用户新增序列进行广告投放分配。其中,DAU:Daily Active User,意为日活跃用户数量,本发明实施例中表示为每天用户新增数量。CPI:平均每个买量新用户的单价。LTV:平均每个买量新用户给产品带来的收益,如消费,充值。ROI:投资回报率,通过投资而应返回的价值的比率,即收益与成本的比值。买量意为通过广告投放的方式进行推广,增加的使用产品的用户新增数量。
本发明实施例中,将预设第一时间长度内每个广告投放渠道对应的用户新增数量看成一个第一用户新增序列,将第一用户新增序列复制得到预设数量的第二用户新增序列,让不同的第二用户新增序列随机变化。随机变化主要分为选择和繁殖,其中繁殖包括交配和突变。选择是根据用户新增序列的投资回报率相关的,投资回报率越大的用户新增序列被选择的概率越大。渐渐的投资回报率越大的用户新增序列会被在迭代中遗传,而投资回报率越小的用户新增序列将消失在迭代的过程中。被选择的用户新增序列将进入交配过程。交配过程是被选择的用户新增序列与之前的用户新增序列有一定概率可以进行互相交换各自的元素,从而产生新的用户新增序列,代替原来的“老”用户新增序列。交配完后,新的用户新增序列中的元素也会有一定概率发生突变,通过突变产生最终的新用户新增序列。图3为本发明实施例提供用户新增序列交配变异示意图,如图3所示,A用户新增序列为[A1 A2 A3 A4 …… An],B用户新增序列为[B1 B2 B3 B4 …… Bn],A用户新增序列和B用户新增序列交配得到子用户新增序 列[A1 B2 A3 B4 …… Bn]。子用户新增序列进行变异,得到最终的用户新增序列[A1 B2+rand() A3-rand() B4 …… Bn]。此过程类似于是生物学的繁衍过程,后代不断从父辈获取优良基因,再自身通过变化达到调整的效果。
在迭代的过程中,每个用户新增序列需要满足以下限制条件,以保证优化的方向必须满足需求。
条件1:
Figure PCTCN2020142288-appb-000001
Figure PCTCN2020142288-appb-000002
条件2:
Figure PCTCN2020142288-appb-000003
Figure PCTCN2020142288-appb-000004
条件3:
Figure PCTCN2020142288-appb-000005
Dau_predict(Ins i)是用户新增序列每个广告投放渠道下的新增用户数量。其中,用户新增数量Ins i的变化会对应影响用户单价Cpi i和用户单体收益Ltv i,从而影响用户新增序列的成本和收益,Cpi_predict(Ins i)和Ltv_predict(Ins i)是根据每个广告投放渠道下用户新增数量与用户单价的对应关系、每个广告投放渠道下用户新增数量与用户单体收益的对应关系确定出的用户单价和用户单体收益。
实施例7:
图4为本发明实施例提供的广告投放分配装置结构示意图,该装置包括:
获取模块41,用于获取预设第一时间长度内的第一用户新增序列,其中,所述第一用户新增序列中的元素为每个广告投放渠道对应的用户新增数量;
第一确定模块42,用于将所述第一用户新增序列复制得到预设数量的第二用户新增序列,针对每个第二用户新增序列,对该第二用户新增序列中的元素进行随机变化,得到该第二用户新增序列对应的第三用户新增序列;
第二确定模块43,用于根据预先确定的所述每个广告投放渠道下用户新增数量与用户单价的对应关系、用户新增数量与用户单体收益的对应关系,确定用户新增序列的投资回报率,判断是否满足预设的收敛条件,如果否,将每个第三用户新增序列作为第二用户新增序列;如果是,将最大的投资回报率对应的用户新增序列作为目标用户新增序列;根据所述目标用户新增序列进行广告投放分配。
所述第一确定模块42,具体用于针对每个第二用户新增序列,对该第二用户新增序列中的元素进行随机变化,得到该第二用户新增序列对应的第四用户新增序列;针对每个第四用户新增序列,将该第四用户新增序列和对应的第二用户新增序列中对应渠道的元素按照预设的随机概率进行交换,得到与该第四用户新增序列对应的第五用户新增序列;将每个第五用户新增序列作为第三用户新增序列。
所述装置还包括:
选择模块44,用于根据每个第四用户新增序列的投资回报率确定所述每个第四用户新增序列被选择的概率,其中,投资回报率越大的第四用户新增序列被选择的概率越大;根据所述每个第四用户新增序列被选择的概率,在所述每个第四用户新增序列中进行预设次数的选择,采用每次选中的第四用户新增序列,触发所述第一确定模块42。
所述装置还包括:
更新模块45,用于针对每个第五用户新增序列,对该第五用户新增序列中的元素进行随机变化,得到与该第五用户新增序列对应的第六用户新增序列;采用该第六用户新增序列对该第五用户新增序列进行更新。
所述装置还包括:
删除模块46,用于根据预先确定的用户新增数量与用户单价的对应关系确定所述用户新增序列的成本,根据预先确定的用户新增数量与用户单体收益的对应关系,确定所述用户新增序列的收益;当所述用户新增序列的收益不小于预设的收益阈值,所述用户新增序列的成本不大于预设的成本阈值,所述用户新增序列的新增用户数量不小于预设的用户数量阈值时,保留所述用户新增序列,否则,删除所述用户新增序列。
所述第二确定模块43,具体用于判断所述第二用户新增序列的迭代次数是否达到预设的次数阈值,或判断相邻两次迭代中,确定出的最大的投资回报率的差值是否小于预设的收敛阈值。
所述装置还包括:
第三确定模块47,用于在预设第二时间长度内,针对所述每个广告投放渠道,获取该广告投放渠道下每天的用户新增数量和用户单价;将所述每天的用户新增数量和用户单价进行曲线拟合,确定该广告投放渠道下用户新增数量与用户单价的对应关系;
所述第三确定模块47,还用于在预设第二时间长度内,针对所述每个广告投放渠道,获取该广告投放渠道下每天的用户新增数量和用户单体收益;将所述每天的用户新增数量和用户单体收益进行曲线拟合,确定该广告投放渠道下用户新增数量与用户单体收益的对应关系。
实施例8:
在上述各实施例的基础上,本发明实施例中还提供了一种电子设备,如图5所示,包括:处理器301、通信接口302、存储器303和通信总线304,其中,处理器301,通信接口302,存储器303通过通信总线304完成相互间的通信;
所述存储器303中存储有计算机程序,当所述程序被所述处理器301执行如下步骤:首先获取预设第一时间长度内的第一用户新增序列,经过复制得到预设数量的第二用户新增序列,对每个第二用户新增序列中的元素进行随机变化得到与第二用户新增序列对应的第三用户新增序列。将每 个第三用户新增序列作为第二用户新增序列,重复上述随机变化的操作,直至满足预设的收敛条件。然后将最大的投资回报率对应的用户新增序列作为目标用户新增序列;根据所述目标用户新增序列进行广告投放分配。
本发明实施例提供的一种电子设备可用于执行上述任意实施例提供的广告投放分配方法,具备相应的功能和有益效果。
实施例9:
在上述各实施例的基础上,本发明实施例还提供了一种计算机存储可读存储介质,所述计算机可读存储介质内存储有可由电子设备执行的计算机程序,当所述程序在所述电子设备上运行时,使得所述电子设备执行如下步骤:首先获取预设第一时间长度内的第一用户新增序列,经过复制得到预设数量的第二用户新增序列,对每个第二用户新增序列中的元素进行随机变化得到与第二用户新增序列对应的第三用户新增序列。将每个第三用户新增序列作为第二用户新增序列,重复上述随机变化的操作,直至满足预设的收敛条件。然后将最大的投资回报率对应的用户新增序列作为目标用户新增序列;根据所述目标用户新增序列进行广告投放分配。
本发明实施例还提供的计算机存储可读存储介质,内存储有可由电子设备执行的计算机程序,当所述程序在所述电子设备上运行时,可用于执行上述任意实施例提供的广告投放分配方法,具备相应的功能和有益效果。
显然,本领域的技术人员可以对本发明进行各种改动和变型而不脱离本发明的精神和范围。这样,倘若本发明的这些修改和变型属于本发明权利要求及其等同技术的范围之内,则本发明也意图包含这些改动和变型在内。

Claims (10)

  1. 一种广告投放分配方法,其特征在于,所述方法包括:
    获取预设第一时间长度内的第一用户新增序列,其中,所述第一用户新增序列中的元素为每个广告投放渠道对应的用户新增数量;
    将所述第一用户新增序列复制得到预设数量的第二用户新增序列,针对每个第二用户新增序列,对该第二用户新增序列中的元素进行随机变化,得到该第二用户新增序列对应的第三用户新增序列;
    根据预先确定的所述每个广告投放渠道下用户新增数量与用户单价的对应关系、用户新增数量与用户单体收益的对应关系,确定用户新增序列的投资回报率,判断是否满足预设的收敛条件,如果否,将每个第三用户新增序列作为第二用户新增序列;如果是,将最大的投资回报率对应的用户新增序列作为目标用户新增序列;根据所述目标用户新增序列进行广告投放分配。
  2. 如权利要求1所述的方法,其特征在于,所述针对每个第二用户新增序列,对该第二用户新增序列中的元素进行随机变化,得到该第二用户新增序列对应的第三用户新增序列包括:
    针对每个第二用户新增序列,对该第二用户新增序列中的元素进行随机变化,得到该第二用户新增序列对应的第四用户新增序列;
    针对每个第四用户新增序列,将该第四用户新增序列和对应的第二用户新增序列中对应渠道的元素按照预设的随机概率进行交换,得到与该第四用户新增序列对应的第五用户新增序列;
    将每个第五用户新增序列作为第三用户新增序列。
  3. 如权利要求2所述的方法,其特征在于,所述得到该第二用户新增序列对应的第四用户新增序列之后,针对每个第四用户新增序列,将该第四用户新增序列和对应的第二用户新增序列中对应渠道的元素按照预设的随机概率进行交换之前,所述方法还包括:
    根据每个第四用户新增序列的投资回报率确定所述每个第四用户新增序列被选择的概率,其中,投资回报率越大的第四用户新增序列被选择的概率越大;
    根据所述每个第四用户新增序列被选择的概率,在所述每个第四用户新增序列中进行预设次数的选择,采用每次选中的第四用户新增序列,进行后续针对每个第四用户新增序列,将该第四用户新增序列和对应的第二用户新增序列中对应渠道的元素按照预设的随机概率进行交换的步骤。
  4. 如权利要求2所述的方法,其特征在于,所述得到与该第四用户新增序列对应的第五用户新增序列之后,将每个第五用户新增序列作为第三用户新增序列之前,所述方法还包括:
    针对每个第五用户新增序列,对该第五用户新增序列中的元素进行随机变化,得到与该第五用户新增序列对应的第六用户新增序列;采用该第六用户新增序列对该第五用户新增序列进行更新。
  5. 如权利要求1至4任一项所述的方法,其特征在于,每当确定出用户新增序列之后,所述方法还包括:
    根据预先确定的用户新增数量与用户单价的对应关系确定所述用户新增序列的成本,根据预先确定的用户新增数量与用户单体收益的对应关系,确定所述用户新增序列的收益;
    当所述用户新增序列的收益不小于预设的收益阈值,所述用户新增序列的成本不大于预设的成本阈值,所述用户新增序列的新增用户数量不小于预设的用户数量阈值时,保留所述用户新增序列,否则,删除所述用户新增序列。
  6. 如权利要求1所述的方法,其特征在于,所述判断是否满足预设的收敛条件包括:
    判断所述第二用户新增序列的迭代次数是否达到预设的次数阈值,或判断相邻两次迭代中,确定出的最大的投资回报率的差值是否小于预设的收敛阈值。
  7. 如权利要求1所述的方法,其特征在于,预先确定所述每个广告投放渠道下用户新增数量与用户单价的对应关系的过程包括:
    在预设第二时间长度内,针对所述每个广告投放渠道,获取该广告投放渠道下每天的用户新增数量和用户单价;将所述每天的用户新增数量和用户单价进行曲线拟合,确定该广告投放渠道下用户新增数量与用户单价的对应关系;
    预先确定所述每个广告投放渠道下用户新增数量与用户单体收益的对应关系的过程包括:
    在预设第二时间长度内,针对所述每个广告投放渠道,获取该广告投放渠道下每天的用户新增数量和用户单体收益;将所述每天的用户新增数量和用户单体收益进行曲线拟合,确定该广告投放渠道下用户新增数量与用户单体收益的对应关系。
  8. 一种广告投放分配装置,其特征在于,所述装置包括:
    获取模块,用于获取预设第一时间长度内的第一用户新增序列,其中,所述第一用户新增序列中的元素为每个广告投放渠道对应的用户新增数量;
    第一确定模块,用于将所述第一用户新增序列复制得到预设数量的第二用户新增序列,针对每个第二用户新增序列,对该第二用户新增序列中的元素进行随机变化,得到该第二用户新增序列对应的第三用户新增序列;
    第二确定模块,用于根据预先确定的所述每个广告投放渠道下用户新增数量与用户单价的对应关系、用户新增数量与用户单体收益的对应关系,确定用户新增序列的投资回报率,判断是否满足预设的收敛条件,如果否,将每个第三用户新增序列作为第二用户新增序列;如果是,将最大的投资回报率对应的用户新增序列作为目标用户新增序列;根据所述目标用户新增序列进行广告投放分配。
  9. 一种电子设备,其特征在于,包括处理器、通信接口、存储器和通信总线,其中,处理器,通信接口,存储器通过通信总线完成相互间的通信;
    存储器,用于存放计算机程序;
    处理器,用于执行存储器上所存放的程序时,实现权利要求1-7任一项所述的方法步骤。
  10. 一种计算机可读存储介质,其特征在于,所述计算机可读存储介质内存储有计算机程序,所述计算机程序被处理器执行时实现权利要求1-7任一项所述的方法步骤。
PCT/CN2020/142288 2020-12-31 2020-12-31 一种广告投放分配方法、装置、电子设备及存储介质 WO2022141456A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/CN2020/142288 WO2022141456A1 (zh) 2020-12-31 2020-12-31 一种广告投放分配方法、装置、电子设备及存储介质
CN202080003840.7A CN115066702A (zh) 2020-12-31 2020-12-31 一种广告投放分配方法、装置、电子设备及存储介质

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2020/142288 WO2022141456A1 (zh) 2020-12-31 2020-12-31 一种广告投放分配方法、装置、电子设备及存储介质

Publications (1)

Publication Number Publication Date
WO2022141456A1 true WO2022141456A1 (zh) 2022-07-07

Family

ID=82258924

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/142288 WO2022141456A1 (zh) 2020-12-31 2020-12-31 一种广告投放分配方法、装置、电子设备及存储介质

Country Status (2)

Country Link
CN (1) CN115066702A (zh)
WO (1) WO2022141456A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118014661A (zh) * 2024-02-02 2024-05-10 广州多游信息技术有限公司 一种基于大数据的广告投放管理系统

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101980477A (zh) * 2010-10-09 2011-02-23 北京星网锐捷网络技术有限公司 检测影子用户的数目的方法和装置及网络设备
CN102708496A (zh) * 2012-01-09 2012-10-03 合一网络技术(北京)有限公司 一种自适应视频广告投放方法及装置
CN110991875A (zh) * 2019-11-29 2020-04-10 广州市百果园信息技术有限公司 一种平台用户质量评估系统
US20200219129A1 (en) * 2019-01-07 2020-07-09 Alphonso Inc. Bidding Agent with Optimized Reach Limitation
CN112036963A (zh) * 2020-09-24 2020-12-04 深圳市万佳安物联科技股份有限公司 一种基于多层随机隐特征模型的网页广告投放装置和方法
CN112150219A (zh) * 2020-10-22 2020-12-29 广州万唯邑众信息科技有限公司 一种在线广告投放系统及其使用方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101980477A (zh) * 2010-10-09 2011-02-23 北京星网锐捷网络技术有限公司 检测影子用户的数目的方法和装置及网络设备
CN102708496A (zh) * 2012-01-09 2012-10-03 合一网络技术(北京)有限公司 一种自适应视频广告投放方法及装置
US20200219129A1 (en) * 2019-01-07 2020-07-09 Alphonso Inc. Bidding Agent with Optimized Reach Limitation
CN110991875A (zh) * 2019-11-29 2020-04-10 广州市百果园信息技术有限公司 一种平台用户质量评估系统
CN112036963A (zh) * 2020-09-24 2020-12-04 深圳市万佳安物联科技股份有限公司 一种基于多层随机隐特征模型的网页广告投放装置和方法
CN112150219A (zh) * 2020-10-22 2020-12-29 广州万唯邑众信息科技有限公司 一种在线广告投放系统及其使用方法

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118014661A (zh) * 2024-02-02 2024-05-10 广州多游信息技术有限公司 一种基于大数据的广告投放管理系统

Also Published As

Publication number Publication date
CN115066702A (zh) 2022-09-16

Similar Documents

Publication Publication Date Title
CN104299151B (zh) 网络广告流量分配系统和方法
CN108574706B (zh) 媒体文件的推送方法及装置
WO2022141456A1 (zh) 一种广告投放分配方法、装置、电子设备及存储介质
CN106296247B (zh) 网络信息资源在线排序方法和装置
CN111582947A (zh) 一种优惠券处理的方法及相关装置
US20140337304A1 (en) Application retention metrics
Durgadevi et al. Resource allocation in cloud computing using SFLA and cuckoo search hybridization
CN113256345B (zh) 广告投放策略的自定义方法、装置和计算机设备
US20110022460A1 (en) Explicit online advertising exposure terms
CN109976901A (zh) 一种资源调度方法、装置、服务器及可读存储介质
CN109087136B (zh) 一种广告资源值的调整方法及装置
CN110334099A (zh) 基于数据库的表单计算方法和系统
US20120030036A1 (en) Booking and management of inventory atoms in content delivery systems
CN112395552A (zh) 数据处理方法以及装置
CN112631612B (zh) 一种基于遗传算法的kubernetes云平台配置的优化方法
Zheng et al. Understanding rollover data
CN108563504A (zh) 一种资源管控方法和装置
CN111311310A (zh) 广告订单推送方法及装置、存储介质及电子装置
CN115115144B (zh) 考虑源荷的电力市场分区和出清方法、装置、设备和介质
Choi et al. Optimizing reserve prices in display advertising auctions
KR102486997B1 (ko) 전문가 매칭 플랫폼에 대한 운영 방법 및 시스템
CN110851716B (zh) 基于最大和谐度的个性化群体推荐方法
CN114240416A (zh) 数据处理方法、装置、计算机设备和存储介质
CN111882359A (zh) 一种广告投放方法和装置
CN108062605A (zh) 电能按质定价方法及装置

Legal Events

Date Code Title Description
NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20967795

Country of ref document: EP

Kind code of ref document: A1