KR102122883B1 - Mobile apparatus and method for providing contextual advertisement reflecting user interest - Google Patents

Abstract

The present invention relates to a mobile device and a method for providing a contextual advertisement reflecting a user's interest. A mobile device according to an embodiment of the present invention includes a display; And a processor storing a program providing advertisement content, and a processor executing the program stored in the memory. The processor determines user interest from user content stored in the memory based on a pre-stored topic classification tree according to the execution of the program, and when a request to display an advertisement in an application currently being executed is obtained, based on the topic classification tree Determine a category of application content, determine each category of at least one advertisement content previously obtained based on a topic classification tree, and use user interest, similarity between a category of application content and a category of advertisement content, and One advertisement content is selected from the advertisement contents and output on the display.

Description

Mobile device and method for providing contextual advertisements reflecting user interests{MOBILE APPARATUS AND METHOD FOR PROVIDING CONTEXTUAL ADVERTISEMENT REFLECTING USER INTEREST}

The present invention relates to a mobile device and a method for providing a contextual advertisement reflecting a user's interest.

Recently, as the spread of mobile devices increases, the mobile advertisement market providing advertisements to users through mobile devices is growing. These mobile advertisements are mainly made in the form of in-app advertisements that provide advertisements inside the application of the mobile device, and recognize the context of the user from the physical situation such as the user's location and time and expose the related advertisement. The method is used.

Meanwhile, as one method for effectively providing advertisements, interest in contextual advertisements is increasing. The contextual advertisement provides advertisement content related to a web page being used by a user, and there are syntactic methodology and semantic methodology as a method for searching for such advertisement content. Syntactic methodology is a method of measuring similarity by simply matching words included in web pages and advertisement content, and semantic relations are not taken into account. As a way to compensate for these limitations, a semantic method that considers the topical similarity between pages and advertisements has been proposed. However, these semantic methods have not been commercialized due to limitations such as limited resources of mobile smart devices.

In particular, most in-app advertisements provide contextual advertisements using the user's physical context, as described above. This is limited in that the same advertisement content is provided to people in the same physical situation. That is, even if they are located at the same time and region, each interest is not considered different.

Meanwhile, in order to provide a contextual advertisement reflecting the user's interest, the user's personal information must be transmitted to the advertisement providing server. This has room for problems in protecting the user's information.

Republic of Korea Registered Patent Publication No. 10-1252245 (Name of invention: subject classification module and contextual advertising system using the same)

The present invention is to solve the problems of the prior art described above, and provides a context advertisement providing method and a mobile device for providing a context advertisement content based on a user's intention by a mobile smart device without access to an external server. However, the technical problem to be achieved by the present embodiment is not limited to the technical problem as described above, and other technical problems may exist.

As a technical means for achieving the above-described technical problem, the mobile device according to the first aspect of the present invention includes a display; And a processor storing a program providing advertisement content, and a processor executing the program stored in the memory. The processor determines user interest from user content stored in the memory based on a pre-stored topic classification tree according to the execution of the program, and when a request to display an advertisement in an application currently being executed is obtained, based on the topic classification tree Determine a category of application content, determine each category of at least one advertisement content previously obtained based on a topic classification tree, and use user interest, similarity between a category of application content and a category of advertisement content, and One advertisement content among advertisement content is output from the display. At this time, the subject classification tree includes a hierarchical category calculated from an open directory project.

In addition, a method for providing contextual advertisement of a mobile device according to a second aspect of the present invention includes determining a user interest from pre-stored user content on a mobile device based on a pre-stored subject classification tree; Determining a category of application content based on a topic classification tree when an ad serving request in the currently running application is obtained; Determining each category of at least one advertisement content previously obtained based on a topic classification tree; And outputting one advertisement content among at least one advertisement content by using user interest, a category of application content, and similarity between each category of advertisement content. At this time, the subject classification tree includes hierarchical categories calculated from the open directory project.

In addition, a third aspect of the present invention includes a computer readable recording medium recording a program for performing the method of the second aspect on a computer.

The present invention can provide a contextual advertisement reflecting the interest of the user of the mobile device and the currently running application content. At this time, the present invention can determine the user's interest level internally by the mobile device, so that the user's personal information is not exposed.

1 is an exemplary view for explaining a conventional method for providing advertising content.
Figure 2 shows the configuration of a mobile device according to an embodiment of the present invention.
3 is an example of a subject classification tree stored in a subject classification tree storage module according to an embodiment of the present invention.
4 is a flowchart illustrating a method of providing advertisement content by a control unit according to an embodiment of the present invention.
5 illustrates an algorithm for extracting at least one advertisement content that is semantically similar to a category of application content among at least one candidate advertisement content extracted based on a user's interest according to an embodiment of the present invention.
6 is an example in which context advertisement is provided according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art to which the present invention pertains may easily practice. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. In addition, in order to clearly describe the present invention in the drawings, parts irrelevant to the description are omitted, and like reference numerals are assigned to similar parts throughout the specification.

Throughout the specification, when a part is "connected" to another part, this includes not only "directly connected" but also "electrically connected" with another element in between. . Also, when a part is said to "include" a certain component, this means that other components may be further included rather than excluding other components, unless otherwise stated.

1 is an exemplary view for explaining a conventional method for providing advertising content. Conventional advertisement content providing method is provided by extracting the related content from the content that the user selects or uses through the user terminal. At this time, the use may be that the user purchases, subscribes, creates, or modifies the content.

Specifically, the conventional advertisement content providing method provides a contextual advertisement by extracting a category (or subject) with respect to nouns included in the content that the user is using. For example, referring to FIG. 1, the conventional advertisement content providing method extracts a category from content C consumed by a first user U1 interested in soccer. For example, the conventional method may classify the content used by the first user into a sports category, and deliver advertisement content related to the content to the first user U1.

In addition, the conventional method extracts a category from content C consumed by a second user U2 interested in baseball. At this time, since the first user U1 and the second user U2 use the same content C, the same advertisement content Ad may be provided to the first user U1 and the second user U2. .

As described above, the conventional method provides the advertisement content in consideration of the category of the content that the user is using, but the same advertisement content is provided to the first user and the second user having different interests, thereby reducing the efficiency of the advertisement content. That is, the conventional method has limitations in determining the user's intention and providing customized advertisement content suitable for the user's intention.

Figure 2 shows the configuration of a mobile device 100 according to an embodiment of the present invention. Referring to FIG. 2, the mobile device 100 includes a communication unit 110, a display unit 120, a memory 130, and a control unit 140.

The communication unit 110 includes at least one component for the mobile device 100 to communicate with other devices and servers. As a non-limiting example, the communication unit 110 includes at least one of a mobile communication chip (not shown), a Wi-Fi chip (not shown), a Bluetooth chip (not shown), a wireless communication chip (not shown), and an NFC chip (not shown). It may include.

The display 120 outputs various information processed by the mobile device 100. The display unit 120 may include a display panel (not shown) and a controller (not shown) that controls the display panel. The display panel may be implemented as various types of displays such as a liquid crystal display (LCD), organic light emitting diodes (OLED) display, active-matrix organic light-emitting diode (AM-OLED), and plasma display panel (PDP). . Also, the display 120 may be combined with a touch panel (not shown) and provided as a touch screen (not shown). For example, a touch screen (not shown) may include an integrated module in which a display panel and a touch panel are combined in a stacked structure.

The memory 130 stores a program that provides advertisement content based on a user's interest level. The program stored in the memory 130 is executed by the control unit 140. Also, the memory 130 temporarily or continuously stores various pieces of information required for the program to be performed. For example, the memory 130 may include a subject classification tree storage module 131, a user content storage module 132 in which content used by a user of the mobile device 100 (eg, generated, read, modified, etc.) is stored, Advertisement content storage module 133 for storing at least one advertisement content obtained from an advertisement providing server (not shown) or an application providing server (not shown), and an application content storage module for storing application content as the application is executed (134) and the like.

First, the topic classification tree storage module 131 includes a topic classification tree including categories processed from an open directory project. Here, the open directory project subdivides text such as web documents on the world wide web into categories using hierarchical ontology, and organizes the tree data structure. It is classified using. In addition, the subject classification tree includes central vectors based on a vector space model based on words representing hierarchical categories of the open directory project. Also, the subject classification tree may be stored in the form of a mobile device file system or a database (eg, SQLite). In addition, the subject classification tree may be generated and provided by a web search engine, but may be generated by the controller 140 based on web documents collected on the mobile device 100, but is not limited thereto.

Equation 1 below defines a central vector according to an embodiment of the present invention.

<Equation 1>

In the above equation, v denotes the dimension of the central vector (c) as the number of words. And the following equation (2) represents the word weight (w) that is a component of each word vector.

<Equation 2>

는 단어 t의 가중치를 나타낸다. 또한, tf는 웹 문서 등에 포함된 단어 t의 빈도수를 나타내며, N는 전체 웹 문서 등의 수를 나타내며, df는 전체 웹 문서 등에서 단어 t가 발생한 웹 문서 등의 빈도수를 나타낸다. In the above equation, t represents the word,

Denotes the weight of the word t. In addition, tf represents the frequency of the word t included in the web document, etc., N represents the total number of web documents, etc., and df represents the frequency of the web document and the like in which the word t occurs in the entire web document.

3 is an example of a topic classification tree stored in the topic classification tree storage module 131. For example, when a category of specific content is included in a “MLB” node and a “Major league” node, a category of target content may extract “MLB” and “Major league”. At this time, the central vector is calculated based on the subject classification tree and may be a path from a root node of the subject classification tree to a node including the target content. For example, when the target content is included in the "MLB" node of the subject classification tree, the processor may extract all categories included in the parent node or ancestor node of the "MLB" node in the subject classification tree. And, according to the hierarchical structure of the subject classification tree, the central vector of the category corresponding to the target content may be "top/sport/baseball/MLB". Similarly, if the target content is included in the "Major league" node of the subject classification tree, the central vector of the category corresponding to the target content is "top/sport/baseball/Major league".

Meanwhile, the memory 130 is a non-volatile storage device that keeps stored information even when power is not supplied and a volatile storage device that requires power to maintain the stored information.

Referring to FIG. 2 again, the control unit 140 may be implemented by including at least one processing unit (CPU, micro-processor, DSP, etc.), random access memory (RAM), read-only memory (ROM), and the like. , The advertisement content providing program stored in the memory 130 may be read into RAM and executed through at least one processing unit. Meanwhile, according to an embodiment, the term “control unit” may be interpreted in the same meaning as terms such as “processor”, “controller”, and “computation device”.

Hereinafter, FIG. 4 is a flowchart illustrating a method of providing advertisement content by the control unit 140 according to an embodiment of the present invention.

The controller 140 determines user interest from the user content stored in the user content storage module 132 based on the subject classification tree stored in the subject classification tree storage module 131 (S410). At this time, the subject classification tree may be preset according to the above-described process.

More specifically, the control unit 140 collects user content (eg, text content used by the user, text content of a web page accessed by the user, a list of installed applications, a list of downloaded files, etc.) previously stored in the mobile device 100. do. In addition, the control unit 140 identifies individual words constituting the title, description, and body of the collected user content, and calculates a word vector using the weight of each word. The word vector may be calculated in the same form as Equation 1 (ie, the central vector of the categories included in the subject classification tree). Also, a method of identifying words from user content will be described in detail with reference to FIG. 7.

Next, the controller 140 calculates the similarity between the central vector of each category included in the topic classification tree stored in the topic classification tree storage module 131 and the word vector extracted from user content. At this time, the control unit 140 may be calculated based on cosine similarity, and may be expressed by Equation 3 below.

<Equation 3>

In the above equation, the vector user represents a word vector of user content, and vector c represents a central vector of categories in the subject classification tree. However, in addition to the above-described method, the control unit 140 may calculate weights using a jaccard coefficient, a dice coefficient, and a humming distance, but is not limited thereto.

The control unit 140 determines a set of at least one category of the subject classification tree indicating the similarity of the first threshold value or more as the user's interest.

Subsequently, when a request to display an advertisement in the currently running application is obtained, the controller 140 infers the category of the application content based on the subject classification tree (S420). For example, the controller 140 may obtain application content currently being displayed on the screen of the mobile device 100 among a plurality of application contents constituting an application running in the foreground. At this time, the control unit 140 may receive text content among application contents using a predetermined application programming interface (API).

The controller 140 identifies words included in the received application content and calculates a word vector of the application content. At this time, since the method of calculating the word vector of the application content is the same or similar to the method of calculating the central vector of the categories of the above-described subject classification tree, detailed description is omitted.

Next, the control unit 140 calculates the similarity between the central vector of each category included in the topic classification tree stored in the topic classification tree storage module 131 and the word vector calculated from the application content stored in the application content storage module 134. do. This can be expressed by the following equation (4).

<Equation 4>

The control unit 140 determines a set of at least one category of the subject classification tree indicating the similarity of the second threshold or more as a category of application content.

Thereafter, the control unit 140 determines a category of each advertisement content stored in the advertisement content storage module 133 (S430). As in step S420 described above, the controller 140 identifies words included in each advertisement content and calculates a word vector of each advertisement content. At this time, the method of calculating the word vector of the advertisement content is the same or similar to the method of calculating the central vector of the categories of the above-described subject classification tree, so a detailed description is omitted.

Next, the control unit 140 calculates the similarity between the central vector of each category included in the topic classification tree stored in the topic classification tree storage module 131 and the word vector calculated from advertisement content. This can be expressed as Equation 5 below.

<Equation 5>

The control unit 140 determines a set of at least one category of the subject classification tree indicating the similarity of the third threshold or more as a category of advertisement content. Meanwhile, the first to third threshold values may be experimentally determined values, but are not limited thereto.

Thereafter, the control unit 140 outputs one advertisement content based on the degree of similarity between the user interest, the category of the application content, and the category of the advertisement content (S440).

For example, the control unit 140 calculates a degree of similarity between the user's interest and the category of advertisement content, and extracts at least one candidate advertisement content that is semantically similar to the user's interest. For example, the controller 140 may calculate the degree of similarity between the user's interest and the category of advertisement content through Equation 6 below, and extract at least one candidate advertisement content having a degree of similarity equal to or greater than a preset value.

<Equation 6>

및

는 상기한 수학식 3 및 수학식 5를 통해 산출된 유사도 값을 나타낸다. 또는, cw는 기계학습(machine learning) 알고리즘 또는 분류(classification) 기법에 기초하여 생성된 분류기를 통하여 사용자의 관심도와 광고 콘텐트의 카테고리 간 유사도를 산출하는 함수 일 수 있다. In the above equation, UC represents a set of categories corresponding to a user's interest level, and AC represents a set of categories of advertisement content. In addition,

And

Denotes a similarity value calculated through Equation 3 and Equation 5. Alternatively, cw may be a function for calculating similarity between a user's interest and a category of advertisement content through a classifier generated based on a machine learning algorithm or a classification technique.

(weighted personalized PageRank)는 사용자 관심도에 포함된 (주제 분류 트리의) 카테고리(uc)와 각 광고 콘텐트의 카테고리(ac) 간의 의미적 연관성을 산출하는 함수로서, 등록특허 제10-1083274호(발명의 명칭: 유사도 그래프를 활용한 문맥 광고 시스 템)에 기재된 의미적 연관성 산출부를 사용할 수 있다.In addition,

(weighted personalized PageRank) is a function that calculates the semantic association between the category (uc) (of the subject classification tree) included in the user's interest and the category (ac) of each advertisement content. Name: The contextual advertisement system using the similarity graph) can be used.

Next, the control unit 140 outputs a predetermined number of advertisement contents that are semantically similar to the topic of the application contents from at least one candidate advertisement content.

5 illustrates an algorithm for extracting at least one advertisement content that is semantically similar to a category of application content among at least one candidate advertisement content extracted based on a user's interest according to an embodiment of the present invention. Through the algorithm illustrated in FIG. 5, the control unit 140 for each advertisement content included in the set of candidate advertisement content (UA) based on the user's interest level, the category (AC) of each advertisement content and the category of application content The score of the intersection category (UC ∩ AC) between (UC) is calculated. Here, since the score may be similarity calculated using Equation 6, detailed description is omitted. Subsequently, the controller 140 sorts each advertisement content based on the score. That is, the advertisement content positioned at a higher rank among the aligned advertisement contents is related to both user interest and application content.

이다. 여기서, β는 사용자 관심도에 대한 가중치 값으로서, 0 과 1 사이의 값일 수 있다. The control unit 140 determines an upper predetermined number of advertisement contents among advertisement contents arranged based on the score. For example, when k advertisement contents are determined, the number of advertisement contents related to the user's interest is

to be. Here, β is a weight value for user interest, and may be a value between 0 and 1.

The control unit 140 may output the advertisement content having the highest rank among the determined k advertisement contents. Depending on the implementation, the controller 140 may convert and output k advertisement contents at regular time intervals.

6 is an example in which context advertisement is provided according to an embodiment of the present invention. As illustrated in FIG. 6, the mobile device 100 according to an embodiment of the present invention determines candidate advertisement content based on a user's interest level, and then advertising content from one of the candidate advertisement contents based on a category of application content By outputting, it is possible to provide different advertisement contents 620 and 630 even when different users U1 and U2 use the same application content 610.

One embodiment of the present invention can also be implemented in the form of a recording medium including instructions executable by a computer, such as program modules, being executed by a computer. Computer readable media can be any available media that can be accessed by a computer and includes both volatile and nonvolatile media, removable and non-removable media. In addition, computer readable media may include computer storage media. Computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data.

Although the methods and systems of the present invention have been described in connection with specific embodiments, some or all of their components or operations may be implemented using a computer system having a general purpose hardware architecture.

The above description of the present invention is for illustration only, and those skilled in the art to which the present invention pertains can understand that the present invention can be easily modified into other specific forms without changing the technical spirit or essential features of the present invention. will be. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as distributed may be implemented in a combined form.

The scope of the present invention is indicated by the following claims rather than the above detailed description, and it should be interpreted that all changes or modified forms derived from the meaning and scope of the claims and equivalent concepts thereof are included in the scope of the present invention. do.

Claims (11)

In a mobile device providing contextual advertising content,
display;
A memory in which a program providing advertisement content is stored, and
A processor for executing a program stored in the memory,
The processor according to the execution of the program,
Based on a pre-stored subject classification tree, a user interest is determined from the user content stored in the memory, and when a request to display an advertisement in the currently running application is obtained, among a plurality of application contents constituting the running application in the foreground, Acquiring application content being output on the screen of a mobile device, identifying words included in the application content, calculating a word vector using the weight of each word,
By calculating the similarity between the central vector of the subject classification tree and the word vector, a category of the application content is determined, and each category of at least one advertisement content previously obtained based on the subject classification tree is determined,
Calculate the similarity between the user interest and the category of each advertisement content, extract at least one candidate advertisement content having a similarity of more than a predetermined value, and the category of each advertisement content included in the candidate advertisement content and the category of the application content Calculates the similarity of the intersection set between
The advertisement content of the highest rank among the at least one candidate advertisement content sorted based on a score according to the similarity and a weight value for user interest is output,
The score is calculated by the following equation (1),
The subject classification tree includes a hierarchical category calculated from an open directory project.
[Equation 1]

uc: a set of categories that correspond to the user's interest
ac: a set of categories of advertising content
cw(uc), cw(ac): Similarity values calculated by the preset similarity algorithm
wPPR (weighted personalized pagerank): A function that calculates the semantic association between the category (uc) of the subject classification tree included in user interest and the category (ac) of each advertisement content According to claim 1,
The processor
The user content is collected, words included in the user content are identified, a word vector is calculated using the weight of each word, and the similarity between the center vector and the word vector of each category included in the subject classification tree is determined. By calculating, the mobile device is to determine the category of the subject classification tree having similarity above a threshold value as the user interest. delete According to claim 1,
The processor
Determining the category of the advertisement content by identifying words included in each advertisement content, calculating a word vector using the weight of each word, and calculating the similarity between the center vector of the subject classification tree and the word vector Mobile devices. delete A method for a mobile device to provide contextual advertising,
Determining a user interest from user content stored in the mobile device based on a pre-stored subject classification tree;
Determining a category of application content based on the topic classification tree when an advertisement serving request in the currently running application is obtained;
Determining each category of at least one advertisement content previously obtained based on the subject classification tree; And
And outputting one advertisement content among the at least one advertisement content by using the similarity between the user interest level, the category of the application content, and each category of the advertisement content,
Determining the category of the application content is
Obtaining, among the plurality of application contents constituting the application running in the foreground, the application contents currently being displayed on the screen of the mobile device;
Identifying words included in the application content, and calculating a word vector using the weight of each word; And
Determining a category of the application content by calculating a similarity between the center vector of the subject classification tree and the word vector,
The step of outputting the advertisement content is
Calculating the similarity between the user interest and the category of each advertisement content, and extracting at least one candidate advertisement content having a similarity level higher than a preset value;
Calculating the similarity of the intersection category between the category of each advertisement content included in the candidate advertisement content and the category of the application content; And
And outputting advertisement content of a higher rank among the at least one candidate advertisement content sorted based on a score according to the similarity and a weight value for user interest,
The score is calculated by the following equation (1),
The subject classification tree includes a hierarchical category calculated from an open directory project.
[Equation 1]

uc: a set of categories that correspond to the user's interest
ac: a set of categories of advertising content
cw(uc), cw(ac): Similarity values calculated by the preset similarity algorithm
wPPR (weighted personalized pagerank): A function that calculates the semantic association between the category (uc) of the subject classification tree included in user interest and the category (ac) of each advertisement content The method of claim 6,
Determining the user interest is
Collecting the user content, identifying words included in the user content, and calculating a word vector using the weight of each word; And
A method for providing contextual advertisement, wherein a category of the subject classification tree having a similarity of a threshold or higher is determined as the user interest by calculating the similarity between the central vector of the subject classification tree and the word vector. delete The method of claim 6,
Determining the category of the advertising content is
Identifying words included in each advertisement content and calculating a word vector using the weight of each word; And
And determining a category of each advertisement content by calculating the similarity between the center vector of the subject classification tree and the word vector. delete A computer-readable recording medium recording a program for performing the method according to claim 6 on a computer.

Images