KR20110089473A - System and method for matching between contents and advertizing medium - Google Patents

Abstract

PURPOSE: A matching system between content and advertisement and method thereof are provided to supply advertisement to a user of content by analyzing content. CONSTITUTION: A keyword database(102) stores interest information of a keyword. An advertisement database(104) stores targeted advertisement layer information of the advertisement. A keyword extractor(106) extracts the frequency of each keyword from a web page. A keyword processing unit(108) uses interest information in keyword database.

Description

System and method for matching between contents and advertizing medium}

The present invention is directed to a system and method for matching advertisements most relevant to content.

As the Internet is popularized, more information is searched and acquired through the Internet than other media. In addition, in line with this trend, the advertising market has also increased exponentially in the case of using the Internet away from the conventional TV, newspapers, and the like.

In general, advertisements on the Internet are implemented in the form of short texts, banners, or widgets, and are provided in a form of being displayed on a web browser together with newspaper articles or blog posts. However, since these advertisements are provided separately from newspaper articles or blog posts that are published together, advertisements irrelevant to the newspaper articles are displayed on one page. The effect is often halved.

An object of the present invention is to provide an appropriate advertisement to a user of the content by analyzing the content such as the Internet and selecting the advertisement having the highest similarity with the content.

According to an aspect of the present invention, there is provided a matching system between a content and an advertisement, the keyword database storing a plurality of keywords and interest level information for each hierarchical section of the keyword; An advertisement database storing a plurality of advertisements and target advertisement layer information of the advertisements; A keyword extraction unit for extracting keywords and the frequency of each keyword from the web page; A keyword processing unit that calculates an interest propensity of the web page by using the extracted keyword and frequency and interest level information for each hierarchical section of each keyword stored in the keyword database; And an advertisement selector configured to select advertisements corresponding to the web page in consideration of the interest propensity of the web page calculated by the keyword processor and target advertisement layer information for each advertisement stored in the advertisement database.

On the other hand, the matching method between the content and the advertisement according to an embodiment of the present invention for solving the above problem, in the matching system between the content and the advertisement, extracting a keyword and the frequency of each keyword from the web page; Calculating an interest propensity of the web page by using the extracted keyword and frequency and interest information for each hierarchical section of the keyword in the matching system between the content and the advertisement; And selecting, in the matching system between the content and the advertisement, the advertisement corresponding to the web page by comparing the calculated propensity to interest of the web page and target advertisement layer information for each advertisement.

The present invention analyzes content such as the Internet and selects an advertisement having the highest similarity with the content so that the advertisement can be displayed together with the content, thereby increasing the attention and click rate of the advertisement and increasing the purchase rate of a product to be advertised. There are advantages to it.

1 is a view showing the configuration of a matching system 100 between content and advertisements according to an embodiment of the present invention.
2 is a flowchart 200 illustrating a process of calculating a propensity of interest of the web page from keywords and frequencies extracted from the web page according to an embodiment of the present invention.
3 is a flowchart illustrating a matching method 300 between content and advertisements according to an embodiment of the present invention.

Hereinafter, specific embodiments of the present invention will be described with reference to the drawings. However, this is only an example and the present invention is not limited thereto.

In describing the present invention, when it is determined that the detailed description of the known technology related to the present invention may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted. The following terms are defined in consideration of the functions of the present invention, and may be changed according to the intention or custom of the user, the operator, and the like. Therefore, the definition should be based on the contents throughout this specification.

The technical spirit of the present invention is determined by the claims, and the following embodiments are merely means for efficiently explaining the technical spirit of the present invention to those skilled in the art.

1 is a view showing the configuration of a matching system 100 between content and advertisements according to an embodiment of the present invention.

As shown, the matching system 100 between the content and the advertisement according to an embodiment of the present invention is a keyword database 102, the advertisement database 104, keyword extraction unit 106, keyword processing unit 108 and The advertisement selection unit 110 is included.

The keyword database 102 is a database in which a plurality of keywords and interest information for each hierarchical section of the keyword are stored. The keyword is mainly composed of Korean nouns, it may be composed of words frequently used in everyday life. For example, words such as "bag", "furniture", "car", etc. may be selected as keywords stored in the keyword database 102.

In addition, the interest level information for each hierarchical section indicates information on whether there is a difference in interest for each preset hierarchical section for each keyword stored in the keyword database 102. For example, the interest level information for each hierarchical section may be information on a difference in interest of “by gender and age hierarchy” for the keyword. For this, the gender and age sections may be referred to as “South teens, South 20s, South 30s”. Divide into 10 sections of "large, 40s, 50s, 10s, 20s, 30s, 40s, 50s" and divide the percentage of interest in the corresponding keywords for each section. It can be measured in (%). Table 1 below illustrates the interest information for each gender and age section for the keywords “furniture” and “bag”.

keyword south female 10-19 20-29 30-39 40-49 50-59 10-19 20-29 30-39 40-49 50-59 furniture 7.2 9.6 17.2 8.2 4.4 8.2 11.0 19.7 9.3 5.2 bag 9.6 11.4 7.8 4.3 2.0 17.6 21.1 14.3 8.2 3.7

The keyword and interest information for each hierarchical section may be generated using, for example, a search word input from a search engine such as a portal and hierarchical information of a user who inputs the search word. That is, if a user inputs a keyword corresponding to a specific section (for example, South 20s) with respect to each keyword, the interest percentage of the corresponding section may increase. Therefore, by using the keyword and the interest level for each hierarchical section, it is possible to determine the interest propensity for each hierarchical section of a specific keyword. The hierarchical information of the user may be input directly from the user, or may use information input when the user subscribes.

Meanwhile, in the present embodiment, the example of the hierarchical section has been mainly described as a gender and an age section, but various criteria may be applied to the hierarchical section in the present invention. For example, a user's salary level, occupation, region, race, single / married status, etc. may be defined as hierarchical sections in the present invention, and the interest level of each keyword for each defined hierarchical section may be stored.

The advertisement database 104 is a database in which a plurality of advertisements and target advertisement layer information of the advertisements are stored. The advertisement may be, for example, information about a website to be advertised by an advertiser. In this case, the advertisement may include URL information of a website to be advertised. In addition, the target advertisement layer information of the advertisement is preference information according to the hierarchical section of the prospective visitor for the website. For example, if the website you want to advertise is a women's clothing sale site in your twenties and thirties, your target audience will be primarily in your twenties and female thirties. It will be mainly in their 30s and 50s. Advertisement database 104 may also store information about keywords associated with the website to be advertised. For example, a keyword associated with a women's clothing sales site may be a skirt, a skirt, or the like, and in the case of an external automobile information site, the keyword may be Mercedes, BMW, or the like.

Table 2 illustrates the target advertisement layer stored in the advertisement database 104 as described above. The expected target advertisement layer of each advertisement for the gender and age intervals identical to the keyword database 102 is represented as a percentage (%). It is represented as.

Advertisement south female 10-19 20-29 30-39 40-49 50-59 10-19 20-29 30-39 40-49 50-59 Women's Clothing Store 0 0 0 0 0 10 40 40 10 0 Foreign car information site 0 10 25 25 15 0 0 10 10 5

The keyword extracting unit 106 extracts the keyword and the frequency of each keyword from the web page. According to an embodiment of the present invention, the web page is a page storing content for combining advertisements. For example, the web page may be a newspaper article, a blog post, or another website post.

In order to extract the keyword and the frequency of each keyword from the web page, the keyword extractor 106 needs to extract only the meaningful text from the web page without the HTML tag. Removal of HTML tags from such web pages is accomplished using an HTML parser. The technical configuration of the HTML parser, etc. correspond to those known in the art, and a detailed description thereof will be omitted herein. In addition, in the text extraction process, unnecessary parts such as advertisements included in the web page may also be removed.

As described above, when the HTML tag is removed from the web page and the text is extracted, the text is then morphologically analyzed to calculate keywords and frequencies in the text of each keyword. In the process of morphological analysis of the text, unnecessary symbols, English letters or stopwords are removed, and grammatical morphemes are removed using syllable information dictionaries, search dictionaries, and ending dictionaries. Will be extracted.

The keyword processing unit 108 calculates an interest propensity of the web page using the keyword and frequency extracted from the keyword extraction unit 106 and the interest level information for each hierarchical section of each keyword stored in the keyword database 102. The interest tendency of the web page is an index indicating how the hierarchy section interested in the web page is concerned. For example, the difference in the relative interest of the web page in the same hierarchy section as distinguished from the keyword database 102 is determined. It can be expressed as a percentage.

2 is a flowchart 200 illustrating a process of calculating a propensity of interest of the web page from keywords and frequencies extracted from the web page according to an embodiment of the present invention.

First, hierarchical interest information for each keyword is extracted from the keyword database 102 for each extracted keyword (202). Next, the frequency value of the keyword is multiplied by the value of each section of the extracted interest information (204), and the interest information for each keyword multiplied by the frequency is summed for all keywords of the web page, and the sum of all sections is 100. By normalizing to 206 (206), the propensity of interest can be calculated.

The interest propensity information of the web page generated in this manner may be expressed as a vector having a value of each section as an element. For example, as described above, when the interest information for each gender and age section is divided into 10 sections from South teens to 50s, the interest propensity information of the web page is expressed as a 10-dimensional vector having 10 elements. Can be.

For example, the process of extracting a disposition of interest from such a web page is as follows. For example, suppose that the keywords extracted from a given web page are two of "furniture" and "bag", and the frequencies of each are 2 and 3. It is assumed that interest information for each gender and age section of the keywords is as described in Table 1, respectively.

First, multiplying the frequency of interest information by gender and age hierarchy of "furniture" shown in Table 1 by 3, which is the frequency of "furniture" in the web page, is as follows. (The order of each value is the same as in Table 1.)

(21.6, 28.8, 51.6, 24.6, 13.2, 24.6, 33, 59.1, 27.9, 15.6)

Next, multiplying the degree of interest information corresponding to "bag" by 2, which is the frequency of "bag," as follows.

(19.2, 22.8, 15.6, 8.6, 4, 35.2, 42.2, 28.6, 16.4, 7.4)

Next, adding interest multiplied by frequency for all the keywords (furnitures and bags) on a webpage:

(43.2, 53.4, 57.8, 29.3, 14.8, 69.2, 85.3, 82.3, 43.2, 21.5)

This is divided by the total frequency of 5 and normalized so that the sum is 100.

(8.64, 10.68, 11.56, 5.86, 2.96, 13.84, 17.06, 16.46, 8.64, 4.3)

As described above, it can be seen through the above example that the normalized interest information can be expressed as a 10-dimensional vector. The interest information calculated as described above represents interests of the corresponding web pages by gender and age group, and the higher the value, the higher the interest level of the corresponding web page of the corresponding gender or hierarchy.

Lastly, the advertisement selection unit 110 corresponds to the web page by comparing the interest propensity of the web page calculated by the keyword processing unit 108 and target advertisement layer information for each advertisement stored in the advertisement database 104. Select the advertisement. In this case, the advertisement selecting unit 110 generates an advertisement propensity vector having an expected preference for each hierarchical section corresponding to each advertisement stored in the advertisement database 104, and generates the advertisement propensity vector. And the advertisement corresponding to the web page is selected by comparing the magnitude of the value of the interest propensity vector of the web page.

For example, when the advertisement propensity vector is generated using the target advertisement floor information of the women's clothing shopping mall and the foreign vehicle information site shown in Table 2, it is as follows.

Women's Clothing Store: (0, 0, 0, 0, 0, 10, 40, 40, 10, 0)

Foreign car information site: (0, 10, 25, 25, 15, 0, 0, 10, 10, 5)

Next, the advertisement propensity vector is normalized such that the sum of the values is 1, and the internal product value of the propensity vector of interest generated by the keyword processor 108 is obtained as follows.

Women's Clothing Store: 78.28

Foreign car information site: 42.96

The inner value represents the similarity between the interest propensity layer of the web page and the target advertisement layer of the advertisement, and the larger the inner value, the more similar the interest propensity layer of the web page and the target advertisement layer of the advertisement. The advertisement selecting unit 110 may compare the inner product values and select one advertisement having the largest value or a predetermined number of advertisements in the order of the inner product values as the advertisements matching the web page. In the above example, since the women's clothing shopping mall has a higher internal value compared to the foreign car information site, the women's clothing shopping mall may be selected as an advertisement that matches the corresponding web page.

Meanwhile, the matching system 100 between the content and the advertisement according to an embodiment of the present invention may further include a keyword expansion unit (not shown) and an extended keyword database (not shown).

If there is a keyword that is not stored in the keyword database 102 among the keywords extracted by the keyword extraction unit 106, the keyword expansion unit selects the corresponding keyword as an expansion keyword, and interest level information for each of the hierarchical sections of the expansion keyword. And calculate the frequency in the web page. That is, in the embodiment of the present invention, the keyword database 102 mainly stores keywords that are frequently used by users or that can easily grasp the propensity of the users. When extracted as a keyword, the keyword is stored in the extended keyword database by the keyword expansion unit.

The keyword expansion unit may calculate the interest of the extended keyword in the same way as to calculate the interest of each of the hierarchical sections of the keywords stored in the keyword database 102. That is, the frequency of the hierarchical layer in which the extended keyword is input may be calculated using the keyword input from the user and the personal information of the user who inputs the keyword.

The extended keyword database stores interest information for each layer section of the extended keyword calculated by the extended keyword calculator.

When the extended keyword is stored as described above, the keyword processing unit 108, when there is a keyword that is not stored in the keyword database 102 among the keywords extracted by the keyword extraction unit 106, the keyword is stored in the extended keyword database. It is determined whether the keyword is a keyword, and if the keyword is stored in the extended keyword database, interest information for each hierarchical section of the corresponding keyword is extracted from the extended keyword database. However, since the keyword stored in the extended keyword database should be applied differently from the keyword stored in the keyword database 102, the extracted propensity information is multiplied by a separate weight. The weight is determined according to the frequency in the web page of the corresponding extended keyword. Specifically, the weight may be determined as follows.

First, when the frequency is a predetermined lower bound, the weight is set to zero. That is, since an extension keyword having a frequency less than or equal to a certain value is most likely not related to the main word of the web page, the interest of the extension keyword is excluded from the calculation.

Next, when the frequency is an upper bound, the maximum value of the preset weight is applied. For example, if the maximum value of the preset weight is A, in case of the extended keyword whose frequency exceeds the preset maximum value, A is multiplied by the value of each element.

Finally, when the frequency is between the minimum value and the maximum value, the weight may be applied as much as the maximum value of the weight is multiplied by the ratio of the frequency and the maximum value of the corresponding extended keyword. For example, when the maximum frequency is 10, the frequency of the specific extended keyword is 6 and the maximum weight is 0.8, the weight of the corresponding extended keyword may be set to 0.8 * 6/10 = 0.48.

3 is a flowchart illustrating a matching method 300 between content and advertisements according to an embodiment of the present invention.

First, keywords and frequency of each keyword are extracted from a web page (302). The keyword and the frequency extraction process of the keyword have been described above.

Next, the interest propensity of the web page is calculated using the extracted keyword and frequency and the interest level information for each hierarchical section of each keyword extracted from each keyword database 102 (304). As described above, the interest propensity of the web page may be represented as a multidimensional vector having an interest value of each layer section as an element.

Next, an advertisement corresponding to the web page is selected in consideration of the calculated interest propensity of the web page and target advertisement layer information for each advertisement stored in the advertisement database 104. As described above, the advertisement selection may be performed by inner product of the interest tendency expressed by the vector and target advertisement layer information for each layer, and comparing respective inner products.

Meanwhile, an embodiment of the present invention may include a computer readable recording medium including a program for performing the methods described herein on a computer. The computer-readable recording medium may include program instructions, local data files, local data structures, etc. alone or in combination. The media may be those specially designed and constructed for the purposes of the present invention, or they may be of the kind well-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 recording media such as CD-ROMs, DVDs, magnetic-optical media such as floppy disks, and ROM, RAM, flash memory, and the like. Hardware devices specifically configured to store and execute program instructions are included. Examples of program instructions may include high-level language code that can be executed by a computer using an interpreter as well as machine code such as produced by a compiler.

Although the present invention has been described in detail with reference to exemplary embodiments above, those skilled in the art to which the present invention pertains can make various modifications to the above-described embodiments without departing from the scope of the present invention. Will understand.

Therefore, the scope of the present invention should not be limited to the described embodiments, but should be defined by the claims below and equivalents thereof.

100: Matching system between content and advertisement
102: keyword database 104: advertisement database
106: keyword extraction unit 108: keyword processing unit
110: advertisement selection unit

Claims (14)

A keyword database for storing a plurality of keywords and interest information for each hierarchical section of the keyword;
An advertisement database storing a plurality of advertisements and target advertisement layer information of the advertisements;
A keyword extraction unit for extracting keywords and the frequency of each keyword from the web page;
A keyword processing unit that calculates an interest propensity of the web page by using the extracted keyword and frequency and interest level information for each hierarchical section of each keyword stored in the keyword database; And
An advertisement selection unit configured to select advertisements corresponding to the web page by comparing the interest propensity of the web page calculated by the keyword processor and target advertisement layer information for each advertisement stored in the advertisement database;
Matching system between the content and the advertisement comprising a. The method of claim 1,
The target advertisement layer information of the advertisement is a matching system between content and advertisement, which is an expected preference of the advertisement for each layer section. The method of claim 2,
The keyword extracting unit generates a text from which the HTML code value is removed from the web page, and extracts the keyword and the frequency by stemming the text. The method of claim 2,
The keyword processing unit,
The keyword extracting unit multiplies the interest level information of each layer section corresponding to each keyword by the frequency of each keyword, sums up the interest information of each keyword of the web page multiplied by the frequency, and adds the interest level information to each layer section. A matching system between the content and the advertisement, which generates a propensity vector of interest whose element is. The method of claim 4, wherein
The advertisement selection unit,
Generates an advertisement propensity vector having an element of the predicted preference for each layer section corresponding to each advertisement stored in the advertisement database, and internalizes the generated advertisement propensity vector and the interest propensity vector of the web page. Matching content between the content and the advertisement by comparing the sizes of the web pages. The method of claim 1,
If there is a keyword that is not stored in the keyword database among the keywords extracted by the keyword extracting unit, the keyword is selected as an extended keyword, and the interest information for each hierarchical section of the extended keyword and the frequency in the web page are calculated. A keyword expansion unit; And
An extended keyword database for storing the interest information for each of the extension keywords and the hierarchical sections of the extension keyword;
Further comprising, the matching system between the content and the advertisement. The method of claim 6,
The keyword processing unit,
If there is a keyword that is not stored in the keyword database among the keywords extracted by the keyword extracting unit, it is determined whether the keyword is a keyword stored in the extended keyword database. Matching system between the content and the advertisement, extracting the interest level information for each layer of the keyword from the keyword database. The method of claim 7, wherein
When the keyword processing unit extracts the interest level information for each of the hierarchical sections of the corresponding keyword from the extended keyword database, the keyword processor applies a weight to the interest information for each of the hierarchical sections according to the frequency of the web page of the extended keyword. And matching system between advertisements. Extracting keywords and a frequency of each keyword from a web page in a matching system between the content and the advertisement;
Calculating an interest propensity of the web page by using the extracted keyword and frequency and interest information for each hierarchical section of the keyword in the matching system between the content and the advertisement;
Selecting an advertisement corresponding to the web page in the matching system between the content and the advertisement, by comparing the calculated interest inclination of the web page and target advertisement layer information for each advertisement;
Matching method between the content and the advertisement comprising a. 10. The method of claim 9,
The target advertisement layer information of the advertisement is a predicted preference of the advertisement for each of the hierarchical sections. The method of claim 10,
The keyword and frequency extraction step,
Generating text in which the HTML code value has been removed from the web page; And
Morphologically analyzing the text to extract the keyword and the frequency;
Matching method between the content and the advertisement comprising a. The method of claim 10,
Calculating interest propensity of the web page,
Multiplying the frequency information of each keyword by the interest information for each hierarchical section corresponding to each keyword extracted in the keyword and frequency extraction step;
Generating an interest propensity vector having the interest of each section as the sum of all interest information for each keyword of the web page multiplied by the frequency;
Matching method between the content and the advertisement further comprising. The method of claim 12,
The advertisement selection step,
Generating an advertisement propensity vector having an element of the preferred preference for each of the hierarchical sections corresponding to each advertisement and comparing the magnitude of the generated advertisement propensity vector with the interest propensity vector of the web page. Matching method between the content and the advertisement to select the advertisement corresponding to the web page. A computer-readable recording medium storing a program for performing a method recorded in claim 9 on a computer.

Images