Method, device, and program for big data-based advertising optimization

Abstract

Disclosed is a big data-based advertising optimization method according to various embodiments of the present invention. The method comprises a step of predicting advertising efficiency for each of a plurality of advertising media on the basis of big data and a step of obtaining an advertising optimization result based on the predicted advertising efficiency and advertising condition information, wherein the step of predicting advertising efficiency may comprise the steps of: predicting first advertising efficiency by using a first model that predicts advertising efficiency for a specific week on the basis of historical data corresponding to the specific week in the big data; predicting second advertising efficiency by using a second model that predicts advertising efficiency for the specific week by setting the week as a binary categorical variable in the big data; and predicting the advertising efficiency by using one, of the first model or second model, which has relatively higher prediction accuracy.

Description

Big data-based advertising optimization methods, devices, and programs

[0001] The present invention relates to a method, device, and program for optimizing advertising based on big data, and more particularly, to a method, device, and program for optimizing and distributing an advertising budget based on big data.

[0002]

[0003] The global digital advertising market is estimated to be worth hundreds of billions of dollars and is driven by advertising across various digital media platforms. Major digital advertising platforms (e.g., Google, Naver, Meta, etc.) offer fully automated ad planning, execution, and evaluation processes on their platforms. This allows marketers to take a quantitative approach to maximize the utility of the resulting data.

[0004] Specifically, for each medium, the total marketing budget, number of visits, conversions, and revenue generated over a given period are provided through the advertising media platform. From an advertiser's perspective, the low tracking costs of the digital economy allow for tracking individual user behavioral data, rather than aggregated statistics.

[0005] However, despite the advantages of the digital environment, advertisers still face the challenge of lacking a single advertising medium that provides a unified perspective on advertising performance, including that of competitors. For example, while certain digital advertising mediums evaluate their own advertising, they do not assess the performance of ads run on other media. Consequently, advertisers with advertising budgets distributed across multiple digital advertising mediums face difficulties obtaining a unified assessment of advertising performance across each medium.

[0006] Typically, when conducting digital marketing, advertisers do not rely solely on a single advertising medium, but rather run advertisements across multiple advertising media available in the market, and there is a need to efficiently distribute advertising budgets across each advertising medium.

[0007] Therefore, considering the evaluation scope currently offered by advertising media and the needs of advertisers, it is expected that there will be a demand in the industry for methods to optimize advertising. In this regard, Republic of Korea Patent No. 10-0573410 discloses a method and system for allocating advertising budgets by media in online advertising.

[0008]

[0009] The present invention has been devised in response to the aforementioned background technology and aims to provide a method, device, and program for optimizing big data-based advertising.

[0010] The technical problems of the present invention are not limited to the technical problems mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art from the description below.

[0011]

[0012] According to one embodiment of the present invention for solving the aforementioned problem, a big data-based advertising optimization method is disclosed. The method includes: a step of predicting advertising efficiency for each of a plurality of advertising media based on big data; and a step of obtaining an advertising optimization result based on the predicted advertising efficiency and advertising condition information; wherein the step of predicting advertising efficiency may include a step of predicting a first advertising efficiency through a first model that predicts the advertising efficiency of a specific parking lot based on past data corresponding to the specific parking lot in the big data; a step of predicting a second advertising efficiency through a second model that predicts the advertising efficiency of the specific parking lot by setting the parking lot in the big data as a categorical variable in the form of a binary number; and a step of predicting the advertising efficiency using any one model having a relatively high prediction accuracy among the first model and the second model.

[0013] In an alternative embodiment, the method may further include the steps of: subsampling advertising efficiency data corresponding to each parking lot from the big data; generating learning data corresponding to advertising efficiency for each parking lot based on the subsampled data; and generating a first model that predicts advertising efficiency for each parking lot for each of the plurality of advertising media using the learning data.

[0014] In an alternative embodiment, the step of generating the training data may include: generating a plurality of training data by assigning a weight corresponding to recency to the sub-sampled data; or generating a plurality of training data by differently setting a past period parameter in the sub-sampled data; and the step of predicting the first advertising efficiency through the first model may include: generating a plurality of preliminary models that predict the advertising efficiency based on the plurality of training data; comparing the plurality of advertising efficiencies predicted by each of the plurality of preliminary models with actual advertising efficiencies to identify the first model having the highest prediction accuracy; and predicting the advertising efficiency for each week for each of the plurality of advertising media using the first model.

[0015] In an alternative embodiment, the method may further include: obtaining parking characteristic data by setting parking as a categorical variable in the form of a binary number in the big data; generating learning data by combining advertising efficiency data related to the parking characteristic data; and generating a second model that predicts advertising efficiency for each parking lot for each of the plurality of advertising media using the learning data.

[0016] In an alternative embodiment, the step of generating the learning data may include: generating a plurality of learning data by assigning a weight corresponding to recency to the parking characteristic data; or generating a plurality of learning data by differently setting a past period parameter in the parking characteristic data; and the step of predicting the second advertising efficiency through the second model may include: generating a plurality of preliminary models that predict advertising efficiency based on the plurality of learning data; comparing the plurality of advertising efficiencies predicted by each of the plurality of preliminary models with actual advertising efficiency to identify the second model having the highest prediction accuracy; and predicting advertising efficiency for each parking lot for each of the plurality of advertising media using the second model.

[0017] In an alternative embodiment, the method may further include: a step of recognizing the number of times a holiday is included in each week in the big data; a step of updating the learning data by reflecting the number of times a holiday is included in each week in the learning data; and a step of generating the first model or the second model using the updated learning data to predict advertising efficiency according to the number of times a holiday is included in each week.

[0018] In an alternative embodiment, the step of predicting the advertising efficiency may include: a step of dividing data from the big data into a plurality of time intervals from a specific advertising execution time point in each of the plurality of advertising media; a step of repeatedly training a model for predicting advertising efficiency using the data divided into the plurality of time intervals; a step of comparing advertising performance for each of the plurality of time intervals included in output data of the repeatedly trained model to recognize a delay effect occurring in each of the plurality of advertising media; and a step of predicting advertising efficiency for a specific time interval after advertising execution for each of the plurality of advertising media by reflecting the delay effect.

[0019] In an alternative embodiment, the advertising condition information may include information about an advertising budget and information about an advertising objective, and the step of obtaining an advertising optimization result may include: determining at least one advertising medium to run an advertisement among the plurality of advertising mediums based on the predicted advertising efficiency, the advertising budget, and the advertising objective; and determining an advertising cost distribution ratio for the at least one advertising medium based on the predicted advertising efficiency and the advertising budget.

[0020] According to one embodiment of the present invention for solving the above-described problem, a device is disclosed. The device includes: a memory storing one or more instructions; and a processor executing the one or more instructions stored in the memory, wherein the processor can perform the above-described methods by executing the one or more instructions.

[0021] According to one embodiment of the present invention for solving the above-described problem, a computer program stored in a computer-readable recording medium is disclosed, which is combined with a computer as hardware and can perform the above-described methods.

[0022] Other specific details of the present invention are included in the detailed description and drawings.

[0023]

[0024] The present invention maximizes advertising effectiveness through a big data-based advertising optimization method. Specifically, the invention reflects predicted advertising effectiveness and advertising condition information to enable optimal advertising budget allocation and channel determination for each advertising medium. Furthermore, it considers variations in advertising performance, such as the delay effect, to develop more precise advertising strategies. This allows for efficient use of advertising budgets and maximization of target performance.

[0025] The effects of the present invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description below.

[0026]

[0027] FIG. 1 is a diagram illustrating a system according to one embodiment of the present invention.

[0028] Figure 2 is a hardware configuration diagram of a computing device according to one embodiment of the present invention.

[0029] FIG. 3 and FIG. 4 are diagrams for explaining an example of a big data-based advertising optimization method according to one embodiment of the present invention.

[0030] FIG. 5 is a drawing for explaining an example of a method for generating a first model according to one embodiment of the present invention.

[0031] FIG. 6 is a drawing for explaining an example of a method for generating a second model according to one embodiment of the present invention.

[0032] FIG. 7 is a diagram illustrating an example of a method for predicting advertising efficiency by reflecting a delay effect according to one embodiment of the present invention.

[0033]

[0034] Various embodiments are now described with reference to the drawings. In this specification, various descriptions are provided to facilitate an understanding of the present invention. However, it will be apparent that these embodiments may be practiced without these specific details.

[0035] As used herein, the terms "component," "module," "system," and the like refer to computer-related entities, hardware, firmware, software, a combination of software and hardware, or an execution of software. For example, a component may be, but is not limited to, a procedure running on a processor, a processor, an object, a thread of execution, a program, and / or a computer. For example, both an application running on a computing device and the computing device may be a component. One or more components may reside within a processor and / or a thread of execution. A component may be localized within a single computer. A component may be distributed between two or more computers. Furthermore, these components may execute from various computer-readable media having various data structures stored therein. Components may communicate via local and / or remote processes, for example, by signals comprising one or more data packets (e.g., data from one component interacting with another component in a local system, a distributed system, and / or data transmitted to another system via a network such as the Internet via signals).

[0036] Furthermore, the term "or" is intended to mean an inclusive "or" rather than an exclusive "or." That is, unless otherwise specified or clear from context, "X employs A or B" is intended to mean either of the natural inclusive permutations. That is, if X employs A; X employs B; or X employs both A and B, "X employs A or B" can apply to any of these cases. Furthermore, the term "and / or" as used herein should be understood to refer to and include all possible combinations of one or more of the associated items listed.

[0037] Additionally, the terms "comprises" and / or "comprising" should be understood to imply the presence of the features and / or components in question. However, it should be understood that the terms "comprises" and / or "comprising" do not exclude the presence or addition of one or more other features, components, and / or groups thereof. Furthermore, unless otherwise specified or clear from the context to refer to the singular form, the singular in the specification and claims should generally be construed to mean "one or more."

[0038] Those skilled in the art should further recognize that the various illustrative logical blocks, components, modules, circuits, means, logics, and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate the interchangeability of hardware and software, various illustrative components, blocks, components, means, logics, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application. However, such implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.

[0039] The description of the disclosed embodiments is provided to enable those skilled in the art to make or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art. The general principles defined herein may be applied to other embodiments without departing from the scope of the present invention. Therefore, the present invention is not limited to the disclosed embodiments. The present invention is to be construed in the widest scope consistent with the principles and novel features disclosed herein.

[0040] In this specification, the term "computer" refers to any type of hardware device including at least one processor, and may also be understood to encompass software components operating on the hardware device, depending on the embodiment. For example, the term "computer" may be understood to encompass, but is not limited to, smartphones, tablet PCs, desktops, laptops, and all user clients and applications running on each device.

[0041] Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings.

[0042] Although each step described in this specification is described as being performed by a computer, the subject of each step is not limited thereto, and at least some of each step may be performed by different devices depending on the embodiment.

[0043]

[0044] FIG. 1 is a diagram illustrating a system according to one embodiment of the present invention.

[0045] Referring to FIG. 1, a system according to one embodiment of the present invention may include a computing device (100), a user terminal (200), and an external server (300). The system illustrated in FIG. 1 is according to one embodiment, and its components are not limited to the embodiment illustrated in FIG. 1, and may be added, changed, or deleted as needed.

[0046] In one embodiment, the computing device (100) can perform a big data-based advertising optimization method. Furthermore, the computing device (100) can provide an advertising budget optimization solution based on the advertising optimization method.

[0047] Specifically, the computing device (100) can predict advertising effectiveness for multiple advertising media based on big data. Here, advertising effectiveness corresponds to advertising performance and may include, but is not limited to, clicks, conversion rates, impressions, click-through rates (CTR), and cost-per-acquisition (CPA).

[0048] Additionally, the computing device (100) can obtain advertising optimization results based on predicted advertising effectiveness and advertising condition information for each advertising medium. Furthermore, the computing device (100) can provide the advertising optimization results to the user terminal (200). Here, the advertising condition information may include information about the advertising budget and advertising objectives.

[0049] Therefore, the computing device (100) of the present invention can maximize the advertising efficiency of a user (specifically, an advertiser).

[0050] More specifically, when predicting advertising effectiveness, the computing device (100) can predict the first advertising effectiveness through a first model that predicts the advertising effectiveness of a specific week based on past data corresponding to the specific week in big data. In addition, the computing device (100) can predict the second advertising effectiveness through a second model that predicts the advertising effectiveness of a specific week by setting the week as a binary categorical variable in the big data. In addition, the computing device (100) can predict the advertising effectiveness using either the first model or the second model, whichever model has a relatively high prediction accuracy.

[0051] Accordingly, the computing device (100) of the present invention can increase the accuracy of advertising efficiency prediction and further derive an advertising strategy optimized for each advertising medium.

[0052] Hereinafter, an example of a method in which a computing device (100) performs a big data-based advertising optimization method will be described with reference to FIGS. 3 to 7.

[0053] In one embodiment, the computing device (100) can predict advertising effectiveness for each of a plurality of advertising media based on big data.

[0054] Specifically, when predicting advertising effectiveness, the computing device (100) can remove abnormal values ​​contained in big data. Furthermore, the computing device (100) can generate a plurality of training data sets based on valid data from which abnormal values ​​have been removed. Furthermore, the computing device (100) can generate a plurality of advertising effectiveness prediction models based on the plurality of training data sets. Furthermore, the computing device (100) can compare the advertising effectiveness predicted by each of the plurality of advertising effectiveness prediction models with actual advertising effectiveness to identify a specific model with the highest prediction accuracy. Furthermore, the computing device (100) can predict advertising effectiveness for each advertising medium based on the specific model.

[0055] In one embodiment, when generating multiple training data sets, the computing device (100) may generate multiple training data sets by assigning weights corresponding to the recency of valid data sets. Furthermore, the computing device (100) may generate multiple training data sets by dividing the valid data sets into multiple period parameters.

[0056] In various embodiments, when predicting advertising effectiveness, the computing device (100) may repeatedly train multiple advertising effectiveness prediction models for each advertising medium, thereby recognizing the delay effects of each advertising medium. Here, the advertising effectiveness prediction models may be trained to predict advertising effectiveness by reflecting the delay effects of each advertising medium.

[0057] A computing device (100) can obtain advertising optimization results by inputting predicted advertising effectiveness and advertising condition information for each advertising medium into an optimization model. Here, the advertising condition information may include information about the advertising budget and advertising objectives.

[0058] In one embodiment, previous studies have presented contribution models and performance measurement methods rather than from an advertising budget optimization perspective, limiting their ability to provide concrete solutions to advertisers' desired budget optimization needs. To bridge the gap between academia and the practical market, the present invention provides a method for solving multi-channel budget allocation problems.

[0059] In one embodiment, the present invention applies the same core elements that business practitioners refer to in the budget allocation process when designing the objective function and constraints of an optimization model.

[0060] Specifically, the objective function of the optimization model can be designed to include conversions or revenue as a result of digital advertising execution.

[0061] In one embodiment, the computing device (100) can learn advertising big data, measure efficiency by medium, and derive an optimized budget result that fits the input parameters based on the measured efficiency.

[0062] For example, if the goal of an advertisement is to maximize the number of clicks, the computing device (100) can allocate the budget for each channel in a direction that maximizes the number of clicks.

[0063] In the present invention, channels can be set in various ways, such as search advertisements, display advertisements, mobile, PC, and web, and the computing device (100) can measure the efficiency of each of these various channels.

[0064] When measuring the efficiency of each advertising medium (or channel), the computing device (100) can perform learning based on big data (e.g., customer data).

[0065] Additionally, the computing device (100) can perform outlier removal on big data prior to learning. For example, the computing device (100) can comprehensively review the data and remove outliers that are over or underperforming compared to budget. Thereafter, the computing device (100) can perform learning based on the data from which the outliers have been removed.

[0066] In one example, the advertising field can be greatly influenced by recency due to its nature.

[0067] Reflecting this, the computing device (100) can assign weights to data through a process of determining weights according to recency (i.e., assigning higher weights to recent data), and select an advertising efficiency prediction model with high prediction accuracy through comparison between predicted data and actual data.

[0068] For example, the computing device (100) can repeatedly train an advertising efficiency prediction model using data collected over a period of 1 to 24 months based on advertising-related data over the past two years to determine the influence over each period.

[0069] For example, the computing device (100) can repeatedly and variously divide the period parameters of the data used for learning to train an advertising efficiency prediction model using data collected over the past three months, train an advertising efficiency prediction model using data collected over the past four months, train an advertising efficiency prediction model using data collected over each of the three and four months, and so on, to find a specific model with the highest prediction accuracy. This prediction accuracy may vary depending on the medium or conditions.

[0070] In one example, advertising may have a delayed effect. For example, an advertisement that ran last Saturday may begin to have an impact later this week.

[0071] Reflecting this, the computing device (100) can improve overall prediction performance by performing preprocesses that take into account the characteristics of the advertisement and applying a function appropriate to those characteristics. These characteristics may vary depending on the specific conditions, such as the product, advertisement, media, or company.

[0072] Specifically, the computing device (100) can perform a process of repeatedly learning a model for each channel to determine how the delay effect of each medium appears.

[0073] For example, the computing device (100) can execute an advertisement that is the source of influence on Monday from Monday to Sunday, and analyze the extent of Monday's influence by setting different periods such as Monday, Monday-Tuesday, Monday-Tuesday-Wednesday, etc. In addition, the computing device (100) can perform learning in various combinations, such as dividing 7 days' worth of data into daily units and adding up two days' worth of data during period-based analysis.

[0074] Additionally, the computing device (100) may reflect the day of the week characteristics, including holidays, among the characteristics of the advertisement.

[0075] For example, the computing device (100) can train a model to consider data on how many of the characteristics of days of the week corresponding to holidays are included in the advertising period, such as how many weekend days, such as Saturdays, are included, and how many holidays are included, after the advertisement is executed.

[0076] In one embodiment, accurately analyzing advertising efficiency can be a key technology for optimizing advertising budget allocation. For example, if advertising budget allocation optimization is performed without an accurate analysis of advertising efficiency, it may be difficult to achieve desirable results. In other words, the performance of efficiency prediction techniques can be linked to optimization performance.

[0077] Accordingly, the present invention discloses a method for accurately predicting advertising effectiveness, enabling appropriate optimization of advertising budget distribution. If the analysis of advertising effectiveness is accurate, various optimization functions available in open libraries can be utilized to optimize advertising budget distribution.

[0078] In one embodiment, the big data may include advertising performance data.

[0079] Specifically, advertising performance data can include information about how the budget was allocated, the number of impressions and clicks generated, and the final conversion value, based on information such as advertising costs, the campaign for which the ad was advertised, the media, channel type (search, display, video, etc.), and device information. Here, conversion can refer to whether the intended outcome of the ad was achieved, such as an app installation, a link visit, a purchase, or an addition to a shopping cart. For example, conversion can be interpreted as a purchase conversion, but is not limited thereto.

[0080] In one embodiment, an advertising budget optimization solution may predict and compare advertising performance metrics, informing users that, for example, distributing 50 million won across different channels would be more beneficial for overall performance than spending 100 million won entirely on a specific channel. The optimization algorithm used to derive this information may be an algorithm that identifies inputs (e.g., advertising budget allocation, channel selection, etc.) that maximize the desired values ​​(e.g., advertising efficiency in terms of impressions, clicks, etc.) desired by the user (e.g., advertiser).

[0081] In various embodiments, the computing device (100) may provide web- or application-based services, but is not limited thereto.

[0082] The computing device (100) may include any type of computer system or computer device, such as, but not limited to, a microprocessor, a mainframe computer, a digital processor, a portable device, and a device controller.

[0083] Below, a description of the hardware configuration of the computing device (100) will be provided with reference to FIG. 2.

[0084] Meanwhile, the user terminal (200) can be connected to the computing device (100) via a network (400) and can be a terminal of a user who uses the advertising budget optimization solution provided by the computing device (100).

[0085] Here, the user terminal (200) may include, for example, various types of computer devices. For example, the user terminal (200) may refer to various terminal devices such as a smartphone, tablet PC, desktop, or laptop.

[0086] The user terminal (200) includes a display on at least a portion of the terminal, and may include an operating system for driving an application or extension program-based service provided from the computing device (100). For example, the user terminal (200) may be a smart phone, but is not limited thereto, and the user terminal (200) may include all types of handheld-based wireless communication devices such as navigation, PCS (Personal Communication System), GSM (Global System for Mobile communications), PDC (Personal Digital Cellular), PHS (Personal Handyphone System), PDA (Personal Digital Assistant), IMT (International Mobile Telecommunication)-2000, CDMA (Code Division Multiple Access)-2000, W-CDMA (W-Code Division Multiple Access), Wibro (Wireless Broadband Internet) terminals, smart pads, tablet PCs, etc., as a wireless communication device that ensures portability and mobility.

[0087] An external server (300) can be connected to a computing device (100) via a network (400), and can transmit and receive various information / data necessary for the computing device (100) to perform a big data-based advertising optimization method, and can store and manage various information / data generated as the computing device (100) performs the big data-based advertising optimization method.

[0088] For example, the external server (300) may be a database server that stores information used in a big data-based advertising optimization method. As another example, the external server (300) may be a server that provides information used in a big data-based advertising optimization method.

[0089] A network (400) may refer to a connection structure that enables information exchange between each node, such as a computing device, multiple terminals, and servers. For example, the network (400) includes a local area network (LAN), a wide area network (WAN), the Internet (WWW), a wired and wireless data communication network, a telephone network, a wired and wireless television communication network, etc.

[0090] Wireless data communication networks include, but are not limited to, 3G, 4G, 5G, 3GPP (3rd Generation Partnership Project), 5GPP (5th Generation Partnership Project), LTE (Long Term Evolution), WIMAX (World Interoperability for Microwave Access), Wi-Fi, the Internet, LAN (Local Area Network), Wireless LAN (Wireless Local Area Network), WAN (Wide Area Network), PAN (Personal Area Network), RF (Radio Frequency), Bluetooth network, NFC (Near-Field Communication) network, satellite broadcasting network, analog broadcasting network, and DMB (Digital Multimedia Broadcasting) network.

[0091]

[0092] Figure 2 is a hardware configuration diagram of a computing device according to one embodiment of the present invention.

[0093] Referring to FIG. 2, a computing device (100) according to one embodiment of the present invention may include one or more processors (110), a memory (120) for loading a computer program (151) executed by the processor (110), a bus (130), a communication interface (140), and a storage (150) for storing the computer program (151). Here, only components related to the embodiment of the present invention are illustrated in FIG. 2. Therefore, a person skilled in the art to which the present invention pertains will understand that other general components may be included in addition to the components illustrated in FIG. 2.

[0094] The processor (110) controls the overall operation of each component of the computing device (100). The processor (110) may be configured with one or more cores, and may include a processor for data analysis and deep learning, such as a central processing unit (CPU), a general purpose graphics processing unit (GPGPU), or a tensor processing unit (TPU) of the computing device. Alternatively, the processor (110) may be configured to include any type of processor well known in the technical field of the present invention.

[0095] Additionally, the processor (110) may perform operations for at least one application or program for executing a method according to embodiments of the present invention, and the computing device (100) may have one or more processors.

[0096] In various embodiments, the processor (110) may further include a Random Access Memory (RAM, not shown) and a Read-Only Memory (ROM, not shown) that temporarily and / or permanently store signals (or data) processed within the processor (110). In addition, the processor (110) may be implemented in the form of a system on chip (SoC) that includes at least one of a graphics processing unit, RAM, and ROM.

[0097] The memory (120) stores various data, commands, and / or information. The memory (120) can load a computer program (151) from the storage (150) to execute methods / operations according to various embodiments of the present invention. When the computer program (151) is loaded into the memory (120), the processor (110) can perform the method / operation by executing one or more instructions constituting the computer program (151). The memory (120) may be implemented as a volatile memory such as RAM, but the technical scope of the present invention is not limited thereto.

[0098] The bus (130) provides a communication function between components of the computing device (100). The bus (130) may be implemented as various types of buses, such as an address bus, a data bus, and a control bus.

[0099] The communication interface (140) supports wired and wireless Internet communication of the computing device (100). Furthermore, the communication interface (140) may support various communication methods other than Internet communication. To this end, the communication interface (140) may be configured to include a communication module well known in the technical field of the present invention. In some embodiments, the communication interface (140) may be omitted.

[0100] Storage (150) can non-temporarily store a computer program (151). When performing a process according to an embodiment of the present invention through a computing device (100), storage (150) can perform a method according to the disclosed embodiment or store various information necessary to provide a service.

[0101] Storage (150) may be configured to include non-volatile memory such as ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable ROM), flash memory, a hard disk, a removable disk, or any type of computer-readable recording medium well known in the art to which the present invention pertains.

[0102] The computer program (151) may include one or more instructions that cause the processor (110) to perform a method / operation according to various embodiments of the present invention when loaded into the memory (120). That is, the processor (110) may perform the method / operation according to various embodiments of the present invention by executing the one or more instructions.

[0103] In one embodiment, the computer program (151) may include one or more instructions for performing various methods associated with various tasks related to learning a neural network model.

[0104] The steps of a method or algorithm described in connection with an embodiment of the present invention may be implemented directly in hardware, implemented as a software module executed by hardware, or implemented by a combination thereof. The software module may reside in a random access memory (RAM), a read only memory (ROM), an erasable programmable ROM (EPROM), an electrically erasable programmable ROM (EEPROM), a flash memory, a hard disk, a removable disk, a CD-ROM, or any other form of computer-readable recording medium well known in the art to which the present invention pertains.

[0105] The components of the present invention may be implemented as programs (or applications) and stored on a medium to be executed in conjunction with a computer, which is hardware. The components of the present invention may be implemented as software programs or software elements. Similarly, the embodiments may be implemented in a programming or scripting language such as C, C++, Java, or an assembler, including various algorithms implemented as a combination of data structures, processes, routines, or other programming components. Functional aspects may be implemented as algorithms that are executed on one or more processors.

[0106]

[0107] FIG. 3 and FIG. 4 are diagrams for explaining an example of a big data-based advertising optimization method according to one embodiment of the present invention.

[0108] Referring to FIG. 3, the computing device (100) can predict advertising effectiveness for multiple advertising media based on big data (S100). Here, advertising effectiveness corresponds to advertising performance and may include, but is not limited to, clicks, conversion rates, impressions, click-through rates (CTR), and cost-per-acquisition (CPA).

[0109] Specifically, referring to FIG. 4, when predicting advertising efficiency in step (S100), the computing device (100) can predict the first advertising efficiency through a first model that predicts the advertising efficiency of a specific parking lot based on past data corresponding to a specific parking lot in big data (S101).

[0110] For example, past data from Monday to Sunday of a specific week can be collected, advertising performance during that period can be analyzed, and then advertising efficiency for a specific future week can be predicted. For example, the computing device (100) can collect advertising performance data for two weeks each month and use the first model learned from the data to predict advertising efficiency for the following month. For example, the computing device (100) can collect advertising performance data for the second week from January to November, train the first model, and predict advertising efficiency for the second week of December using the trained first model.

[0111] Through this, the computing device (100) can provide advertisers with predictive information on which advertising mediums are expected to deliver higher advertising effectiveness in a specific future period. Furthermore, this information can help advertisers optimize their budgets, focusing their advertising spending on the most effective times.

[0112] When predicting advertising efficiency in step (S100), the computing device (100) can predict the second advertising efficiency through a second model that predicts the advertising efficiency of a specific parking lot by setting the parking lot as a categorical variable in the form of a binary in big data (S102).

[0113] For example, the computing device (100) can classify parking data into "weekdays" and "weekends" and analyze advertising performance for each category. Furthermore, the computing device (100) can predict advertising effectiveness for each weekday and weekend within a specific future parking lot using a second model learned based on advertising performance classified into weekdays and weekends within a specific parking lot.

[0114] For example, the computing device (100) may analyze data from Monday to Friday by setting it as "weekdays" and data from Saturday and Sunday by setting it as "weekends," and based on this, may predict differences in advertising performance between weekdays and weekends for a specific week. Additionally, the computing device (100) may analyze data from holidays other than weekends by setting it as "holidays," and may predict advertising performance for cases where a holiday is included in a specific week.

[0115] Through this, the computing device (100) can more accurately predict the effectiveness of advertisements concentrated on weekends or on specific days of the week, and optimize advertising effectiveness for each category. Furthermore, advertisers can develop more effective advertising strategies for weekends and holidays.

[0116] When the first advertising efficiency and the second advertising efficiency are predicted in step (S100), the computing device (100) can predict the advertising efficiency using one of the first and second models with relatively high prediction accuracy (S103).

[0117] Specifically, the computing device (100) can compare the results predicted by the first model and the second model with actual advertising performance and select a model with higher accuracy.

[0118] For example, assuming that it is currently November, the computing device (100) can compare the advertising efficiency of November predicted by each of the two models trained with advertising performance data for the period from January to October with the actual advertising efficiency of November, and recognize a specific model with higher prediction accuracy.

[0119] For example, if the advertising effectiveness predicted by the first model based on past data for a specific parking lot is more consistent with actual performance, the computing device (100) may select the first model to predict the advertising effectiveness for that parking lot. Conversely, if the second model provides a more accurate prediction, the computing device (100) may use the second model to predict the advertising effectiveness for that parking lot.

[0120] Through this, the computing device (100) can optimize advertising efficiency based on the most reliable prediction results.

[0121] Referring again to FIG. 3, the computing device (100) can obtain advertising optimization results based on predicted advertising effectiveness and advertising condition information (S200). Here, the advertising condition information may include information about the advertising budget and advertising objectives.

[0122] Specifically, the computing device (100) can determine at least one advertising medium to run an advertisement among a plurality of advertising mediums based on predicted advertising efficiency, advertising budget, and advertising goals.

[0123] For example, the computing device (100) may select a medium among multiple advertising media predicted to exhibit the highest advertising effectiveness to achieve a specific advertising goal. For example, assuming the advertising goal is to maximize conversion rates, the computing device (100) may identify at least one advertising medium predicted to have a relatively high conversion rate and determine that medium as the medium for advertising.

[0124] Additionally, the computing device (100) can determine an advertising cost distribution ratio for at least one advertising medium based on predicted advertising efficiency and advertising budget.

[0125] For example, the computing device (100) may proportionally distribute advertising costs based on the predicted advertising efficiency of each medium to efficiently allocate the advertising budget for the selected advertising medium. For example, if the predicted advertising efficiency of medium A is higher than that of medium B, the computing device (100) may allocate a larger budget to medium A and a relatively smaller budget to medium B, thereby maximizing overall advertising performance.

[0126]

[0127] FIG. 5 is a drawing for explaining an example of a method for generating a first model according to one embodiment of the present invention.

[0128] According to various embodiments of the present invention, the computing device (100) can generate a first model for predicting advertising effectiveness.

[0129] Referring to FIG. 5, the computing device (100) can subsample advertising efficiency data corresponding to each parking lot from big data (S301).

[0130] Specifically, the computing device (100) can extract data from a specific parking lot and reduce (i.e., subsample) it into a subset suitable for analysis. This may be a process of selecting data that meets specific conditions to efficiently process a large amount of data.

[0131] For example, the computing device (100) can subsample data collected during a specific time period or from a specific advertising medium for each week, thereby extracting representative data on the advertising effectiveness of that week. This can improve the model's learning performance by focusing on key data that can significantly impact advertising performance.

[0132] The computing device (100) can generate learning data corresponding to advertising efficiency by parking lot based on the subsampled data (S302).

[0133] When generating training data, the computing device (100) can generate multiple training data by assigning weights corresponding to recency to subsampled data.

[0134] For example, the computing device (100) can generate training data to enable predictions that are more sensitive to recent trends and market changes by assigning a higher weight to the most recent advertising performance data. Specifically, the computing device (100) can assign a weight of 1.5 times to data from the past month, a weight of 1.2 times to data from one to three months ago, and a default weight to data older than three months. This allows the computing device (100) to reflect patterns in advertising performance changes over time.

[0135] Additionally, when generating training data in step (S302), the computing device (100) may generate multiple training data sets by setting different past period parameters in the subsampled data. For example, the computing device (100) may generate training data sets by setting advertising performance data collected over various periods, such as the past one month, three months, and six months, as respective period parameters. Through this, the computing device (100) may enable predictions that take into account both long-term advertising performance patterns and short-term advertising performance changes.

[0136] And, the computing device (100) can use learning data to create a first model that predicts advertising efficiency for each parking lot for each of multiple advertising media (S303).

[0137] Specifically, the computing device (100) inputs the generated training data into a first model, thereby training a model capable of predicting advertising effectiveness for each parking period. In this process, the first model analyzes the performance of advertising media for each parking period based on subsampled data and assigned weights, and based on this analysis, learns the ability to predict advertising effectiveness for future parking periods.

[0138] For example, the computing device (100) can train a first model using advertising data for each week from January to November, and use this model to predict advertising effectiveness for each week in December. This allows advertisers to develop more accurate advertising strategies for the upcoming week.

[0139] Meanwhile, in step S101, when predicting the first advertising efficiency through the first model, the computing device (100) may generate multiple preliminary models for predicting advertising efficiency based on multiple learning data. Furthermore, the computing device (100) may compare the multiple advertising efficiencies predicted by each of the multiple preliminary models with the actual advertising efficiency to identify the first model with the highest prediction accuracy. Furthermore, the computing device (100) may use the first model to predict advertising efficiency for each of the multiple advertising media by week.

[0140] Specifically, the computing device (100) compares the prediction results of each preliminary model with actual advertising performance to analyze the prediction error. In this process, the preliminary model with the lowest error is selected and set as the final first model, thereby enabling prediction of advertising effectiveness for each week.

[0141] For example, the computing device (100) can generate preliminary models using data from 3 months, 6 months, and 12 months, and then evaluate the predictive performance of each preliminary model to select the model that provides the most accurate prediction. For example, the computing device (100) can compare the predicted advertising efficiency of a first preliminary model generated using data from the past 3 months, a second preliminary model generated using data from the past 6 months, and a third preliminary model generated using data from the past 12 months with the actual advertising efficiency, and determine a specific preliminary model with the lowest error as the first model. The first model selected in this way can then be used to predict the advertising efficiency of each advertising medium for each week.

[0142] According to various embodiments of the present invention, when generating the first model, the computing device (100) may consider the number of times a week includes a holiday. Here, a holiday may refer to a statutory holiday such as Children's Day or Christmas.

[0143] Specifically, the computing device (100) can recognize the number of holidays included in each week in the big data. Furthermore, the computing device (100) can update the learning data by reflecting the number of holidays included in each week in the learning data. Furthermore, the computing device (100) can use the updated learning data to create a first model that predicts advertising effectiveness based on the number of holidays included in each week.

[0144] More specifically, the computing device (100) can analyze differences in advertising performance between weeks that include and do not include a holiday, thereby generating learning data that reflects the impact of holidays on advertising effectiveness. This allows the computing device (100) to incorporate patterns of increased or decreased advertising effectiveness during the holiday period into the first model.

[0145] For example, if the computing device (100) analyzes data from the past three years and assumes that the number of advertisement clicks in weeks that include holidays increases by 20% more than the average, the computing device (100) can reflect this data in the learning data to create a first model, thereby increasing the accuracy of advertising efficiency prediction for weeks that include holidays.

[0146]

[0147] FIG. 6 is a drawing for explaining an example of a method for generating a second model according to one embodiment of the present invention.

[0148] According to various embodiments of the present invention, the computing device (100) can generate a second model for predicting advertising effectiveness.

[0149] Referring to FIG. 6, the computing device (100) can obtain parking characteristic data by setting a binary type categorical variable for parking in big data (S401).

[0150] In the present invention, setting the parking number as a categorical variable in binary form may mean, for example, dividing the parking number into 'weekdays' and 'weekends' and classifying the advertising performance of each parking number in binary form.

[0151] That is, the computing device (100) can divide data by weekday and weekend periods from big data and set these as variables to independently analyze advertising performance for each period. This allows the computing device (100) to make predictions that reflect differences in advertising effectiveness between weekdays and weekends, and to determine whether a specific advertising medium is more effective during the week or on weekends.

[0152] The computing device (100) can generate learning data by combining advertising efficiency data related to parking characteristic data (S402).

[0153] When generating learning data, the computing device (100) can generate multiple learning data by assigning weights corresponding to recency to parking characteristic data.

[0154] For example, by assigning a higher weight to the most recent advertising performance data, the computing device (100) can generate training data to enable predictions that are more sensitive to recent trends and market changes. This data, generated in this way, reflects the latest advertising performance on weekdays and weekends, enabling more accurate predictions of advertising effectiveness.

[0155] In addition, when the computing device (100) generates learning data in step (S402), it can generate multiple learning data by setting past period parameters differently in the parking characteristic data.

[0156] For example, the computing device (100) can generate learning data by setting data for each parking period, for example, for 1 month, 3 months, and 6 months, as respective period parameters. This allows the computing device (100) to perform predictions that reflect both short-term changes in advertising effectiveness based on long-term parking characteristics.

[0157] The computing device (100) can use learning data to create a second model that predicts advertising efficiency for each parking lot for each of multiple advertising media (S403).

[0158] Specifically, the computing device (100) inputs the generated training data into a second model, thereby training a model capable of predicting advertising effectiveness for each parking lot. This allows the second model to predict advertising performance for each advertising medium based on parking lot characteristics, such as weekdays and weekends, based on training data related to parking lot characteristic data.

[0159] Meanwhile, in step (S102), when predicting the second advertising efficiency using the second model, the computing device (100) may generate multiple preliminary models for predicting advertising efficiency based on multiple learning data. Furthermore, the computing device (100) may compare the multiple advertising efficiency predicted by each of the multiple preliminary models with the actual advertising efficiency to identify the second model with the highest prediction accuracy. Furthermore, the computing device (100) may use the second model to predict advertising efficiency for each of the multiple advertising media by week.

[0160] According to various embodiments of the present invention, when generating a second model, the computing device (100) may consider the number of times a parking lot includes a holiday.

[0161] Specifically, the computing device (100) can recognize the number of holidays included in each week in the big data. Furthermore, the computing device (100) can update the learning data by reflecting the number of holidays included in each week in the learning data. Furthermore, the computing device (100) can use the updated learning data to create a second model that predicts advertising effectiveness based on the number of holidays included in each week.

[0162] More specifically, the computing device (100) can compare and analyze advertising performance between weeks that include and do not include a holiday, thereby generating learning data that reflects the impact of holidays on advertising performance. This allows the computing device (100) to build a model capable of more accurately predicting advertising effectiveness during the holiday period.

[0163]

[0164] FIG. 7 is a diagram illustrating an example of a method for predicting advertising efficiency by reflecting a delay effect according to one embodiment of the present invention.

[0165] According to various embodiments of the present invention, the computing device (100) may reflect the delay effect when predicting advertising efficiency in step (S100).

[0166] Referring to FIG. 7, the computing device (100) can divide the data into multiple time intervals from a specific advertisement execution time point in each of multiple advertising media in the big data (S111).

[0167] For example, the computing device (100) can divide data into time intervals, such as the first day, three days later, one week later, and two weeks later, based on the time of advertisement execution, and analyze advertisement performance in each interval. This allows the computing device (100) to independently evaluate the impact of each time interval on advertisement effectiveness.

[0168] In addition, the computing device (100) can repeatedly learn a model for predicting advertising efficiency using data divided into multiple time intervals (S112).

[0169] For example, a computing device (100) can input data collected at each time interval and train a model capable of predicting advertising performance during that period. By learning performance at each interval, the model can predict the level of advertising effectiveness of a specific advertising medium over time.

[0170] In addition, the computing device (100) can recognize the delay effect occurring in each of the plurality of advertising media by comparing the advertising performance for each of the plurality of time intervals included in the output data of the repeatedly learned model (S113).

[0171] Specifically, the computing device (100) can determine whether the advertising effect is immediate or delayed by comparing and analyzing changes in advertising performance over each time period. This allows the computing device (100) to identify whether a specific advertising medium exhibits its greatest advertising effect several days after its execution.

[0172] In addition, the computing device (100) can predict the advertising efficiency for a specific time period after advertising execution for each of multiple advertising media by reflecting the delay effect (S114).

[0173] Specifically, the computing device (100) can predict future advertising execution timing and the resulting performance based on the identified delay effect for each advertising medium. For example, if a delay effect is identified for a specific advertising medium, where maximum advertising effectiveness occurs after three days, the advertising strategy can be optimized by adjusting the subsequent advertising execution timing to achieve optimal performance. Specifically, the computing device (100) can optimize the subsequent advertising execution timing by reflecting the delay effect, where maximum advertising effectiveness occurs after three days. In this way, advertisers can achieve higher advertising effectiveness through an optimized advertising strategy that reflects the delay effect.

[0174]

[0175] Throughout this specification, the terms model, neural network, network function, and neural network may be used interchangeably. Hereinafter, they are collectively referred to as neural networks. A data structure may include a neural network. Furthermore, a data structure including a neural network may be stored on a computer-readable medium. A data structure including a neural network may also include data input to the neural network, weights of the neural network, hyperparameters of the neural network, data obtained from the neural network, activation functions associated with each node or layer of the neural network, and loss functions for training the neural network. A data structure including a neural network may include any of the components described above. That is, a data structure including a neural network may be configured to include all or any combination of data input to the neural network, weights of the neural network, hyperparameters of the neural network, data obtained from the neural network, activation functions associated with each node or layer of the neural network, and loss functions for training the neural network. In addition to the aforementioned components, a data structure including a neural network may include any other information that determines the characteristics of the neural network. Additionally, the data structure may include any form of data used or generated in the computational process of the neural network, and is not limited to the aforementioned. The computer-readable medium may include a computer-readable recording medium and / or a computer-readable transmission medium. The neural network may be composed of a set of interconnected computational units, which may generally be referred to as nodes. These nodes may also be referred to as neurons. The neural network is composed of at least one node.

[0176] While the embodiments of the present invention have been described above with reference to the attached drawings, those skilled in the art will appreciate that the present invention can be implemented in other specific forms without altering the technical concept or essential features thereof. Therefore, the embodiments described above should be understood to be illustrative in all respects and not restrictive.

Claims

1. A method performed by a computing device including at least one processor, A step of predicting advertising efficiency for multiple advertising media based on big data; and A step of obtaining advertising optimization results based on predicted advertising efficiency and advertising condition information; Including, The step of predicting the above advertising efficiency is: A step of predicting first advertising efficiency through a first model that predicts advertising efficiency of a specific parking lot based on past data corresponding to a specific parking lot in the big data; A step of predicting the second advertising efficiency through a second model that predicts the advertising efficiency of the specific parking lot by setting the parking lot as a categorical variable in the form of a binary in the above big data; and A step of predicting the advertising efficiency using one of the first model and the second model, which has a relatively high prediction accuracy; including, How to optimize advertising based on big data.

2. In paragraph 1, The above method, A step of subsampling advertising efficiency data corresponding to each parking lot from the above big data; A step of generating learning data corresponding to advertising efficiency by parking lot based on the sub-sampled data; and A step of generating a first model that predicts advertising efficiency by week for each of the plurality of advertising media using the above learning data; including more, How to optimize advertising based on big data.

3. In paragraph 2, The step of generating the above learning data is: A step of generating multiple training data by assigning weights corresponding to recency to the subsampled data; or A step of generating multiple learning data by setting past period parameters differently from the above subsampled data; Including, The step of predicting the first advertising efficiency through the first model is: A step of generating a plurality of preliminary models for predicting advertising efficiency based on the plurality of learning data; A step of comparing the advertising efficiency predicted by each of the plurality of preliminary models with the actual advertising efficiency, and recognizing the first model with the highest prediction accuracy; and A step of predicting advertising efficiency for each parking lot by using the first model; including, How to optimize advertising based on big data.

4. In paragraph 1, The above method, A step of obtaining parking characteristic data by setting parking as a binary categorical variable in the above big data; A step of generating learning data by combining advertising efficiency data related to the above parking characteristic data; and A step of generating a second model that predicts advertising efficiency by week for each of the plurality of advertising media using the above learning data; including more, How to optimize advertising based on big data.

5. In paragraph 4, The step of generating the above learning data is: A step of generating multiple learning data by assigning weights corresponding to recency to the above parking characteristic data; or A step of generating multiple learning data by setting past period parameters differently in the above parking characteristic data; Including, The step of predicting the second advertising efficiency through the second model is as follows: A step of generating a plurality of preliminary models for predicting advertising efficiency based on the plurality of learning data; A step of comparing the advertising efficiency predicted by each of the plurality of preliminary models with the actual advertising efficiency, and recognizing the second model with the highest prediction accuracy; and A step of predicting advertising efficiency for each of the plurality of advertising media by using the second model; including, How to optimize advertising based on big data.

6. In paragraph 2 or paragraph 4, The above method, A step of recognizing the number of times a holiday is included in each week in the above big data; A step of updating the learning data by reflecting the number of times a holiday is included in each week in the learning data; and A step of generating the first model or the second model using the updated learning data to predict advertising efficiency according to the number of times a holiday is included in a parking lot; including more, How to optimize advertising based on big data.

7. In paragraph 1, The step of predicting the above advertising efficiency is: A step of dividing the data into multiple time intervals from the point in time at which a specific advertisement was executed in each of the multiple advertising media in the above big data; A step of repeatedly training a model for predicting advertising efficiency using data divided into the above multiple time intervals; A step of recognizing a delay effect occurring in each of the plurality of advertising media by comparing the advertising performance for each of the plurality of time intervals included in the output data of the repeatedly learned model; and A step of predicting advertising efficiency for a specific time period after advertising execution for each of the plurality of advertising media by reflecting the above delay effect; including, How to optimize advertising based on big data.

8. In paragraph 1, The above advertising conditions information is: Includes information about your advertising budget and advertising objectives; The steps to obtain advertising optimization results are: A step of determining at least one advertising medium to run an advertisement among the plurality of advertising mediums based on the predicted advertising efficiency, the advertising budget, and the advertising objective; and A step of determining an advertising cost distribution ratio for the at least one advertising medium based on the predicted advertising efficiency and the advertising budget; including, How to optimize advertising based on big data.

9. Memory that stores one or more instructions; and A processor that executes one or more instructions stored in the memory. Including, The processor executes one or more of the instructions, A device for performing the method of claim 1.

10. Combined with a computer, which is hardware, A step of predicting advertising efficiency for multiple advertising media based on big data; and A step of obtaining advertising optimization results based on predicted advertising efficiency and advertising condition information; Including, The step of predicting the above advertising efficiency is: A step of predicting first advertising efficiency through a first model that predicts advertising efficiency of a specific parking lot based on past data corresponding to a specific parking lot in the big data; A step of predicting the second advertising efficiency through a second model that predicts the advertising efficiency of the specific parking lot by setting the parking lot as a categorical variable in the form of a binary in the above big data; and A step of predicting the advertising efficiency using one of the first model and the second model, which has a relatively high prediction accuracy; A computer program stored on a recording medium readable by a computing device to perform a big data-based advertising optimization method including:

Images