KR20200017318A - AI-based Management Server for Advertising Keyword - Google Patents

Abstract

The present invention relates to an advertisement keyword managing server based on artificial intelligence, which comprises an advertisement keyword selecting unit, an exposure ranking determination unit, and a bidding price determination unit. The advertisement keyword selecting unit extracts an associated keyword associated with a request keyword received from an advertiser terminal, and calculates an ROI for each keyword with respect to the associated keyword on the basis of keyword advertisement data received from a search server. The advertisement keyword selecting unit selects a keyword corresponding to the ROI for each keyword and an advertisement condition as an advertisement keyword. The exposure ranking determination unit extracts an ROI for each exposure ranking from an exposure region of the search server, and selects the exposure ranking corresponding to the ROI for each exposure ranking and the advertisement condition. The bidding price determination unit derives a bidding price function from learning data including the keyword advertisement data with respect to a plurality of bidding variables using a multi-variable linear regression method, and generates a bidding price for the advertisement keyword and the exposure ranking by substituting a bidding variable value for the bidding price function.

Description

AI-based Management Keyword for Advertising Keyword {AI-based Management Server for Advertising Keyword}

The present invention relates to keyword advertising, and more particularly, to an advertising keyword management server that can maximize the cost-effective advertising effect based on artificial intelligence (AI).

Keyword advertising is an advertisement method that lists a client company (called an advertiser) site in a predetermined advertisement area in each portal site when a specific keyword is searched. Keyword advertising does not expose the advertisement to an unspecified number of people, but the advertisement effect is high because the advertisement is immediately exposed to the potential customers who need the service of the advertiser.

Keyword advertising includes Pay for Performance (P4P), Cost Per Millennium (CPM), Cost Per Action (CPA), etc., according to the method of determining the unit price of advertising, and P4P is widely used.

The P4P method prepays advertising costs, and subtracts the amount of cost per click (CPC) determined by the bid for each keyword whenever a click occurs. The cost per click can be determined by the advertisers via competitive bidding. In other words, if a minimum unit price (eg, 70 won) is set and advertisers bid more for a certain amount of money (eg, 10 won), the advertiser's advertisement is exposed at a desired position.

Competitive bidding can usually go through the following steps: First, the advertiser creates an account (i.e. username and password), deposits advertising costs, and registers information about the advertisement. In response, the portal site (search server) determines whether to approve the advertisement. Here, the advertisement information may include a keyword to expose the advertisement, page information to be linked after a click (URL to be shown, URL to be actually connected), a title, and a description. Thereafter, the advertiser may perform a bidding by transmitting bids every set time with respect to a desired advertisement area and an exposure rank.

As described above, in the P4P scheme, an advertisement area, an exposure rank, and the like are determined through a competitive bidding scheme, and an advertisement effect thereof is changed, and therefore, an advertiser needs to manage an advertisement keyword separately. However, in order to maximize the advertisement effect, since similar keywords, ie, a plurality of related keywords, are registered in addition to the representative keywords, it is not easy to manage a plurality of keywords. In particular, small and mid-sized businesses have more difficulty in online advertising as the number of search keywords increases. If you have enough budget, you can use an advertising agency to do your keyword advertising, but most small businesses are small.

[Prior Art Documents] Korean Patent Publication No. 2017-0037765 (Automatic Bidding System, Method, and Bid Management Server for Keyword Ads Based on Competitive Situation)

The present invention is to solve this problem,

First, you can run online keyword ads outside of business hours.

Second, to provide an artificial intelligence-based advertising keyword management server that can maximize the cost-effectiveness of online keyword advertising.

The artificial intelligence-based advertising keyword management server according to the present invention for achieving the above object may include an advertisement keyword selecting unit, an exposure ranking determiner, a bid determination unit, a controller and the like.

The advertisement keyword selecting unit may extract a related keyword related to the request keyword received from the advertiser terminal. The advertisement keyword selecting unit may calculate a keyword-specific ROI for the requested keyword and the related keyword based on the keyword advertisement data received from the search server. The advertisement keyword selecting unit may select a keyword matching the keyword ROI and the advertisement condition as the advertisement keyword.

The exposure ranking unit may extract the ROI for each exposure ranking from the exposure area of the search server. The exposure ranking unit may select an exposure ranking that meets the ROI and advertisement conditions for each exposure ranking.

The bid determination unit may derive a bid function from the training data including the keyword advertisement data by multivariate linear regression for a plurality of bid variables. The bid determiner may generate a bid for an advertisement keyword and an impression ranking by substituting a bid variable value in a bid function.

The controller may perform bidding by transmitting bid information including an advertisement keyword, an exposure ranking, and a bid to the search server.

In the advertisement keyword management server according to the present invention, the ROI for each keyword may include the number of page views, the number of effective clicks, and the sales / expense advertising costs of the previous month.

In the advertisement keyword management server according to the present invention, the advertisement condition may include the number of page views, effective clicks, and sales / expenses.

In the advertisement keyword management server according to the present invention, the exposure ranking ROI may include the effective click count, the previous month's sales, and the sales / expense advertising expenses.

In the keyword management server according to the present invention, the bid variable is at least one selected from the number of page views, effective clicks, average bids, maximum bids, advertisers, average selling price, time information (month, day, etc.), industry Can be.

In another form of the advertisement keyword management server according to the present invention, the controller may receive from the search server keyword advertising data related to the real-time popular search word and the real-time popular search word included in the request keyword and the related keyword.

Another form of the advertisement keyword management server according to the present invention may include an advertiser selection unit that selects an advertiser to bid on based on a real-time popular search term as an advertisement keyword according to an advertiser selection criterion.

In another form of the advertisement keyword management server according to the present invention, the advertisement keyword selecting unit may select a real-time popular search term as an advertisement keyword.

In another form of the advertisement keyword management server according to the present invention, the exposure ranking determiner, the bid determiner, and the controller may perform the exposure ranking, the bid determination, and the bidding, respectively, using the real-time popular search term as the advertisement keyword.

In another form of the advertisement keyword management server according to the present invention, the advertiser selection criterion is a primary selection criterion for determining whether to have a request keyword that may include or include a real-time popular search term as a related keyword, an advertisement condition, a product sale price, and a membership registration. It may include a second selection criteria selected from the order.

According to the advertisement keyword management server according to the present invention having such a configuration, the maximum ROI is automatically determined and determined through advertisement conditions, keyword advertisement data, learning data, and the like. By performing the keyword advertisement to generate, it is possible not only to reduce the advertiser's keyword advertising burden, but also to maximize the cost effectiveness of the advertisement.

1 is a block diagram of a keyword advertising agency system including an ad keyword management server according to the present invention.
1 illustrates a keyword advertisement agency process of the advertisement keyword management server according to the present invention.
3 illustrates a process of selecting an advertisement keyword selecting unit in the advertisement keyword management server according to the present invention.
4 illustrates a process of determining an exposure ranking unit in an advertisement keyword management server according to the present invention.
5 illustrates a determination process of a bid determination unit in an advertisement keyword management server according to the present invention.
6 is a configuration diagram of a keyword advertising agency system including a modification of the advertisement keyword management server according to the present invention.
7 illustrates a process of selecting an advertiser selection unit in a variation of an advertisement keyword management server according to the present invention.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

1 is a configuration diagram of a keyword advertising agency system including an advertisement keyword management server according to the present invention.

As shown in FIG. 1, the keyword advertisement agency system may include an advertiser terminal 100, a search server 200, an advertisement keyword management server 300, and the like.

The advertiser terminal 100 is a terminal of an advertiser who wants a keyword advertisement, and may transmit advertiser information, advertisement information, advertisement conditions, product selling price, etc. to the advertisement keyword management server 300.

The advertiser information may include a business name, a representative name, a business address, a contact information, an email, and the like. The advertiser information may include account information such as ID and password.

The advertisement information may include a request keyword, a display URL, a connection URL, a title, a brief description, and the like. The request keyword may be a representative keyword such as a pension, a bicycle, a refrigerator, or the like, or a related keyword of the representative keyword, for example, a gapyeong pension or a spa pension. There may be one or more request keywords.

The advertisement conditions may include effective clicks, page views, ROI (effective clicks / advertising spending, total sales / advertising spending, etc.), advertising costs and the like. Ad conditions are prerequisites for keyword ads. If your ad conditions include only "effective clicks," select the keywords with the highest valid clicks among the requested keywords or associated keywords, regardless of pageviews or ROI. Can be selected as an advertising keyword. The advertising cost is the maximum cost that the advertiser pays for the requested keyword (or the selected advertising keyword) and may be, for example, 500,000 won, 1.5 million won, 2 million won, or the like. Advertising costs can be added on condition of execution such as 500,000 won / week, 1.5 million won / month.

The product selling price refers to the selling price of the product sold by the advertiser. When the product is a pension lodging, the selling price of the product may include individual selling prices of 80,000 won, 100,000 won, 150,000 won, 200,000 won, and an average selling price of 130,000 won.

The advertiser terminal 100 may perform a keyword advertisement, and then transmit the revenue obtained through the keyword advertisement to the advertisement keyword management server 300.

The advertiser terminal 100 may directly perform advertiser registration. In this case, the advertiser terminal 100 accesses the search server 200 to create an account (ID, password, etc.), deposit an advertisement fee, and ad information (advertisement). Keywords, URLs shown, connection URLs, titles, brief descriptions, etc.) may be transmitted to the search server 200.

The advertiser terminal 100 may allow the advertiser registration to the advertisement keyword management server 300. In this case, the advertiser terminal 200 may transmit an account, advertisement cost, advertisement information, etc. to the advertisement keyword management server 300.

The search server 200 may receive an advertisement keyword, an advertisement cost, etc. from the advertiser terminal 100 or the advertisement keyword management server 300. The search server 200 may bid on an advertisement keyword and execute a keyword advertisement on the advertisement keyword. The keyword advertisement may include selection of an exposure area, selection of an exposure rank, and the like.

The search server 200 may transmit keyword advertisement data to the advertisement keyword management server 300 when a request is received from the advertisement keyword management server 300. Keyword ad data includes the highest bids, lowest bids, pageviews, effective clicks, deposited advertising costs, spent advertising costs, and ROI (pageviews / spent advertising costs, effective clicks / spend advertising costs, etc.) And the like.

The search server 200 may include a portal site such as Naver, Daum, Google, as well as a server of a keyword advertising agency such as Overture.

The advertisement keyword management server 300 may perform keyword advertisement through the search server 200 on behalf of the advertiser terminal 100. The advertisement keyword management server 300 may include an advertisement keyword selecting unit 310, an exposure ranking determiner 320, a bid determining unit 330, a database 340, a controller 350, and the like.

The controller 350 may receive the advertiser information, the advertisement information, the advertisement condition, the product sale price, etc. from the advertiser terminal 100 and store it in the database 340.

The controller 350 may access the search server 200 to register an advertiser, deposit an advertisement fee, and register advertisement information (ad keyword, showing URL, connection URL, title, brief description, etc.).

The controller 350 may request and receive keyword advertisement data from the search server 200. The controller 350 may utilize a screen scraping program and the like to request and receive keyword advertisement data.

The advertisement keyword selecting unit 310 is based on the keyword selection criteria in the keyword group related to the requested keyword based on the advertisement information received from the advertiser terminal 100, the advertisement condition, the product sale price, and the like, and the keyword advertisement data received from the search server 200. Matching ad keywords can be selected.

Exposure ranking unit 320 is based on the advertisement information received from the advertiser terminal 100, advertising conditions, product selling price, etc. based on the keyword advertising data received from the search server 200 in the exposure area that meets the impression selection criteria The ranking can be determined.

The bid determiner 330 may generate a bid through a bid function based on the advertisement information received from the advertiser terminal 100, the advertisement condition, the product sale price, and the like, and the keyword advertisement data received from the search server 200.

The controller 350 may transmit bid information including the advertisement keyword, the exposure ranking, and the bid selected and determined to the search server 200.

When the controller 350 receives a successful bid for the keyword from the search server 200, the keyword 350 matches the keyword in the search server 200 in accordance with the advertisement conditions (advertising fee, execution period, etc.) of the advertiser terminal 100 during the set advertisement time. Can be performed.

The control unit 350 repeatedly performs the above-described process of selecting an advertisement keyword, determining an exposure rank, determining a bid, and performing a bid, for each set advertisement time through the search server 200 under the advertisement information of the advertiser terminal 100 and the advertisement conditions. can do.

The controller 350 may store the result of the above operation as the training data in the database 340.

2 illustrates a keyword advertisement agency process of the advertisement keyword management server according to the present invention.

As shown in FIG. 2, first, the advertiser terminal 100 includes advertiser information (business name, representative name, business address, contact information, e-mail, etc.), advertisement information (request keyword, URL, etc.), advertisement conditions (advertising costs, execution period). , Advertisement keyword selection condition, exposure ranking decision condition, etc.), product sales price, etc. may be transmitted to the advertisement keyword management server 300.

The advertisement keyword management server 300 may store advertiser information received from the advertiser terminal 100 in the database 340.

The advertisement keyword management server 300 may register an advertiser as a member of the search server 200. The advertisement keyword management server 300 may transmit advertiser information (business name, representative name, business address, contact information, email, etc.), advertisement information (request keyword, URL, etc.) to the search server 200.

The advertisement keyword management server 300 requests and receives keyword advertisement data from the search server 200 including related keywords (gapyeong pension, spa pension, etc.) related to the requested keyword (pension, etc.), and the database 340. ) Can be stored.

Thereafter, the advertisement keyword management server 300 may select an advertisement keyword, determine an exposure rank in the exposure area, and generate a bid for the selected advertisement keyword and the exposure rank.

The advertisement keyword management server 300 may transmit bid information including an advertisement keyword, an exposure rank, and a bid to the search server 200 to perform a bid.

The search server 300 may perform an auction based on the bidding information, and then may transmit the successful bid to the advertisement keyword management server 300.

3 illustrates a process of selecting an advertisement keyword selecting unit in the advertisement keyword management server according to the present invention.

In operation S311, the advertisement keyword selecting unit 310 may extract a request keyword, a related keyword, keyword advertisement data related thereto, a product selling price, and the like from the database 340. For example, if the request keyword is 'pension', the related keyword may be 'gapyeong pension', 'spa pension', 'pool pension' and the like. The request keyword and the related keyword are keywords registered in the search server 200 and may be received from the search server 200 and stored in the database 340.

In operation S313, the advertisement keyword selecting unit 310 may calculate an ROI for each of the extracted keywords (a request keyword and a related keyword). The ROI may be the number of page views (keyword inputs), the number of valid clicks (the number of clicks on any of the ads listed by the user on the search results page), the sales / expenses cost, and the like.

In operation S315, the advertisement keyword selector 310 may select an advertisement keyword according to an advertisement condition set by the advertiser. The advertisement condition may be page views, effective clicks, sales / expenses, and the like. Different keywords may be optimal keywords for each advertisement condition. For example, the number of page views may include the most 'gapyeong pension' and the number of effective clicks may include the 'spasion'. In this case, if the advertiser sets the advertisement condition to 'pageviews', 'Gapyeong Pension' is selected as the keyword, and if the advertisement condition is set to 'Effective clicks', the keyword 'Spasion' is selected as the keyword. Can be. Advertisement conditions may be set in plural, and in this case, a plurality of advertisement keywords may be selected.

4 illustrates a process of determining an exposure ranking unit in an advertisement keyword management server according to the present invention.

In operation S321, the exposure ranking determiner 320 may select an exposure area in which the advertisement keyword is to be exposed in the advertisement area of the search server 200. For example, in the case of Naver, the exposed area may be a 'power link' area, a 'business site' area, a 'web site' area, or the like.

In operation S323, the exposure ranking determiner 320 may extract an ROI for each ranking in the exposure area. For example, when the exposure area is selected as the 'power link' area, the exposure ranking may be a first rank, a second rank, a third rank, ... from above in the 'power link' area. The ROI can extract or calculate the number of effective clicks, sales, and sales / spend advertising costs for the previous month. For example, the number of effective clicks may be highest in 'first ranking', sales in second ranking, and revenue / spend advertising costs may be highest in third ranking.

In operation S325, the exposure ranking unit 320 may determine an exposure ranking that satisfies the advertisement condition. For example, in the example of step S323 above, if the advertiser sets the advertising condition to 'effective clicks', select the 'first rank', and if the advertising condition is set to 'sales,' the second rank ', And if the advertising conditions are set to' sales / expenses advertising costs, 'the' third ranking 'can be selected.

5 illustrates a determination process of a bid determination unit in an advertisement keyword management server according to the present invention.

In operation S331, the bid determination unit 330 may generate a bid function. The bid function may be expressed by the formula Y = A * X1 + B * X2 + C * X3 + D * X4 + .... Here, Y is a bid, X1 to X4 are first to fourth variables, and A to D are constants for the first to fourth variables.

Variables are: "pageviews last month", "effective clicks last month", "average bids", "best bids", "number of advertisers (bids)", "average product sales", time information (months, days, etc.), Industry, and the like, and one or more of these variables can be selected.

The bid function provides training data ("pageviews," "effective clicks," "average bids," "best bids," "number of advertisers (bids)," And learning algorithms). The learning algorithm may use a machine-learning algorithm using multivariate linear regression.

When the number and type of variables are determined, the bid function may derive the constants A to D through counting iterations (linear regression) by substituting the variables and training data into a learning algorithm. Once the constant is derived, the bid function is complete.

The bid function may run the learning algorithm again when the training data is updated or the number of variables increases / decreases, to generate a modified bid function.

In step S333, the bid function Y derived in step S331 is a variable value, i.e., "pageviews", "effective clicks", "average bids", "best bids", "advertisers" in the previous month. (Number of bids) "," average selling price, "etc., into the bid function, i.e., Y = A * X1 + B * X2 + C * X3 + D * X4 + ... to calculate the bid (Y) have.

6 is a configuration diagram of an advertisement keyword management system including a modified example of an advertisement keyword management server according to the present invention, and FIG. 7 illustrates a process of selecting an advertiser selection unit in a modified example of the advertisement keyword management server according to the present invention.

As shown in FIG. 6, the modified example of the advertisement keyword management server may further include an advertiser selecting unit 360 in the first embodiment.

The controller 350 may receive a real-time popular search word from the search server 200 (S361 of FIG. 7).

The controller 350 may request and receive keyword advertisement data for a real-time popular search from the search server 200.

As illustrated in FIG. 7, the advertiser selecting unit 360 may select an advertiser for bidding using a real-time popular search term as an advertisement keyword. Advertiser selection criteria can be divided into two stages. The first selection may select an advertiser having a request keyword that may include a real-time popular search term as a related keyword (S363 of FIG. 7). The second selection may be selected according to the advertisement conditions (advertising costs, etc.), the product selling price, the membership registration order, etc. (S365 in FIG. 7). For example, if the second selection criterion is 'advertising costs', the advertiser may select the advertiser having the most 'advertising costs' among the first selected advertisers.

The advertisement keyword selecting unit 310 may select a real-time popular search term as an advertisement keyword.

The impression ranking unit 320, the bid determination unit 330, and the control unit 350 may determine the exposure ranking, bid determination, bidding, etc. using the real-time popular search term as an advertisement keyword in the same or similar manner as in the first embodiment. Can be performed.

The embodiments described above are intended to illustrate the present invention. Such embodiments may be modified or modified in other forms. However, since the scope of the present invention is defined by the claims below, such modifications and variations can be interpreted as being included in the scope of the present invention.

100: advertiser terminal 200: search server
300: Ad keyword management server 310: Ad keyword selection unit
320: impression ranking unit 330: bid determination unit
340: Database 350: control unit
360: Advertiser Selection

Claims (7)

Extracting a related keyword related to a request keyword received from an advertiser terminal, calculating ROI for each keyword for the request keyword and the related keyword based on keyword advertisement data received from a search server, and receiving the ROI for each keyword and the advertiser terminal from the advertiser terminal. An advertisement keyword selecting unit that selects a keyword matching the advertisement condition as an advertisement keyword;
An exposure priority determining unit configured to extract an ROI for each exposure ranking from the exposure area of the search server and to select an ROI for each exposure ranking and an exposure ranking corresponding to the advertisement condition;
A bid function for deriving a bid function from the training data including the keyword ad data by a multivariate linear regression method for a plurality of bid variables, and substituting a bid variable value into the bid function to generate bids for the ad keyword and impression rank. Decision unit;
And a controller configured to perform bidding by transmitting bid information including the advertisement keyword, an impression rank, and a bid to a search server. The method of claim 1, wherein the keyword-based ROI is
AI-based advertising keyword management server, including the number of pageviews, effective clicks, revenue / expense of the previous month. The method of claim 1, wherein the advertisement condition
AI-based advertising keyword management server, including page views, effective clicks, revenue / expenses. The method of claim 1, wherein the exposure rank ROI is
Artificial intelligence-based advertising keyword management server, including effective clicks, sales in the previous month, revenue / expenses. The method of claim 1, wherein the bid variable is
AI-based advertising keyword management server, which is at least one selected from the number of page views, effective clicks, average bids, highest bids, advertisers, average selling price, time information (month, day of the week, etc.), industry. The method of claim 1,
The controller receives from the search server real-time popular search terms included in the request keyword and related keywords and keyword advertisement data related to the real-time popular search terms,
And an advertiser selecting unit for selecting an advertiser to bid based on the real-time popular search term as an advertisement keyword according to an advertiser selection criterion.
The advertisement keyword selecting unit selects the real-time popular search term as an advertisement keyword,
The exposure ranking determiner, the bid determiner, and the controller performs the exposure ranking, bid determination, and bidding, respectively, using the real-time popular search term as an advertisement keyword. The method of claim 6, wherein the advertiser selection criteria
Artificial intelligence-based, including the first selection criteria and advertising conditions, product sales price, the second selection criteria selected from the order of membership registration to determine whether the request keyword that can include or include the real-time popular search term as a related keyword Ad Keyword Management Server.

Images