Artificial intelligence based content-tailored ad serving system

Abstract

The present invention relates to a content-customized advertising placement system using artificial intelligence, and more specifically, comprises: a receiving unit that receives page data of websites; an analysis unit that analyzes the page data and classifies the page data by category; and a matching unit that generates match data by matching one or more of the page data with the advertising content data using the classification data generated by the analysis unit classifying the page data by category and the advertising content data.

Description

Technology Field

[0001] The present invention relates to a content-customized advertising placement system using artificial intelligence. Background Technology

[0003] A banner is an advertisement consisting of an image file and a hyperlink located in a corner of an internet webpage.

[0004] When an internet user clicks or selects the above banner advertisement, the advertisement redirects to a landing page equipped with product information and a purchase button.

[0005] The price of the above banner is determined by the traffic volume of the above webpage, the size of the above advertisement image, and the location where it is displayed.

[0006] Meanwhile, the prior art, registered patent No. 10-1615264 (hereinafter referred to as prior art), relates to a banner advertising method and apparatus, and more specifically, includes the steps of: transmitting data constituting an advertising banner including a query input area to a user terminal; receiving a query entered in the query input area of ​​the advertising banner displayed on the user terminal from the user terminal; determining a landing page that matches the entered query; and transmitting data for displaying the matched landing page to the user terminal. By doing so, the user is not passively able to access only pre-specified information through the banner advertisement, but can instead search for and access the information they want among the information presented by the advertiser, thereby reducing the user's aversion to the banner advertisement.

[0007] However, the aforementioned prior art is an invention for selecting a landing page that is displayed when an ad banner is clicked, but it cannot be considered an efficient method for ad exposure when noting that the majority of users do not consciously prefer to be exposed to ads.

[0008] The above prior art had problems such as low advertising effectiveness because it did not include configurations that analyze data of web pages where banner advertisements are displayed and match appropriate advertising content to web pages so that users are unintentionally exposed to advertisements in fields they are interested in. Prior art literature

[0010] Korean Registered Patent No. 10-1615264 The problem to be solved

[0011] The problem to be solved by the present invention is to improve advertising efficiency by analyzing data of a webpage where an advertisement is displayed through artificial intelligence and matching suitable advertising content to said webpage.

[0012] The present invention has the purpose of improving advertising efficiency by matching advertising content suitable for a webpage using the webpage's own content or text, images, and videos of another advertisement.

[0013] The problems of the present invention are not limited to those mentioned above, and other unmentioned technical problems will be clearly understood by those skilled in the art from the description below. means of solving the problem

[0015] The present invention for achieving the above objective comprises: a receiving unit for receiving page data of sites; an analysis unit for analyzing the page data and classifying the page data by category; and a matching unit for generating match data by matching one or more of the page data with the advertising content data using the classification data generated by the analysis unit classifying the page data by category and the advertising content data.

[0016] Additionally, the analysis unit may include a setting unit for setting key words for each of a plurality of categories; and a first classification unit for setting categories of the page data using key words that match the categories and text data of the page data.

[0017] Additionally, the first classification unit may include a frequency unit for calculating the number of each of the first keywords in the text data corresponding to the main words that match each of the categories; and a comparison unit for matching the category in which the main word corresponding to the largest number of the first keywords (hereinafter referred to as the first designated keyword) is set with the page data.

[0018] Additionally, the analysis unit may further include an extraction unit that extracts second keywords from image data of the page data; and a second classification unit that sets the category of the page data using major words that match the categories and the second keywords of the page data.

[0019] Additionally, the analysis unit may further include a collection unit that collects reference images for a plurality of categories; and a third classification unit that sets the category of the page data by comparing the reference images with the images of the page data.

[0020] Additionally, the system further includes a preference unit that stores positive keywords and negative keywords, wherein the preference unit receives the match data from the matching unit and can generate positive data or negative data that matches the match data by using the text data of the page data and the positive keywords and negative keywords from the match data. Effects of the invention

[0022] According to an embodiment of the present invention, by analyzing data of a webpage where an advertisement is displayed and matching suitable advertising content to the webpage, the effect of improving advertising efficiency is achieved.

[0023] The present invention has the effect of improving advertising efficiency by matching advertising content suitable for a webpage using the webpage's own content or text, images, and videos of another advertisement. Brief explanation of the drawing

[0025] FIG. 1 is a schematic diagram illustrating a content-customized ad placement system according to one embodiment of the present invention. Figure 2 is a block diagram illustrating a management server. Specific details for implementing the invention

[0026] The advantages and features of the present invention and the methods for achieving them will become clear by referring to the embodiments described below in detail together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below but may be implemented in various different forms. These embodiments are provided merely to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the invention, and the present invention is defined only by the scope of the claims.

[0027] Throughout this specification, when a part is described as being "connected" to another part, this includes not only cases where they are "directly connected (including electrical connections capable of communication)" but also cases where they are "indirectly connected (including electrical connections capable of communication)" with other elements interposed between them.

[0029] FIG. 1 is a schematic diagram illustrating a content-customized ad placement system according to an embodiment of the present invention. FIG. 2 is a block diagram illustrating a management server.

[0030] A content (website content) customized advertising placement system (S) (a content customized advertising placement system using artificial intelligence) according to one embodiment of the present invention is designed to analyze a website where advertising content is placed using an artificial intelligence algorithm and to place appropriate advertising content on said website, and for convenience of explanation, it will be referred to as ‘this system’ below.

[0031] Referring to FIGS. 1 and 2, the present system (content-customized ad placement system (S)) may include a receiving unit, an analysis unit, and a matching unit.

[0032] The receiving unit, analysis unit, and matching unit, etc., may be components included in the management server (1). The present system may include the management server (1).

[0033] Various control configurations or computing devices known to a person skilled in the art (e.g., various electronic devices or electronic modules such as computers, processors, etc.) may be applied to the management server (1). For example, the management server (1) may be configured to transmit signals or commands generated according to a program or algorithm (such as a conventional artificial intelligence algorithm) to the client (2) and external server (3) described below. For example, the management server (1) may be an independent configuration connected via wired or wireless connection to the client (2) and external server (3) described below. As stated above, the management server (1) may include an artificial intelligence algorithm; however, since the artificial intelligence algorithm is a known technology, a more detailed description will be omitted.

[0034] The receiving unit can receive page data (web page data) from sites (for example, affiliate media sites of a portal site).

[0035] As will be explained in more detail in the following description, the above page data may include text data, image data, and video data.

[0036] The text data, image data, and video data of the above page data may be the content of the webpage itself or may be content related to advertisements displayed on the webpage.

[0037] For example, the above page data may be HTML source (form). For example, the client (2) may access sites where advertisements are to be displayed and output HTML source, and the client (2) may transmit HTML source to the receiver.

[0038] The above client (2) may, for example, be a mobile terminal or PC of a user who wishes to display an advertisement. The present system may include the above client (2).

[0039] The method for extracting the above HTML source is the announced method (right-click on the site webpage and click 'View Source', type 'ctrl+U' on the keyboard, etc.), so a more detailed explanation will be omitted.

[0040] The above page data may be diverse in addition to HTML source, and may be in various formats as long as the page data of the webpage of the above site can be identified.

[0041] The analysis unit receives the above page data from the receiving unit and can classify the above page data by category.

[0042] For example, the above categories may include categories in various fields such as 'food category', 'exercise category', 'travel category', and 'insurance category'.

[0043] The above analysis unit may include a setting unit that sets (matches) the main words of each of a plurality of categories.

[0044] For example, the above analysis unit (setting unit) can set words related to food, such as 'food', 'famous restaurant', 'mukbang', 'meat', 'restaurant', 'cooking', etc., as the above key words of the 'food category'.

[0045] For example, the analysis section (settings section) can set words related to exercise, such as 'health', 'soccer', 'diet', and 'yoga', as key words for the 'exercise category'.

[0046] For example, the analysis section (settings section) can set insurance-related words such as 'insurance' and 'indemnity insurance' as key words for the 'insurance category'.

[0047] For example, the above key word may be input into the client (2) and transmitted from the client (2).

[0048] At this time, the analysis unit may include a first classification unit that sets the category of the page data using the main words set (matched) to each of the above categories and the text data of the page data.

[0049] The first classification unit above can extract text data displayed within tags in the HTML source described above, for example.

[0050] As described above, the management server (1) may include an artificial intelligence algorithm, and the first classification unit may perform natural language processing (NLP) on the text data to tokenize the text data and compare it with the main words.

[0051] The first classification unit may include a frequency unit that calculates the number of first keywords (tokens) of text data corresponding to major words that match each of the categories. That is, the text data may include first keywords.

[0052] In other words, the frequency unit can determine the frequency of each of the first keywords that match the main words on a specific webpage.

[0053] The first classification unit may include a comparison unit that generates first setting data by matching the category in which a major word corresponding to the largest number of the first keywords (hereinafter referred to as the first designated keyword) is set (matched) with the page data.

[0054] The above classification data may include one or more of the above first setting data, the second setting data described below, and the third setting data described below.

[0055] That is, the comparison unit can set the webpage as the aforementioned 'food category' if the page data of the specific webpage includes a first keyword 'food' and the number of first keywords 'food' is the highest among other first keywords.

[0056] The above analysis unit can generate classification data by matching the above page data with categories as described above and classifying them by category.

[0057] That is, the classification data may be in a data frame format and may include the specific category and page data matched with the specific category. The page data may include an address for accessing a webpage.

[0058] The matching unit can receive classification data (data frame) containing page data and categories from the analysis unit by classifying the page data by category.

[0059] The matching unit can generate match data by using the above classification data and advertising content data to match one or more of the page data with the above advertising content data.

[0060] The above matching unit can receive advertising content data from the above client (2).

[0061] The above advertising content data may include, for example, an image to be used as a banner ad, a category name, a product name, etc.

[0062] The above matching unit can generate match data by comparing the category of each of the multiple pages of the classification data with the category name of the advertising content data, or by comparing the product name of the advertising content data with the main word matched to the category, thereby matching the advertising content data with the page data.

[0063] That is, the matching unit can generate match data by matching the page data of the classification data with the advertising content data when the category name of the classification data corresponds to the category name of the advertising content data, or when the aforementioned key word corresponds to the product name. The match data may also be in the form of a data frame. As described above, the page data may include a web page address.

[0064] The above matching unit can transmit the above match data to the client (2), and the client (2) can receive and output the above-described match data.

[0065] The above matching unit can transmit the match data to the web server (external server (3)) of the site where the advertisement is displayed, so that the banner advertisement of the advertisement content data is displayed on the webpage in real time.

[0066] In this way, the present system has the advantage of improving advertising efficiency by analyzing the content of web pages of the site(s) where advertisements are displayed, selecting web pages suitable for the advertisement content data, and ensuring that advertisements are displayed on those web pages.

[0067] As mentioned above, it was stated that the page data may include image data in addition to text data.

[0068] At this time, the analysis unit (image data of the above page data) <video>It may include an extraction unit that extracts a second keyword from an element.

[0069] For example, the extraction part is HTML's <video>Secondary keywords (e.g., 'chestnut') can be extracted from the element and the secondary keywords can be tokenized (the above primary word may include 'chestnut').

[0070] The analysis unit may include a second classification unit that receives second keywords from the extraction unit and generates second setting data by setting (matching) the category of the page data using the second keywords and the main words that match the categories.

[0071] If there is a corresponding word among the above second keywords to the main word of the above category, the page data containing the above second keyword may be set to a specific category in which the main word matching the above second keyword is set by the above second classification unit.

[0072] As described above, the matching unit can generate the matching data using the classification data and the advertising content data and transmit it to the client (2) or the external server (3).

[0073] As such, the present system has the advantage of being able to utilize not only text data of page data but also video data (another advertisement video or video of the webpage's own content) as data for displaying the aforementioned advertisement.

[0074] Referring to FIGS. 1 and 2, the analysis unit may include a collection unit that collects reference images for a plurality of categories.

[0075] The above collection unit can receive and store reference images transmitted from clients (2), etc., by category.

[0076] The above analysis unit receives the reference images from the above collection unit, and the images and the image of the page data (image data of the page data ( It may include a third classification unit that sets the category of the page data by comparing the element) and the image on the path.

[0077] The third classification unit may, for example, include a comparison unit that calculates a match rate by comparing the pixel brightness and color of each of a plurality of reference images with the image (an image on an image data path).

[0078] For example, the above comparison unit may include a conventional artificial intelligence image (object) recognition algorithm, and may extract features, contours, etc. of reference images by category using pixel-by-pixel brightness and color, etc., and compare the pixels of the image (features and contours, etc. by pixels) to obtain the matching rate between the reference images by category and the image.

[0079] The above third classification unit may include a designation unit that generates the above third setting data by matching the category matched with one of the reference images (hereinafter designated reference image) that has the highest (largest) matching rate with the above image with the above page data (image data).

[0080] As described above, the matching unit can generate match data using classification data including third setting data and advertising content data and transmit it to the client (2), etc.

[0081] As such, this system has the advantage of being able to analyze images on web pages and utilize them for displaying advertisements.

[0082] Meanwhile, the above extraction unit is the image data of the page data ( The third keywords of the element (including, for example, 'chestnut') can be extracted.

[0083] The above extraction unit can extract third keywords (e.g., 'chestnut') and tokenize the third keywords.

[0084] At this time, the third classification unit receives third keywords from the extraction unit and can generate third setting data by setting (matching) the category of the page data using the main words that match the categories and the third keywords.

[0085] If there is a corresponding word among the above third keywords to the main word of the above category, the page data containing the above third keyword can be set to a specific category by the above third classification unit.

[0086] That is, the first classification unit (comparison unit) generates the first setting data, the second classification unit generates the second setting data, and the third classification unit or designation unit can generate the third setting data.

[0087] As such, the present system has the advantage of being able to utilize not only text data of page data but also image data as data for displaying the aforementioned advertisement.

[0088] Referring to FIGS. 1 and 2, the analysis unit may include a calculation unit that receives the first keywords and the number of each of the first keywords from a first classification unit (frequency unit), and calculates the usage ratio of the first designated keyword using the total sum of the number of each of the first keywords and the number of the first designated keyword.

[0089] The above operation unit can calculate the usage ratio by dividing the number of the first designated keyword by the total sum.

[0090] The analysis unit may include a selection unit that generates classification data including any one of the first setting data, the second setting data, and the third setting data when receiving any one of the first setting data, the second setting data, and the third setting data.

[0091] The above selection unit can receive first setting data from the first classification unit, receive second setting data from the second classification unit, and receive third setting data from the third classification unit.

[0092] When the selection unit receives the first setting data, the second setting data, and the third setting data, the image data (from the always-on receiving unit) <video 요소)의 영상크기 데이터(웹페이지에서 영상 크기에 관한 데이터)와 이미지 데이터의 크기 데이터(웹페이지에서 상기 이미지의 크기에 관한 데이터)를 수신하고, 상기 연산부로부터 사용비율을 수신할 수 있다.

[0093] Here, image size data is image data ( <video>It may include width and height values ​​of the element, and the size data is the image data ( It can include the width and height values ​​of the element.

[0094] At this time, the selection unit may generate classification data including the first setting data when the usage ratio exceeds a preset reference ratio (e.g., 0.5 or 50%), generate classification data including the second setting data when the image size (width and height) in the image size data exceeds a preset reference image size (e.g., width 200 and height 200), and generate classification data including the third setting data when the image size (width and height) of the size data exceeds a preset reference size (e.g., width '200' and height '200').

[0095] In other words, if a specific webpage (webpage data) includes text data, image data, and video data, categories matching advertising content data can be set only for data that exceeds a threshold value (one or more of text data, image data, and video data). Through this, advertising content can be matched by selecting data with a high proportion on the webpage (one or more of text data, image data, and video data).

[0096] This allows for the selection of noise with low text, image, and video content on the webpage, thereby enabling the more accurate display of advertisements for specific categories.

[0097] Meanwhile, the selection unit, upon receiving the first setting data, the second setting data, and the third setting data, can calculate a first difference value between the usage ratio and the reference ratio when the usage ratio exceeds the reference ratio, calculate a second difference value between the image size and the reference image size when the image size exceeds the reference image size, and calculate a third difference value between the image size and the reference size when the image size exceeds the reference size.

[0098] At this time, the selection unit can generate classification data including first exposure data that matches the first setting data and is proportional to the first difference value, second exposure data that matches the second setting data and is proportional to the second difference value, and third exposure data that matches the third setting data and is proportional to the third difference value.

[0099] Exposure data may include the number of exposures and exposure time of advertising content, and each of the number of exposures and exposure time may increase in proportion to the difference value.

[0100] In other words, the higher the usage rate of the first keyword on the webpage, the more exposure can be given to advertisements classified into a specific category on that webpage, and the larger the size of the image or video on the webpage, the more exposure can be given to advertisements classified into a specific category.

[0101] Through this, when a specific webpage contains text, images, and videos, more efficient advertising can be performed by controlling the ad exposure time and number of impressions for categories set according to their proportions.

[0103] Meanwhile, the management server (1) may include a preference unit that receives and stores positive keywords such as 'recommendation' and 'highly recommended', and negative keywords such as 'fraud', 'damage', and 'not recommended'.

[0104] The above positive keywords and negative keywords may be input and transmitted from the aforementioned client (2), etc., and the preference unit may receive and store the positive keywords and negative keywords from the client (2).

[0105] The above preference unit receives match data from the above matching unit and can generate positive data or negative data that matches the match data by using the text data of the page data and the positive keyword and negative keyword from the match data.

[0106] The above preference section can extract text data displayed within tags from the HTML source.

[0107] As described above, the management server (1) may include an artificial intelligence algorithm, and the preference unit may perform natural language processing (NLP) on the text data to tokenize the text data and compare it with the positive keywords and negative keywords.

[0108] The above preference unit can compare each of the above positive keyword and negative keyword with additional keywords in which the text data has been tokenized, and if there is an additional keyword corresponding to the positive keyword, it can generate positive data, and if there is an additional keyword corresponding to the negative keyword, it can generate negative data and match it with the match data to generate final data.

[0109] The above final data may also be in the form of a data frame. That is, the above final data may include one or more of positive data and negative data and the above match data.

[0110] The above preference unit can transmit the above final data to a client (2) or an external server (3).

[0112] Although embodiments of the present invention have been described above with reference to the attached drawings, those skilled in the art will understand that the present invention may be implemented in other specific forms without changing the technical concept or essential features thereof. Therefore, the embodiments described above should be understood as illustrative in all respects and not restrictive. Explanation of the symbols

[0114] S: Content-customized ad serving system 1: Management Server 2: Client 3: External server< / video> < / video> < / video>

Claims

Claim 1 A receiving unit for receiving page data of sites; an analysis unit for analyzing the page data and classifying the page data by category; and a matching unit for generating match data by matching one or more of the page data with the advertising content data using the classification data generated by the analysis unit classifying the page data by category and the advertising content data, wherein the analysis unit includes a setting unit for setting a key word for each of a plurality of categories; and a first classification unit for setting a category of the page data using the key words matched with the categories and the text data of the page data, wherein the first classification unit includes a frequency unit for calculating the number of each of the first keywords in the text data corresponding to the key words matched with each of the categories; and a comparison unit for matching the page data with a category in which a key word corresponding to one of the first keywords with the largest number (hereinafter referred to as the first designated keyword) is set, and wherein the analysis unit includes an extraction unit for extracting second keywords from the image data of the page data; The method further includes a second classification unit that sets the category of the page data using major words matching the categories and second keywords of the page data, and the analysis unit further includes: a collection unit that collects reference images for each of the multiple categories; and a third classification unit that sets the category of the page data by comparing the reference images with the images of the page data, and the third classification unit includes: a comparison unit that calculates matching rates by comparing the pixel brightness and color of each of the multiple reference images with the image; and a designation unit that generates third setting data by matching the category matched with the page data to any one of the reference images (hereinafter designated reference image) that has the highest matching rate with the image, and the first classification unit sets the category of the page data to generate first setting data, and the second classification unit sets the category of the page data to generate second setting data.The above third classification unit generates third setting data by setting the category of the page data, and the above analysis unit is a calculation unit that calculates the usage ratio of the first designated keyword using the sum of the total number of each of the first keywords and the number of the first designated keyword; The method further includes a selection unit that generates classification data including any one of the first setting data, the second setting data, and the third setting data when receiving any one of the first setting data, the second setting data, and the third setting data. When the selection unit receives the first setting data, the second setting data, and the third setting data, it receives image size data of the image data and image size data of the image from the receiving unit and receives a usage ratio from the operation unit. When the usage ratio exceeds a preset reference ratio, it generates classification data including the first setting data. When the image size in the image size data exceeds a preset reference image size, it generates classification data including the second setting data. When the image size of the size data exceeds a preset reference size, it generates classification data including the third setting data. When the image size of the size data exceeds a preset reference size, the selection unit calculates a first difference value between the usage ratio and the reference ratio when the usage ratio exceeds the reference ratio, while the first setting data, the second setting data, and the third setting data are received. If the above image size exceeds the above reference image size, a second difference value between the above image size and the above reference image size is calculated, and if the above image size exceeds the above reference size, a third difference value between the above image size and the above reference size is calculated, and the selection unit, when receiving the first setting data, the second setting data and the third setting data,A content-customized ad placement system comprising: generating classification data including first exposure data that matches the first setting data and is proportional to the first difference value, second exposure data that matches the second setting data and is proportional to the second difference value, and third exposure data that matches the third setting data and is proportional to the third difference value; further comprising a preference unit that stores positive keywords and negative keywords, wherein the preference unit receives the match data from the matching unit and generates positive data or negative data that matches the match data using the text data of the page data and the positive keywords and negative keywords from the match data. Claim 2 delete Claim 3 delete Claim 4 delete Claim 5 delete Claim 6 delete Claim 7 delete

Images