KR20110063160A - Apparatus and method for providing advertisement - Google Patents

Abstract

PURPOSE: An apparatus and method for providing an advertisement are provided to offer an advertisement corresponding to the interest of a user by determining the interest of a user. CONSTITUTION: An ontology construction unit(210) constructs a relationship between subordinates and superiors. The ontology construction unit establishes the levels of product terms. The ontology construction part collects advertisement content about the product terms. A keyword processor(220) calculates an average level and a maximum level using the levels of the product terms. An interest change determining unit(230) determines a change of the interest level using the average level.

Description

Apparatus and method for providing advertisement

The present invention relates to an advertisement providing apparatus and method, and more particularly, to an advertisement providing apparatus and method for providing an advertisement according to the user's interest.

Recently, with the rapid increase of Internet users, the Internet is playing a role as a new media that enables real-time communication between companies and individuals, individuals and individuals rather than a network connected with computers.

Due to the popular use of the Internet, the exchange among Internet users is expanding. Many Internet business companies have opened various sites to provide users with various services such as information exchange, games, entertainment or e-commerce.

On the other hand, the Internet is in the limelight as a huge and efficient information storage medium and information providing medium connected with numerous computer networks around the world. Thus, as the number of networks connected to the Internet and the number of Internet users have soared, interest in various advertisements on the Internet has increased.

That is, the representative of the services using the Internet is the advertising service. Advertising services using the Internet are less expensive than offline advertising services and it is easy to grasp the advertising preferences after advertisements are displayed. In particular, the size of the advertising service using the Internet can be intensively advertised to a specific user class and its size is continuously expanding.

In general, Internet advertising has been displayed in the form of banners independently of the content. In other words, the Internet advertisement was displayed as content irrelevant to the content. Such internet advertising has a problem that it is difficult to draw the user's attention away from the content focused on the user.

In order to solve this problem of Internet advertising, internet advertising using keywords has been used. An internet advertisement using a keyword is an internet advertisement in which an advertisement corresponding to a keyword corresponding to a user's search word is displayed.

However, in the conventional Internet advertising, not only one website may not have advertisements corresponding to various keywords inputted by users, but also there is a problem that it is difficult to draw user's attention because the level of automatic analysis of search terms is weak.

An object of the present invention is to determine a user's interest change using a user's search word, and to provide an advertisement in accordance with the user's interest change.

According to an aspect of the present invention, an advertisement providing apparatus for providing an advertisement according to a search word input by a user is disclosed.

The advertisement providing apparatus according to the embodiment of the present invention establishes a vertical relationship between a plurality of product terms according to an upper and lower concept by using an ontology hierarchy structure, sets the level of the product term according to an ontology hierarchy, and An ontology construction unit for collecting advertisement content and linking the product term and the advertisement content, extracting a keyword from the search term to generate a keyword set, and using the level of the product term corresponding to the keyword of the keyword set to an average level And a keyword processor for calculating a maximum level, an interest change determination unit for determining a change in interest using the average level, and a product term selected according to the change in interest, and detecting advertisement content connected to the selected product term to perform an advertisement. It includes an advertising content detection unit.

According to another aspect of the present invention, an advertisement providing method performed by an advertisement providing apparatus for providing an advertisement according to a search word input by a user is disclosed.

According to an embodiment of the present invention, a method of providing an advertisement includes a) constructing an up and down relationship of a plurality of product terms according to an upper and lower concept using an ontology hierarchy, and b) setting a level of the product term according to an ontology hierarchy. c) collecting advertisement content for the product term and linking the product term with the advertisement content, d) generating a first set of keywords from the search term entered by the user, and generating a second keyword from the search term entered again. Generating a set, e) first and second average levels for the first and second keyword sets and first and second using the levels of the product terms corresponding to the keywords in the first and second keyword sets; Calculating a second maximum level, f) determining a change in interest using the first and second average levels, and g) selecting a product term according to the change in interest. And, detecting the advertisement content associated with the selected product terms comprises the step of performing the ad.

The present invention has an effect of determining a change in interest of a user using a user's search word and providing an advertisement according to the change in interest of the user.

In addition, the present invention has an effect that can increase the effect of advertising by providing the user with a high interest of the user.

As the invention allows for various changes and numerous embodiments, particular embodiments will be illustrated in the drawings and described in detail in the written description. However, this is not intended to limit the present invention to specific embodiments, it should be understood to include all transformations, equivalents, and substitutes included in the spirit and scope of the present invention. In the following description of the present invention, when it is determined that the detailed description of the related known technology may obscure the gist of the present invention, the detailed description thereof will be omitted.

The terms first, second, etc. may be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another.

The terminology used herein is for the purpose of describing particular example embodiments only and is not intended to be limiting of the present invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this application, the terms "comprise" or "have" are intended to indicate that there is a feature, number, step, operation, component, part, or combination thereof described in the specification, and one or more other features. It is to be understood that the present invention does not exclude the possibility of the presence or the addition of numbers, steps, operations, components, components, or a combination thereof.

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

1 is a configuration diagram schematically illustrating an advertisement providing system according to an embodiment of the present invention.

Referring to FIG. 1, an advertisement providing system according to an exemplary embodiment of the present invention may include a user terminal 10, a web server 20, an advertisement providing apparatus 30, and a database 40.

The user terminal 10 may be provided with a search service by accessing the web server 20 through a wired or wireless communication network. That is, the user terminal 10 may be an electronic device capable of performing a search service through a wired or wireless internet. Here, the user terminal 10 may include a mobile communication terminal (for example, a smartphone, etc.) capable of performing an Internet connection using a user PC or a wireless network connected to a wired network.

The web server 20 is a server that provides various services through the Internet. For example, the web server 20 may include a search site that provides a search service, a portal site that provides various services (for example, mail, shopping, news, games, music, etc.).

The web server 20 according to an embodiment of the present invention may receive a user's search word and display corresponding information or provide a service. In this case, the web server 20 may transmit a user's search word to the advertisement providing apparatus 30 and display the advertisement content delivered by the advertisement providing apparatus 30.

The advertisement providing apparatus 30 may receive a user's search word from the web server 20 and use the same to perform an advertisement according to the user's interest.

That is, the advertisement providing apparatus 30 may calculate a change in interest of the user by tracking a change in a search word input by the user. Thus, the advertisement providing device 30 may perform advertisement by detecting advertisement content corresponding to the change in interest.

For example, a user may repeat input of a search word and check of a search result when searching the Internet. In this case, the advertisement providing apparatus 30 may calculate a change in interest of the user according to the change of the search word before and after using the repeated search word input. In addition, the advertisement providing apparatus 30 may construct an ontology of a product term for a specific product, collect advertisement content corresponding to each product term, and connect the product term. Here, the advertisement providing device 30 may store the collected advertisement content in the database 40. Thus, the advertisement providing apparatus 30 may select a product term according to a change in the interest of the user and perform an advertisement with advertisement content connected to the selected product term.

The advertisement providing apparatus 30 according to the embodiment of the present invention may be implemented as a device connected to the web server 20 as shown in FIG. 1, or may be included in the web server 20.

The database 40 may store advertisement content collected by the advertisement providing device 30. In addition, the database 40 may function as a storage of the advertisement providing apparatus 30 and may be included in the advertisement providing apparatus 30.

2 is a block diagram showing an advertisement providing apparatus according to an embodiment of the present invention, Figure 4 is an exemplary view showing an ontology hierarchy structure according to an embodiment of the present invention. Hereinafter, an advertisement providing apparatus according to an exemplary embodiment of the present invention will be described with reference to FIG. 2 to describe the present invention with reference to FIG. 4.

Referring to FIG. 2, the advertisement providing apparatus 30 may include an ontology constructing unit 210, a keyword processing unit 220, an interest change determining unit 230, and an advertisement content detecting unit 240.

The ontology constructing unit 210 may construct an isA relationship between a plurality of product terms according to an upper and lower concept using an ontology hierarchy. Here, the ontology construction unit 210 may build a hierarchical structure in a tree structure. The ontology hierarchy may have the form of a graph. Graphs are different from trees in that they can have more than one parent node. In the embodiment of the present invention, since the ontology hierarchy is a tree, the ontology hierarchy is converted into a tree structure when the ontology hierarchy is in the form of a graph, which includes nodes having a plurality of parent nodes and descendant nodes of the nodes. This can be done by copying and configuring the subtree of each parent node.

Hereinafter, a method of constructing an ontology of a product term by using a product as an automobile will be described.

For example, the ontology hierarchy structure for a vehicle may be expressed as shown in Table 1 below. Here, referring to FIG. 4, the ontology hierarchy of Table 1 may be expressed as shown in FIG. 4.

TABLE 1

car car SUV Sports car Compact car Medium car Large car - - H1 tea G1 tea H2 tea G2 car H3 tea G3 tea

Here, the ontology construction unit 210 may set a level and a node number for each product term. That is, the ontology construction unit 210 may set the level for each product term to be the root node level 1, and increase by one for the parent-child relationship from the upper level to the lower level. In addition, the ontology construction unit 210 may set node numbers on the ontology hierarchy for each product term in order from upper level (lower level) to lower level (higher level) and from left to right.

For example, the ontology construction unit 210 may manage the ontology hierarchy by setting the ontology hierarchy in a table as shown in Table 2 below.

TABLE 2

Node number Product terms level Parent node number Child node number One car One - 2, 3, 4 2 car 2 One 5, 6, 7 3 SUV 2 One - 4 Sports car 2 One - 5 Compact car 3 2 8, 9 6 Medium car 3 2 10, 11 7 Large car 3 2 12, 13 8 H1 tea 4 5 - 9 G1 tea 4 5 - 10 H2 tea 4 6 - 11 G2 car 4 6 - 12 H3 tea 4 7 - 13 G3 tea 4 7 -

Here, the parent node number and the child node number are the node number of the parent node of the upper level and the node number of the child node of the lower level, respectively, based on the corresponding product term.

The ontology construction unit 210 may define a description of each product term, and extract a keyword therefrom. The ontology constructing unit 210 may generate a term vector using keywords.

For example, the ontology construction unit 210 may define a description for each product term as shown in Table 3 below. Here, the description of each product may be collected from a dictionary, the Internet, a database, or the like.

 [Table 3]

car Vehicles that transport people or cargo on the road car A vehicle that transports people on the road SUV Vehicles that drive on rough roads and are suitable for leisure Sports car Fast racing car on the road Compact car Passenger cars with engine displacement of 1,600cc or less H2 Medium passenger car made by H G1 Small passenger car made by G car … …

The ontology construction unit 210 extracts a keyword from the description of each product term. Here, the ontology construction unit 210 may use a keyword extraction method through morpheme analysis, and may use a keyword dictionary and a stop word dictionary in morpheme analysis. Since a method of extracting keywords through morphological analysis is a well-known technique, a detailed description thereof will be omitted.

In addition, the ontology constructing unit 210 may assign serial numbers to all extracted keywords and generate positions of elements of the term vector in the order of the serial numbers. Here, the element of the term vector may be indicated by the presence or absence of a keyword in the description of the corresponding product term.

For example, the ontology construction unit 210 may select a car, a road, a driving, a person, a transport, or a vehicle as a keyword for describing the car. In addition, the ontology construction unit 210 displays an element value as 1 (if present) or 0 (none) according to the presence or absence of the corresponding keyword at the position of the corresponding keyword (keyword in the description of the car) of the term vector. You can create a vector.

For example, the ontology construction unit 210 may generate a term vector for a car as (0, 1, 0, 1, 1, 0, 0, ..). Here, the number of elements in the term vector is the total number of keywords extracted from all product terms. In the term vector, an element having a value of 1 is an element having a corresponding keyword in the description of a car. The position of an element corresponding to each keyword in the term vector may be a position corresponding to the serial number of the keyword. For example, if the total number of keywords is n (where n is an integer), then the number of elements in the vector is n, and the keywords are given a serial number from 1 to n, and k (where k is an integer, 1 The <= k <= n) th keyword may be mapped to the kth element of the vector. In the example of the term vector, if the k-th keyword exists, the value of the k-th element of the vector may be set to 1, and if not, the value of the k-th element may be set to 0.

In another embodiment, the ontology constructing unit 210 may set the value of the element of the term vector to the frequency (number of times) of occurrence of the keyword.

For example, a plurality of identical keywords may exist in the description of a product term. In this case, the ontology construction unit 210 may set the frequency (number) of appearance of the keyword as an element value of the term vector.

In another embodiment, the ontology construction unit 210 may set an element value (weight of the keyword) of the term vector by varying the adjustment coefficient according to the importance of the keyword.

For example, the ontology construction unit 210 may set the value (weight) of an element of the term vector using Equation 1 below.

[Equation 1]

The value of the kth element of the term vector = Fk * Wk

Here, Fk is the frequency of the kth element (keyword), and Wk is the adjustment coefficient of the kth element. In one example, the adjustment coefficient may be the inverse literature frequency of the keyword.

The ontology construction unit 210 may extract the number of documents (product term descriptions) in which each keyword appears. From this, the inverse document frequency (IDF) of each keyword can be calculated. Backlit frequency is the log value obtained by dividing the total number of documents by the number of documents in which the keyword appears, so that keywords in a small number of documents have a higher value than keywords in a large number of documents. Indicates the scarcity of the keyword. In addition, there are various methods of converting the extracted frequency for each keyword into an element value of a vector corresponding to the keyword by using this as an adjustment coefficient. For example, a TF-IDF (Term Frequency-Inverse Document Frequency) weighting method using keyword frequency and reverse document frequency can be given. Therefore, the value of the element of the vector may be a weight by TF-IDF.

The present invention is applied to all cases of using the presence or absence of a keyword extracted as an element value of a vector as it is, or converting a keyword frequency extracted using a reverse literature frequency or the like into a weight of a keyword in various ways. Since the vector can be generated, it is possible to select and use among various known techniques. On the other hand, the term 'element value' in the vector used in the present invention may be replaced with the 'weight of keyword' used in some documents.

The ontology construction unit 210 may collect advertisement content for each product term and connect the product term and the advertisement content. Here, the ontology builder 210 may store the collected advertisement content in a storage unit (not shown). In this case, the ontology construction unit 210 may store the advertisement content in a storage unit provided in the advertisement providing device 30 or store the advertisement content in a database 40 connected to the outside. Here, the collected advertisement content may be advertisement content provided by an advertiser or an advertising agency, or may be advertisement content retrieved from the Internet or a database in which the advertisement content is stored. The advertisement content may be not only text advertisements but also various types of contents such as image advertisements and multimedia advertisements such as video advertisements.

The ontology construction unit 210 selects a product term that is the subject of each collected advertisement content in the following manner. The ontology construction unit 210 may extract keywords from the collected advertisement content from texts such as titles (titles), advertisement text, advertisement lines, descriptions, and annotations of each advertisement content, and may also describe techniques such as image processing and image processing. A keyword may be extracted from an image or a video of the advertisement content by using. Product terms that are the subject of the advertisement content may be selected using the extracted keywords as follows.

The ontology construction unit 210 may generate the content vector by using the presence or frequency or weight of the keyword in the same manner as the method of generating the term vector from the keywords extracted for each advertisement content. In addition, a product term corresponding to the term vector most similar to the content vector may be selected by comparing the similarity between the content vector and each term vector. Here, the method of comparing the similarity may use the Euclidean distance formula (Equation 3) or the cosine formula (Equation 4) described later, and may also use the similarity comparison method between various other vectors. Alternatively, when a product term appears in the extracted keyword, the product term with the highest frequency among the product term may be selected as the product term that is the subject of the advertisement content. This is because the keyword with the highest frequency is very likely to be the main word. Alternatively, one of the extracted keywords may select one of the product terms existing in the keyword extracted from the title (title) of the document. This is because the title of a document is often related to a subject. Accordingly, the present invention can be selected and used from various methods for selecting a product term that is the subject of the advertisement content.

The ontology construction unit 210 may connect each product term and the corresponding advertisement content by the product term of each selected advertisement content. That is, each product term may be made and linked to a corresponding advertisement content list. The advertising content list may be a list (list) of advertising content identification numbers (IDs) or may be a list of addresses of advertising content. In addition, advertisement content may not exist for each product term, and one or more may exist.

In the ontology construction unit 210, the advertisement content linked to each product term is connected to the abstract image advertisement content with respect to the product term at the upper level, and the specific product advertisement content is linked with the product term at the lower level. This is because the lower (high) level product terms are more specific product terms than the upper (lower) level terms in the ontology hierarchy.

For example, if the product term is a car, the ontology building unit 210 is an image advertisement of the manufacturing company associated with the car (for example, the car sales No. 1 OOO, the safety test No. 1 OOO to improve the image of the automobile company) , Advertising content for the world's top export 000, etc.) can be collected and linked. And, if the product term is the name of the car (for example, H2 car, G2 car, etc.), the ontology construction unit 210 may collect and connect the product advertisement of the car. At this time, the ontology construction unit 210 may store the collected advertisement content directly in the storage unit or the external database 40, and set the ID of the advertisement content as shown in Table 4 below and manage it as a table. Alternatively, if the ontology construction unit 210 does not store the advertisement content in the storage unit or the database 40 directly, the ontology construction unit 210 may manage the Internet address where the advertisement content is located in a table.

[Table 4]

Product terms List of advertising content IDs (or addresses) car 100, 101 SUV 105, 106 … H2 tea 301 G1 tea 210

The keyword processor 220 may generate a keyword set and calculate an average level and a maximum level of the keyword using the generated keyword set.

The keyword processing unit 220 may generate a keyword set by extracting a keyword from a search word input by the user. Here, the keyword processor 220 may generate a first keyword set and a second keyword set to determine a change in interest of the user. That is, the keyword processing unit 220 may generate the first keyword set from the search word input by the user, and then generate the second keyword set from the search word input by the user.

In addition, the keyword processor 220 may generate the first and second keyword sets from documents provided as a search result as well as a search word input by a user. Alternatively, the keyword processing unit 220 may generate the first and second keyword sets from documents read and clicked by the user among search terms input by the user and documents provided as a search result. Therefore, the second keyword set is a keyword set generated later in time (in time) than the first keyword set.

The keyword processor 220 may generate a first ontology term and a second ontology term by extracting a keyword corresponding to a product term from the first and second keyword sets. Here, the product term may be a product term constructed by the ontology constructing unit 210. Since the first and second ontology terms are extracted from the first and second keyword sets, the first and second ontology terms may be extracted when the product terms are included in both the first and second keyword sets. For example, the keyword processor 220 may not generate the first and second ontology terms when the product term is not included in any one of the first and second keyword sets (first and second ontology terms). Is a comparison target). In this case, a change in the degree of interest of the user may be determined by generating a similar ontology term, which will be described later. Therefore, a method of determining interest change using the first ontology term and the second ontology term, which will be described later, may be applied when the first ontology term and the second ontology term are generated. Therefore, when the first ontology term and the second ontology term are both generated, the present invention compares the first ontology term and the second ontology term, and the comparison between the first similar ontology term and the second similar ontology term. Changes in interest can be determined. In addition, when the first ontology term and the second ontology term are not generated, the present invention may determine the change of interest of the user by comparing between the first similar ontology term and the second similar ontology term.

The keyword processing unit 220 may extract the levels corresponding to the first and second ontology terms by using the level of the product term set by the ontology constructing unit 210 (see Table 2).

In addition, the keyword processor 220 may calculate an average level and a maximum level of the first and second ontology terms. Here, the keyword processor 220 may calculate the average level by dividing the sum of the levels for the ontology terms by the number of ontology terms, and calculate the maximum level by extracting the largest level among the levels for the ontology terms.

For example, when the first average level of the first ontology term is OLA1, the first maximum level is OLH1, the second average level of the second ontology term is OLA2, and the second maximum level is OLH2, It can be expressed as

[Equation 2]

First ontology term = (a, b,…)

OLA1 = e

OLH1 = f (e <= f)

Second ontology term = (c, d,…)

OLA2 = g

OLH2 = h (g <= h)

In Equation 2, a, b, c, and d represent product terms, and e, f, g, and h represent levels of an ontology hierarchy. In the above equation, the average level may be a real number rather than an integer, and the average level may be a real number. However, in some cases, the average level may be converted to an integer by a rounding method. In addition, the number of the first ontology term and the second ontology term may be one or more, respectively.

The keyword processor 220 may generate a user vector using the first and second keyword sets and calculate first and second similar ontology terms having high similarity using the user vector. In addition, the keyword processor 220 may calculate an average level and a maximum level of the first and second similar ontology terms.

The keyword processor 220 may generate the user vector in the same way as the ontology builder 210 generates the term vector. That is, the keyword processing unit 220 may generate a user vector by setting the value of an element of the term vector corresponding to the keyword shown in the keyword set to 1 (the value of an element for which the keyword does not appear is set to 0). In this manner, the keyword processor 220 may generate first and second user vectors for the first and second keyword sets, respectively.

The keyword processor 220 may generate the user vector in the same manner as the term vector by using one of keyword presence, frequency, or weighting method as in the method of generating the term vector.

The keyword processing unit 220 may extract a similar ontology term having a high similarity by comparing the term vector and the user vector (the first user vector and the second user vector). Here, the keyword processor 220 may extract the first and second similar ontology terms for the first and second user vectors, respectively.

For example, the keyword processor 220 calculates the similarity using a Boolean search method, an extended Boolean search method, a Euclidian distance formula or a Coscine formula, and the like ontology term. Can be extracted.

For example, a method of applying a Boolean search method is a method of identifying similar numbers of keywords and comparing similarities. The keyword processing unit 220 may calculate the ranking of keywords in a similar order using a method using a Boolean search method. That is, the keyword processor 220 may calculate a higher rank as more keywords match.

In addition, the method using the Euclidean distance formula or the cosine formula is a method of comparing the similarity between the vectors. The keyword processor 220 may calculate the ranking in a similar order by using a method using a Euclidean distance formula or a cosine formula. The Euclidean distance formula or cosine formula can be a more effective method of comparing similarities than the Boolean search method.

For example, assuming that the term vector is P (p1, p2,…, pn), the user vector is Q (q1, q2,…, qn), and n is the total number of keywords (number of vector elements), the Euclidean formula May be represented by Equation 3 below.

&Quot; (3) &quot;

The keyword processing unit 220 may determine the similarity by the distance between two points in the space using the Euclidean distance formula, and may determine that the similarity is higher as the distance is closer.

And, the cosine formula can be represented by the following equation (4).

&Quot; (4) &quot;

The keyword processor 220 may determine the similarity based on the cosine of the angle formed by the two vectors in the space using a cosine formula, and may determine that the similarity is high as the cosine is closer to 1 (the smaller the angle).

Meanwhile, the present invention is not limited to the similarity comparison method using the similarity formula, and in addition to these, the similarity comparison method, the extended Boolean search method modified from the similarity formula, or various similarity comparison methods for determining similarity between vectors may be used. Note that it can be used.

 The keyword processor 220 may extract similar ontology terms according to the similarity calculated by comparing the term vector and the user vector. That is, the keyword processing unit 220 may extract a similar ontology term by extracting a predetermined number of merchandise terms from the upper rank among the merchandise terms set according to the similarity. Thus, the keyword processor 220 may generate first and second similar ontology terms for the first and second user vectors.

In addition, the keyword processing unit 220 may extract a level corresponding to each of the first and second similar ontology terms using the level of the product term set by the ontology constructing unit 210 (see Table 2).

That is, the keyword processor 220 may calculate an average level and a maximum level of the first and second pseudo ontology terms. Here, the keyword processing unit 220 may calculate the average level by dividing the sum of the levels for the similar ontology terms by the number of the similar ontology terms, and calculate the maximum level by extracting the largest level among the similar ontology terms. .

For example, when the first average level of the first pseudo- ontology term SLA1, the first maximum level of SLH1, the second average level of the second pseudo- ontology term SLA2, and the second maximum level of SLH2, the following mathematics It can be expressed as Equation 5.

[Equation 5]

First pseudo-ontology term = (v, w,…)

SLA1 = p

SLH1 = q (p <= q)

Second pseudo-ontology term = (x, y,…)

SLA2 = r

SLH2 = s (r <= s)

In Equation 5, v, w, x, and y represent a product term, and p, q, r, and s represent levels of an ontology hierarchy. In the above equation, the average level may be a real number rather than an integer, and the average level may be a real number. However, in some cases, the average level may be converted to an integer by a rounding method. In addition, the number of the first ontology term and the second ontology term may be one or more, respectively.

The interest change determiner 230 may determine a change in interest of the user by comparing an average level of any one or more of the ontology term and the similar ontology term.

That is, the interest change determiner 230 may determine the change of interest by comparing the magnitudes of the first average level and the second average level. In this case, the interest change determiner 230 may compare the first and second average levels of the first and second ontology terms, or compare the first and second average levels of the first and second similar ontology terms. have.

For example, the interest change determiner 230 may compare OLA1 and OLA2. Thus, the interest level change determiner 230 may determine that the interest level decreases when OLA1> OLA2, and may determine that the interest level is maintained or increased when OLA1 <= OLA2.

As another example, the interest change determiner 230 may compare SLA1 and SLA2. Thus, the interest level change determiner 230 may determine that interest level decreases when SLA1> SLA2, and may determine that interest level is maintained or increased when SLA1 <= SLA2. As described above, if the first and second ontology terms are not generated, the change in the degree of interest of the user may be determined by comparing the average levels SLA1 and SLA2 between the first and second similar ontology terms.

In addition, the interest change determiner 230 may compare OLA1 and OLA2 and compare SLA1 and SLA2. Thus, the interest change determination unit 230 may determine that the interest level decreases when OLA1> OLA2 AND SLA1> SLA2, and may determine that the interest level is maintained or increased when OLA1 <= OLA2 AND SLA1 <= SLA2.

Of course, in addition to the situation it may be determined that the interest is reduced or maintained in various ways. For example, if an ontology term is emphasized as compared to a similar ontology term, it may be determined that the interest level is maintained or increased if OLA1 <= OLA2, and the interest level is reduced if OLA1> OLA2. Alternatively, the degree of interest before and after the comparison may be determined by comparing average levels SLA1 and SLA2 of similar ontology terms. It may also be determined in consideration of an appropriate error range. That is, depending on the situation, an increase in the interest level may be determined to be increased by a certain level or more, and a decrease or decrease in the interest level may be determined to be a decrease in the interest level.

The advertisement content detector 240 may select a product term according to a change in interest degree, detect an advertisement content connected to the selected product term, and perform an advertisement.

The advertisement content detection unit 240 uses the second average level and the second maximum level for the second ontology term, the second average level and the second maximum level for the second similar ontology term, and selects the product term according to the change of interest. You can choose.

When the interest content is maintained or increased, the advertisement content detection unit 240 may select a product term having the highest level from the ontology term and the similar ontology term in order to match the change in the interest of the user. This is for selecting a specific advertisement because the larger level is a more specific product term than the smaller level.

For example, the advertisement content detector 240 may select OLH2 and SLH2 to select the largest level and a product term of the corresponding level. That is, the advertisement content detection unit 240 may select a product term of the largest level and the largest level using Equation 6 below.

&Quot; (6) &quot;

Largest level L = max (OLH2, SLH2)

Product term at the highest level T = {product term at the highest level in the second ontology term and the second similar ontology term}

In Equation 6, max (OLH2, SLH2) yields the larger of two values. The product term T of the largest level is selected as the product term of level L selected as the largest level. In addition, as described above, when the first and second ontology terms are not extracted, SLH2 may be selected as the largest level and a product term of the corresponding level may be selected.

In addition, when the interest level decreases, the advertisement content detection unit 240 may select a product term having an average level from the ontology term and the similar ontology term in order to match the change in the degree of interest of the user.

For example, the advertisement content detector 240 may calculate average levels of the OLA2 and the SLA2 and select a product term. That is, the advertisement content detection unit 240 may select the average level and the product term of the average level using Equation 7 below.

[Equation 7]

Average level L = round [average (OLA2, SLA2)]

Average term product term T = {average level product term in second ontology term and second similar ontology term}

In Equation 7, average (OLA2, SLA2) calculates an average of two values, and since the calculated average value may be a real number, the value calculated by round (rounding) is converted into an integer. Instead of rounding, it can be converted to an integer by a method such as trimming and rounding. And commodity term of average level L is selected. In addition, as described above, when the first and second ontology terms are not extracted, SLA2 may be selected as an average level and a product term of the corresponding level may be selected. If the product term at the average level L does not exist in the second ontology term and the second similar ontology term (the product term may not exist at the average level), the product term at the level closest to the average level L may be selected. have. That is, if the product term of the average level L does not exist, the advertisement content detector 240 may select a product term of a level within a preset range. Alternatively, a product term of average level L may be selected in the ontology hierarchy.

The advertisement content detector 240 detects the advertisement content connected to the selected product term T and performs an advertisement. In this case, the advertisement content detection unit 240 may select and advertise a plurality of advertisement contents sequentially or randomly when there are a plurality of selected product terms and a plurality of advertisement contents, or when there are a plurality of advertisement contents for a single product term. Alternatively, the advertisement content detection unit 240 may preferentially select the latest advertisement or the advertisement having a high advertisement fee.

If there is no advertisement content connected to the selected product term, the advertisement content detector 240 may search for the product term located at the closest distance to the selected product term T from the table representing the ontology hierarchy (see Table 2). Can be.

For example, the advertisement content detection unit 240 assumes the distance between the parent and the child in the ontology hierarchy as 1, and the product term of the closest distance where the advertisement content exists among the product terms located above or below the product terms selected based on the product terms selected. Can be extracted. In this case, when the interest level is maintained or increased, the advertisement content detector 240 may select a product term of a large level among the product terms of the closest distance. This is to advertise specific products. On the contrary, the advertisement content detector 240 may select product terms having the same or lower level when the interest decreases. This is for general advertising.

3 is a flowchart illustrating an advertisement providing method according to an exemplary embodiment of the present invention.

Hereinafter, the functions of each component described with reference to FIG. 2 are processed in the advertisement providing apparatus 30 according to the embodiment of the present invention. In order to understand the present invention, the advertisement providing apparatus 30 is mainly used. Explain.

Referring to FIG. 3, the advertisement providing apparatus 30 constructs an ontology of product terms (S310). The advertisement providing apparatus 30 generates a hierarchical structure in a tree structure by using an ontology hierarchy structure of an isA relationship between a plurality of product terms according to an up-and-down concept, and sets levels and node numbers from top to bottom for each hierarchy. Can be set.

Next, the advertisement providing device 30 generates a term vector (S315). That is, the advertisement providing apparatus 30 defines a description of each product term, and generates a term vector by extracting a keyword therefrom. Here, the advertisement providing apparatus 30 may set the position and the number of elements of the term vector for the extracted whole keywords, and generate the term vectors for the corresponding product terms, respectively. Here, the advertisement providing device 30 may generate the term vector for each product term by displaying the existence of the keyword at the position of the element of the term vector where the keyword for the product term exists. The term vector may be generated by setting the value of an element of the vector to the frequency of the keyword or the weight of the keyword.

Next, the advertisement providing apparatus 30 connects the product term and the advertisement content (S320). That is, the advertisement providing apparatus 30 may collect advertisement content for each product term and connect the product term and the advertisement content. Here, the collected advertising content may be provided from an advertisement provider, or may include any one or more of a text advertisement and a multimedia advertisement collected by a user searching directly.

The advertisement providing apparatus 30 may extract keywords from texts such as titles (titles), advertisement phrases, advertisement lines, descriptions, and annotations of the collected advertisement contents, and may also utilize the techniques such as image processing or image processing. Keywords can be extracted from images or videos of advertisement content. In addition, the advertisement providing apparatus 30 generates a content vector using any one of presence, frequency, and weight of a keyword in the same manner as the method of generating the term vector, and compares the similarity between the content vector and each term vector. In comparison, a product term corresponding to the term vector most similar to the content vector may be selected. That is, the advertisement providing apparatus 30 may select a product term having the highest similarity as the product term of the corresponding advertisement content, and connect the product term to the corresponding advertisement content.

The advertisement providing apparatus 30 may collect and link the abstract image advertisement content to the upper level product term according to the upper and lower relations of the merchandise term on the ontology hierarchy, and collect and link the specific product advertisement content to the lower level product term. Can be.

Next, the advertisement providing apparatus 30 generates a first keyword set and a second keyword set (S325). The advertisement providing apparatus 30 may generate a first keyword set and a second keyword set to identify a change in interest of the user. That is, the advertisement providing apparatus 30 may generate the first keyword set from the search word input by the user, and then generate the second keyword set from the search word input by the user.

In addition, the advertisement providing device 30 may generate the first and second keyword sets from the documents provided as a search result as well as the search word input by the user. Alternatively, the advertisement providing apparatus 30 may generate the first and second keyword sets from the documents read and clicked by the user from among the search terms input by the user and documents provided as a search result.

Next, the advertisement providing apparatus 30 extracts the first and second ontology terms from the first and second keyword sets, respectively (S330). The advertisement providing apparatus 30 may generate the first and second ontology terms by extracting a keyword corresponding to a product term from the first and second keyword sets. In this case, the first and second ontology terms may not be generated.

Next, the advertisement providing apparatus 30 calculates first and second average levels and first and second maximum levels, respectively, for the first and second ontology terms (S335). The advertisement providing apparatus 30 may extract a level corresponding to the first and second ontology terms by using the level of the product term (see Table 2). In addition, the advertisement providing apparatus 30 may calculate an average level and a maximum level for the first and second ontology terms. That is, the advertisement providing apparatus 30 may calculate the average level by dividing the sum of the levels for the ontology terms by the number of ontology terms, and calculate the maximum level by extracting the largest level among the levels for the ontology terms.

Next, the advertisement providing apparatus 30 generates first and second user vectors, respectively, using the first and second keyword sets (S340). The advertisement providing device 30 sets the value of the element of the term vector corresponding to the keyword shown in the keyword set to 1 (the value of the element for which the keyword does not appear is set to 0), or sets the value of the element of the vector to the keyword. The user vector may be generated in the same manner as the term vector by generating the term vector by setting the frequency or the weight of the keyword. In this way, the advertisement providing device 30 may generate the first and second user vectors for the first and second keyword sets, respectively.

Next, the advertisement providing apparatus 30 extracts the first and second similar ontology terms having high similarity, respectively, using the first and second user vectors (S345). The advertisement providing apparatus 30 may extract a similar ontology term having high similarity by comparing the term vector and the user vector (the first user vector and the second user vector). That is, the advertisement providing apparatus 30 may extract a similar ontology term by selecting a predetermined number of merchandise terms from the upper rank among the merchandise terms set according to the similarity. As such, the advertisement providing apparatus 30 may extract the first and second similar ontology terms for the first and second user vectors, respectively.

Next, the advertisement providing apparatus 30 calculates first and second average levels and first and second maximum levels with respect to the first and second similar ontology terms, respectively (S350). The advertisement providing apparatus 30 may extract a level corresponding to the first and second similar ontology terms using the level of the product term (see Table 2). In addition, the advertisement providing apparatus 30 may calculate an average level and a maximum level of the first and second similar ontology terms. That is, the advertisement providing apparatus 30 may calculate the average level by dividing the sum of the levels for the similar ontology terms by the number of ontology terms, and calculate the maximum level by extracting the largest level among the similar ontology terms. .

Next, the advertisement providing apparatus 30 determines a change in interest level by comparing the first average level and the second average level of the ontology term or the similar ontology term (S355). That is, the advertisement providing apparatus 30 may determine a change in interest level by comparing the sizes of the first average level and the second average level. In this case, the interest change determiner 230 may compare the first and second average levels of the first and second ontology terms and compare the first and second average levels of the first and second similar ontology terms. have. If the first and second ontology terms are not selected, the first and second average levels of the first and second similar ontology terms may be compared.

Next, the advertisement providing apparatus 30 selects a product term according to the change in interest level (S360). That is, the advertisement providing apparatus 30 may use the second average level and the second maximum level for the second ontology term, the second average level and the second maximum level for the second similar ontology term, and the product according to the change of interest. You can select a term.

When the interest providing apparatus 30 maintains or increases the interest level, the advertisement providing apparatus 30 may select a product term of the highest level from the ontology term and / or the similar ontology term in order to match the change in the interest degree of the user. Alternatively, when the degree of interest decreases, the advertisement providing apparatus 30 may select a product term of an average level or a level close to the average level in an ontology term and / or a similar ontology term in order to match the change in the interest of the user. That is, when the product term of the average level does not exist, the advertisement providing apparatus 30 may select the product term of a level within a preset range.

Next, the advertisement providing apparatus 30 detects advertisement content connected to the selected product term (S365), and performs advertisement using the advertisement content (S370). The advertisement providing device 30 detects advertisement content connected to the selected product term and performs an advertisement. In this case, the advertisement providing apparatus 30 may select and advertise a plurality of advertisement contents sequentially or randomly when there are a plurality of selected product terms and a plurality of advertisement contents, or when there are a plurality of advertisement contents for a single product. . Alternatively, the advertisement providing device 30 may preferentially select the latest advertisement or the advertisement having a high advertisement fee.

Advertisement providing method according to an embodiment of the present invention is implemented in the form of program instructions that can be executed by various computer means may be recorded on a computer readable medium. Computer-readable media may include, alone or in combination with the program instructions, data files, data structures, and the like.

The program instructions recorded on the computer readable medium may be those specially designed and constructed for the present invention, or may be known and available to those skilled in the computer software arts. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tape, optical media such as CD-ROMs, DVDs, and magnetic disks, such as floppy disks. Hardware devices specially configured to store and execute program instructions such as magneto-optical media and ROM, RAM, flash memory and the like. In addition, the above-described medium may be a transmission medium such as an optical or metal wire, a waveguide, or the like including a carrier wave for transmitting a signal specifying a program command, a data structure, and the like. Examples of program instructions include not only machine code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like.

The hardware device described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

Although the above has been described with reference to a preferred embodiment of the present invention, those skilled in the art to which the present invention pertains without departing from the spirit and scope of the present invention as set forth in the claims below It will be appreciated that modifications and variations can be made.

1 is a configuration diagram schematically illustrating an advertisement providing system according to an embodiment of the present invention.

2 is a block diagram showing an advertisement providing apparatus according to an embodiment of the present invention.

3 is a flowchart illustrating a method for providing advertisement according to an embodiment of the present invention.

4 is an exemplary diagram illustrating an ontology hierarchy according to an embodiment of the present invention.

Claims (20)

An apparatus for providing an advertisement according to a search term entered by a user, Construct the upper and lower relations of a plurality of product terms according to the concept of upper and lower using an ontology hierarchy structure, set the level of the product terms according to the ontology hierarchy, and collect the advertising content for the product terms, the product terms and the advertisement Ontology construction unit for connecting the content; A keyword processing unit which generates a keyword set by extracting a keyword from the search word, and calculates an average level and a maximum level using the level of the product term corresponding to the keyword of the keyword set; An interest change determination unit to determine a change in interest using the average level; And And an advertisement content detection unit for selecting a product term according to the change of interest and detecting an advertisement content connected to the selected product term to perform an advertisement. The method of claim 1, The ontology construction unit extracts a keyword from the description defined for each of the product terms, and generates a term vector for each product term using the extracted keywords. The term vector is an advertisement providing device, characterized in that any one of the presence or absence, frequency, and weight of the extracted keyword is generated for each product term. The method of claim 2, The advertising content may be any one or more of a text advertisement and a multimedia advertisement provided by an advertisement provider or collected by a user searching directly. The ontology construction unit extracts a keyword from the advertisement content, generates a content vector for each advertisement content using the extracted keyword, and compares the similarity between the content vector and the term vector to obtain a product of the term vector having the maximum similarity. And selecting the term as a product term of the corresponding advertisement content, and connecting the product term to the corresponding advertisement content. The method of claim 1, The keyword processing unit generates a first set of keywords from the search terms entered by the user, generates a second set of keywords from the search terms entered again, and extracts keywords corresponding to the product terms from the first and second sets of keywords. Generating first and second ontology terms, and using the level of the product term corresponding to each keyword of the first and second ontology terms, the first average level and the first maximum level of the first ontology term are determined. And calculating a second average level and a second maximum level of the second ontology terminology. 5. The method of claim 4, The keyword processing unit generates first and second user vectors for the first and second keyword sets, And the user vector is generated with respect to the first and second keyword sets by using any one of the extracted keywords, the presence, frequency, and weight of the extracted keyword. The method according to claim 3 or 5, The value of each element of the term vector, the content vector, or the user vector is calculated by multiplying the frequency of the corresponding keyword by the adjustment coefficient, and the adjustment coefficient is an Inverse Document Frequency (IDF) of the keyword. The literature frequency is a log value of the total number of documents (wherein, the total number of documents is the total number of documents in which each keyword appears) divided by the number of documents in which the corresponding keywords appear. The method of claim 5, The keyword processing unit compares the first and second user vectors with the term vector, calculates a similarity, and extracts a product term of a predetermined number of term vectors from a higher rank among term vectors set according to the similarity. Generating first and second similar ontology terms for the first and second user vectors, and using the level of each product term in the first and second similar ontology terms, the first average level of the first similar ontology term and And calculating a first maximum level and calculating a second average level and a second maximum level of the second pseudo-ontology term. The method of claim 7, wherein The interest change determination unit determines the change of interest by comparing the magnitude of the first average level and the second average level of any one or more of the ontology term or the similar ontology term, If the second average level is less than the first average level, it is determined that the degree of interest decreases. And when the second average level is greater than or equal to the first average level, determining that interest is maintained or increased. The method of claim 8, When the interest level is maintained or increased, the advertisement content detection unit selects a large level by comparing a second maximum level of the second ontology term and a second maximum level of the second similar ontology term, and selects a higher level from the second ontology term. In the second similar ontology terminology, a large level of merchandise term is selected, And when the ontology term is not generated, selecting a product term of a second maximum level of the second similar ontology term. The method of claim 8, When the interest level decreases, the advertisement content detector calculates a third average level of the second average level of the second ontology term and a second average level of the second similar ontology term, and calculates the second ontology term and the second mean. 2 In the similar ontology terminology, a product term having a level within a preset range is selected based on the third average level or the third average level, When the ontology term is not generated, a product term having a level within a predetermined range is selected based on a second average level of the second pseudo-tension ontology term or a second average level of the second pseudo-ontology term. Ad serving device. An advertisement providing method performed by an advertisement providing device that provides an advertisement according to a search term input by a user, a) establishing a vertical relationship of a plurality of product terms according to an upper and lower concept using an ontology hierarchy; b) setting the level of the merchandise term according to an ontology hierarchy; c) collecting advertisement content for the product term and linking the product term and the advertisement content; d) generating a first keyword set from the search word input by the user and generating a second keyword set from the search word input again; e) calculating first and second average levels and first and second maximum levels for the first and second keyword sets using the levels of the product terms corresponding to the keywords of the first and second keyword sets. Doing; f) determining a degree of interest change using the first and second average levels; And g) selecting a product term according to the change in interest level, and detecting an advertisement content connected to the selected product term to perform an advertisement. The method of claim 11, Before step c), Extracting a keyword from the description defined for each of the product terms, and using the extracted keywords to generate a term vector for each product term, The term vector is one of the presence, frequency, and weight of the extracted keyword as an element value, the advertisement providing method, characterized in that generated for each product term. The method of claim 12, Step c) Extracts a keyword from the advertisement content and generates a content vector for each advertisement content by using the extracted keyword, and compares the similarity between the content vector and the term vector to select a product term of the term vector having the maximum similarity. Select the product term of the content, link the product term and the corresponding advertisement content, The advertisement content may be any one or more of a text advertisement or a multimedia advertisement provided from an advertisement provider or collected by a user searching directly. The method of claim 11, Step e) Generating first and second ontology terms by extracting keywords corresponding to the product terms from the first and second keyword sets; A first average level and a first maximum level of the first ontology term are calculated using the levels of the respective product terms of the first and second ontology terms, and the second average level and the second maximum of the second ontology term are calculated. Calculating an advertisement level. The method of claim 11, Step e) Generating first and second user vectors for the first and second keyword sets, And the user vector is generated for the first and second keyword sets using any one of the extracted keyword, whether there is a presence, frequency, or weight. The method according to claim 11 or 13, The value of each element of the term vector or the user vector is calculated by multiplying the frequency of the corresponding keyword by the adjustment coefficient, the adjustment coefficient being the Inverse Document Frequency (IDF) of the corresponding keyword, and the inverse document frequency The number of documents (wherein, the total number of documents is the total number of documents in which each keyword appears) is a log value of the log value of the value divided by the number of documents in which the corresponding keyword appears. The method according to claim 13 or 15, Step e) Calculating similarity by comparing the first and second user vectors with the term vector; Generating first and second similar ontology terms for the first and second user vectors by extracting a predetermined number of keywords from an upper level among the keywords set according to the similarity; And Calculating a first average level and a first maximum level of the first pseudo-ontology term using the level of the product term corresponding to each keyword of the first and second pseudo-ontology terms, Calculating a second average level and a second maximum level. The method of claim 11, Step f) The change in the degree of interest may be determined by comparing the magnitudes of the first average level and the second average level of any one or more of the ontology terminology or the similar ontology terminology. If the second average level is less than the first average level, it is determined that the degree of interest decreases. And when the second average level is greater than or equal to the first average level, determining that interest is maintained or increased. The method of claim 11, Step g) When the interest is maintained or increased, a second level is compared by comparing the second maximum level of the second ontology term and the second maximum level of the second similar ontology term, and the second ontology term and the second similar ontology are selected. Choose a higher level product term from the term, And when the ontology term is not generated, selecting a product term of a second maximum level of the second similar ontology term. The method of claim 11, Step g) When the degree of interest decreases, a third average level of the second average level of the second ontology term and a second average level of the second similar ontology term are calculated, and the second ontology term and the second similar ontology term are used. Select a product term of a level within a preset range based on the third average level or the third average level, When the ontology term is not generated, a product term having a level within a predetermined range is selected based on a second average level of the second pseudo-tension ontology term or a second average level of the second pseudo-ontology term. How to serve ads.

Images