JP2016048526A - Extraction device, extraction method, and extraction program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve customer appeal of contents related to keywords in a technical field of distributing such contents.SOLUTION: An extraction device includes a calculation part and an extraction part. The calculation part calculates similarity between feature information on first content associated with a keyword and feature information on second content including an arbitrary word (S13), where feature information is, for example, a word vector corresponding to content. The extraction part extracts a keyword corresponding to feature information calculated by the calculation part and having a similarity equal to or more than a predetermined threshold (S14).SELECTED DRAWING: Figure 2

Description

  The present invention relates to an extraction apparatus, an extraction method, and an extraction program.

  Enormous amounts of information are disclosed on the Internet, and users often use search sites when searching for information they want to know. In such a search site, advertisement distribution called search matching advertisement or search-linked advertisement may be performed.

  In this type of advertisement distribution, a technique is known in which an advertiser designates one or more advertisement-related keywords when submitting an advertisement. In this technique, when an advertisement-related keyword is input as a search keyword from a user, an advertisement related to the advertisement-related keyword is displayed adjacent to a search result corresponding to the search keyword.

JP 2010-211777 A

  However, the above-described conventional technology cannot always enhance the appeal effect of the content. Specifically, the above-described conventional technology does not disclose a method in which an advertiser appropriately specifies an advertisement-related keyword for an advertisement that is a submitted content. In search-linked advertisement distribution, appropriate advertisements cannot be distributed unless advertisement-related keywords are appropriately specified. Thereby, there exists a possibility that the appeal effect of the advertisement distributed may not fully be exhibited.

  In the above, an advertisement is taken as an example of the content, and the point that the appeal effect cannot always be improved has been described. However, it is not always possible to increase the appeal effect for content other than advertising purposes. In other words, the present invention is not limited to the example of search-linked advertisement delivery as described above. In the technical field of delivering content related to a keyword, appropriate content cannot be delivered unless the keyword is appropriately specified. , It is not always possible to increase the appeal of the content.

  The extraction device according to the present application calculates a similarity between the feature information of the first content associated with the keyword and the feature information of the second content including an arbitrary word, and is calculated by the calculation unit An extraction unit that extracts a keyword corresponding to the feature information having a similarity equal to or greater than a predetermined threshold.

  According to one aspect of the embodiment, there is an effect that the appeal effect of the content can be enhanced.

FIG. 1 is a diagram illustrating an example of a web page on which advertisement content according to the embodiment is displayed. FIG. 2 is a diagram illustrating an example of the extraction process according to the embodiment. FIG. 3 is a diagram illustrating a configuration example of the advertising device according to the embodiment. FIG. 4 is a diagram illustrating an example of the LP information storage unit according to the embodiment. FIG. 5 is a diagram illustrating an example of the keyword information storage unit according to the embodiment. FIG. 6 is a conceptual diagram illustrating an example of a generation process according to the embodiment. FIG. 7 is a flowchart illustrating an extraction processing procedure by the advertising device according to the embodiment. FIG. 8 is a diagram illustrating an example of the LP information storage unit according to the modification. FIG. 9 is a hardware configuration diagram illustrating an example of a computer that realizes the function of the advertising device.

  Hereinafter, a mode for carrying out an extraction apparatus, an extraction method, and an extraction program according to the present application (hereinafter referred to as “embodiment”) will be described in detail with reference to the drawings. Note that the extraction apparatus, the extraction method, and the extraction program according to the present application are not limited to the embodiment. Moreover, in the following embodiment, the same code | symbol is attached | subjected to the same site | part and the overlapping description is abbreviate | omitted.

[1. (Extraction process)
Below, an example of the extraction process which concerns on embodiment is demonstrated using FIG.1 and FIG.2. 1 and 2 show an example in which distribution processing and extraction processing are performed by the advertising device 100 corresponding to the extraction device according to the present application. First, an example of advertisement distribution according to the embodiment will be described with reference to FIG. FIG. 1 is a diagram illustrating an example of a web page on which advertisement content according to the embodiment is displayed.

  FIG. 1 shows a search result page 50 displayed on the user terminal 10 as a web page on which advertisement content according to the embodiment is displayed. The search result page 50 is a web page on which information related to a web page searched based on a search query input from a user is displayed.

  As shown in FIG. 1, the search result page 50 includes a search query input unit 52. A user who uses the search result page 50 inputs a search query to the search query input unit 52 via the user terminal 10. FIG. 1 shows an example in which a search result is displayed on the search result page 50 based on a search query “English conversation classroom” input to the search query input unit 52 by the user. That is, the search result page 50 displays information such as a link of a web page related to “English conversation class” as a search result.

  Further, on the search result page 50, advertisement content is displayed separately from the information of the web page that is the search result. For example, the advertising content is displayed in the advertising space 54 or the advertising space 56 included in the search result page 50. The advertising content includes text that advertises the business of the advertiser. The advertisement content is linked to a web page provided by the advertiser. A web page that is a link destination of such advertising content is called a landing page (hereinafter sometimes referred to as “LP”). In other words, advertisers are encouraged to move to LP by displaying advertisement content, and are used for business such as browsing web pages provided by themselves or purchasing products on web pages. .

  In such an advertisement distribution service linked to a search service, an advertiser makes a bid for a predetermined keyword when submitting advertisement content. Then, when the user inputs a predetermined keyword as a search query, the advertiser acquires the right to distribute the advertising content linked to the LP to the user. That is, in the search service, when the user inputs, as a search query, a keyword that is the same as or similar to the keyword associated with the advertising content, the advertiser links the advertising content to the search result page 50 in conjunction with the search service. Can be displayed. Thereby, since the advertiser can distribute the advertising content at the timing when the desire for specific information is generated in the user, it is possible to distribute the advertising content that can be expected to have a higher advertising effect.

  Here, the flow of the advertisement content distribution process linked to the search service will be described. The distribution process of the advertising content linked to the search service is performed by the advertising device 100, the search server 30, and the user terminal 10 via a network N (not shown) (for example, the Internet).

  The user terminal 10 is an information processing device used by a user. The user terminal 10 is, for example, a mobile terminal such as a smartphone, a tablet terminal, or a PDA (Personal Digital Assistant), a desktop PC (Personal Computer), a notebook PC, or the like.

  The search server 30 is a server device that provides a search service. For example, the search server 30 receives an input of a search query from a user at a predetermined search site. Then, the search server 30 provides the user with a search result based on the search query.

  The advertising device 100 is a server device that distributes advertisements in conjunction with the search server 30. The advertising device 100 accepts bids for keywords as well as submission of advertising content from advertisers. Then, the advertising device 100 distributes the advertising content to the user who has input the bid keyword in the search service according to the procedure described below.

  First, the user accesses a search site provided by the search server 30 via the user terminal 10. Then, the user inputs a search query at the search site. Thereby, the user terminal 10 transmits a search query to the search server 30 (step S01). The search server 30 provides the user with a search result page 50, which is a web page that displays search results corresponding to the search query (step S02).

  The search server 30 that has received the search query transmits information related to the search query to the advertising device 100 (step S03). Thereby, advertising device 100 acquires information about a search query inputted by a user. Then, when the search query is a bid keyword, the advertising device 100 selects an advertisement content associated with the bid keyword as a distribution candidate (step S04).

  As described above, the search result page 50 includes the advertising space 54 and the advertising space 56. Therefore, when displaying the search result page 50, the user terminal 10 makes a request for advertisement distribution to the advertising device 100 (step S05). Upon receiving the advertisement distribution request, the advertising device 100 distributes the advertisement content associated with the bid keyword to the user terminal 10 (step S06). As a result, for example, the advertising content provided by the advertiser who has bid on the keyword “English conversation classroom” is displayed in the advertising space 54. In addition, the advertising content provided by other advertisers who have bid on the same keyword is displayed in the advertising space 56. As described above, the advertising device 100 can distribute the advertising content to the user in conjunction with the search service. In addition, although the search server 30 showed the example which transmits query information to the advertising device 100 in step S03, query information may be transmitted to the advertising device 100 from the user terminal 10. FIG.

  By the way, as shown in the example of FIG. 1, the advertiser bids on a predetermined keyword in order to display the advertising content on the search result page 50. In the case of the example in FIG. 1, the advertiser bids on the keyword “English conversation classroom” in order to distribute the advertising content with the LP as the link destination. However, it is not easy for advertisers to bid on appropriate keywords. For example, the advertiser determines whether the correspondence between the bid keyword and the advertisement content provided by the advertiser is appropriate and appropriate advertisement content is distributed to users who are interested in the search results. It is difficult.

  Therefore, the advertising device 100 according to the embodiment extracts bid keywords suitable for the advertiser in the search-linked advertisement distribution service as shown in FIG. Hereinafter, the extraction process according to the embodiment will be described with reference to FIG. FIG. 2 is a diagram illustrating an example of the extraction process according to the embodiment. In FIG. 2, an example of the extraction process will be described by taking the extraction system 1 as an example.

  As shown in FIG. 2, the extraction system 1 includes an advertiser terminal 20 and an advertising device 100. The advertising device 100 is communicably connected to the advertiser terminal 20 via a network N (for example, the Internet) not shown.

  The advertiser terminal 20 is an information processing device used by the advertiser. The advertiser terminal 20 is, for example, a mobile terminal such as a smartphone, a tablet terminal, or a PDA, a desktop PC, a notebook PC, or the like.

  The advertising device 100 is a server device that extracts an appropriate keyword as a bid keyword for an advertiser to bid by an extraction process described below.

  The advertising device 100 holds in advance information regarding existing LPs and bid keywords associated with the existing LPs. The existing LP indicates a link destination of advertisement content that has been submitted to the advertising device 100 in the past by a predetermined advertiser and that may already be distributed to the user terminal 10. For example, the advertising device 100 collects data related to the LP such as an HTML (HyperText Markup Language) file from a predetermined web server based on the URL (Uniform Resource Locator) of the LP, and holds the collected data as an existing LP. A bid keyword associated with an existing LP is a keyword that triggers the display of advertising content with the existing LP as a link destination when input from a user as a search query. This keyword has already accepted a bid.

  Here, the advertising device 100 generates feature information of existing LPs and bid keywords. For example, the advertising device 100 generates a word vector of existing LPs and bid keywords as feature information (step S11). First, the advertising device 100 generates a word vector corresponding to an existing LP. The word vector corresponding to LP is, for example, a representation of LP by a vector model after giving a predetermined weight to the type and number of appearances of words included in LP. Thus, the advertising device 100 generates the feature information of the LP by expressing the tendency of the word appearing in the LP by the word vector.

  Subsequently, the advertising device 100 generates a word vector corresponding to the bid keyword. The advertising device 100 generates the word vector of the bid keyword by using the existing LP word vector associated with the bid keyword. For example, the advertising device 100 sets an existing LP word vector associated with the bid keyword as the word vector of the bid keyword. In addition, when there are a plurality of existing LPs associated with the bid keyword (that is, when a plurality of advertisers are bidding on the same keyword), the advertising device 100 includes a plurality of LPs associated with the bid keyword. A barycentric vector of each word vector of LP is generated. Then, the advertising device 100 sets the generated center-of-gravity vector as the word vector of the bid keyword. The advertising device 100 holds the generated word vector in a predetermined storage unit.

  Then, the advertising device 100 accepts submission of advertising content from the new advertiser CL10 that intends to use the search-linked advertisement distribution service via the advertiser terminal 20 (step S12). In addition, the advertising device 100 receives information (for example, the URL of the new LP) related to the new LP that is the web page linked to the advertising content when the advertising content is submitted. Here, the new LP means a web page linked to the advertising content not linked to the bid keyword because the keyword bid from the advertiser has not yet been accepted.

  Subsequently, the advertising device 100 collects information on the new LP transmitted from the advertiser terminal 20 in the same manner as the existing LP, and acquires the collected information. Then, the advertising device 100 generates feature information that the acquired new LP has. For example, the advertising device 100 generates a word vector corresponding to the new LP.

  Then, the advertising device 100 calculates the similarity between the word vector of the bid keyword and the word vector of the new LP (step S13). In other words, the advertising device 100 calculates the similarity between the existing LP word vector associated with the bid keyword and the new LP word vector. For example, the advertising device 100 calculates the similarity between word vectors by calculating the cosine similarity between word vectors as an example of the similarity. Then, the advertising device 100 extracts a keyword related to a word vector for which the calculated similarity is equal to or greater than a predetermined threshold (step S14).

  Then, the advertising device 100 presents the extracted keyword to the advertiser CL10 as a bid keyword associated with the new LP (step S15). The advertiser CL10 can display the submitted advertisement content on the search result page 50 when the user performs a search by inputting the bid keyword in the search service by using the presented keyword as a bid keyword. it can.

  As described above, the advertising device 100 according to the embodiment calculates the similarity between the feature information of the existing LP linked to the keyword and the feature information of the new LP, and the calculated similarity is equal to or greater than a predetermined threshold. A keyword corresponding to the feature information is extracted.

  In other words, the advertising device 100 according to the embodiment extracts a bid keyword associated with an existing LP that is similar in characteristics to the new LP as a keyword associated with the new LP. In addition, since the advertising device 100 calculates the similarity between the word vector of the new LP and the centroid vectors related to the plurality of existing LPs, a large number of advertisers can be used without focusing on the keywords related to the specific existing LPs. Can extract the keywords that are bidding. As described above, it is difficult for an advertiser who newly uses a search-linked advertisement distribution service to determine whether or not the bid keyword and the LP are appropriately associated with each other. On the other hand, according to the advertising device 100 according to the embodiment, a new LP is created based on information on keywords currently existing by other advertisers who use the search-linked advertisement distribution service and existing LPs. Keywords to be linked can be extracted. Thereby, an advertiser who provides a new LP can specify an appropriate bid keyword. Then, the advertiser can distribute the advertising content to a user who is assumed to be interested in the advertising content by using an appropriate keyword for the new LP as a bid keyword. For this reason, according to the advertising device 100 according to the embodiment, the advertising effect of the advertising content such as the appealing effect exhibited by the distributed advertising content can be enhanced.

[2. Configuration of Advertising Device 100]
Next, the configuration of the advertising device 100 according to the embodiment will be described with reference to FIG. FIG. 3 is a diagram illustrating a configuration example of the advertising device 100 according to the embodiment. As illustrated in FIG. 3, the advertising device 100 includes a communication unit 110, a storage unit 120, and a control unit 130.

(About the communication unit 110)
The communication unit 110 is connected to the network N by wire or wireless, and transmits / receives information to / from the user terminal 10, the advertiser terminal 20, and the search server 30. For example, the communication unit 110 is realized by a NIC (Network Interface Card) or the like.

(About the storage unit 120)
The storage unit 120 is realized by, for example, a semiconductor memory device such as a RAM (Random Access Memory) or a flash memory, or a storage device such as a hard disk or an optical disk. In the example of FIG. 3, the storage unit 120 includes an LP information storage unit 121 and a keyword information storage unit 122.

(About LP information storage unit 121)
The LP information storage unit 121 stores information about LP. Here, FIG. 4 shows an example of the LP information storage unit 121 according to the embodiment. In the example illustrated in FIG. 4, the LP information storage unit 121 includes items such as “advertiser ID”, “LP”, “advertisement content”, and “bid keyword”.

  The “advertiser ID” indicates identification information for identifying the advertiser or the advertiser terminal 20. In the following, the identification information stored in the advertiser ID shown in FIG. 4 may be used as an advertiser reference code. For example, the advertiser identified by the advertiser ID “CL01” may be referred to as “advertiser CL01”.

  “LP” indicates information about LP provided by the advertiser. The notation “LP01” shown in FIG. 4 represents identification information for identifying the LP and conceptually represents information about the LP. The information regarding the LP stored in the LP information storage unit 121 may be an HTML file or the like constituting the LP, or may be information indicating a URL (Uniform Resource Locator) where the LP data is placed. . In the following, the information stored in the LP item shown in FIG. 4 may be used as the LP reference code. For example, an LP having information indicated by LP “LP01” may be expressed as “LP01”.

“Advertising content” indicates the advertising content submitted from the advertiser terminal 20. FIG. 4 shows an example in which conceptual information such as “AD01” and “AD02” is stored in “advertising content”, but in actuality, text data that is advertising content of advertising content, or these data Stores the file path name indicating the storage location.

  “Bid keyword” indicates a bid keyword associated with each LP (that is, each advertisement content). As shown in FIG. 4, the advertiser can bid on a plurality of keywords for one LP. In addition, the bid keyword may be a combination of a plurality of words instead of a single word.

  For example, in FIG. 4, in order for the advertiser CL01 identified by the advertiser ID “CL01” to display the advertising content “AD01” with “LP01” as the link destination on the search result page 50, the bid keyword “K001” is displayed. , “K002”, “K003”, “K004”,...

(Keyword information storage unit 122)
The keyword information storage unit 122 stores information on keywords. Here, FIG. 5 shows an example of the keyword information storage unit 122 according to the embodiment. In the example illustrated in FIG. 5, the keyword information storage unit 122 includes items such as “bid keyword”, “LP”, “LP word vector”, and “keyword word vector”.

  “Bid keyword” corresponds to the item of the bid keyword shown in FIG. “LP” corresponds to the item of LP shown in FIG. Note that “LP” in FIG. 5 indicates the LP associated with the bid keyword.

  “LP word vector” indicates a word vector corresponding to each LP. The notation “V01” shown in FIG. 5 is a conceptual representation of a word vector. A word vector corresponding to each LP is generated by the generation unit 132 described later, and is stored in the keyword information storage unit 122 by the generation unit 132.

  “Keyword word vector” indicates a word vector corresponding to a bid keyword. The notation “G01” shown in FIG. 5 is a conceptual representation of a word vector. As will be described in detail later, the word vector corresponding to the bid keyword is generated based on the LP word vector associated with the bid keyword.

  For example, in FIG. 5, the bid keyword “K001” is linked to LPs “LP01”, “LP02”, “LP03”,..., And the word vectors corresponding to each LP are LP word vectors. In this example, the word vector corresponding to the bid keyword “K001” is “V01”, “V02”, “V03”,.

(About the control unit 130)
In the control unit 130, for example, various programs stored in a storage device inside the advertising device 100 are executed by using a RAM (Random Access Memory) as a work area by a CPU (Central Processing Unit), an MPU (Micro Processing Unit), or the like. It is realized by doing. The control unit 130 is realized by an integrated circuit such as an application specific integrated circuit (ASIC) or a field programmable gate array (FPGA).

  As shown in FIG. 3, the control unit 130 includes a submission reception unit 131, a generation unit 132, a calculation unit 133, an extraction unit 134, a presentation unit 135, a request reception unit 136, and a distribution unit 137. And realizes or executes the functions and operations of information processing described below. Note that the internal configuration of the control unit 130 is not limited to the configuration illustrated in FIG. 3, and may be another configuration as long as the information processing described below is performed. Further, the connection relationship between the processing units included in the control unit 130 is not limited to the connection relationship illustrated in FIG. 3, and may be another connection relationship.

(About the submission acceptance part 131)
The submission accepting unit 131 accepts submission of advertisement content. In addition, the submission accepting unit 131 accepts the URL of the LP as a link destination of the advertising content and the bid keyword corresponding to the advertising content together with the submission of the advertising content. Then, the submission receiving unit 131 stores the received advertisement content, information about the LP, and the bid keyword in the LP information storage unit 121.

  The submission accepting unit 131 accepts submission of advertisement content from an advertiser who newly uses the advertisement distribution service. In this case, the submission receiving unit 131 sends information related to the received new LP to the generating unit 132 in order to generate a word vector of the new LP that is the link destination of the received advertising content.

(About the generator 132)
The generation unit 132 generates feature information of LPs and bid keywords received by the submission receiving unit 131. For example, the generation unit 132 generates an existing LP, a new LP, and a word vector corresponding to a bid keyword as feature information. Then, the generation unit 132 stores the generated word vector in the keyword information storage unit 122.

  When generating the word vector corresponding to the LP, the generation unit 132 collects the LP data based on the LP URL, and acquires the collected LP data. Then, for example, the generation unit 132 performs morphological analysis on the text included in the LP. That is, the generation unit 132 decomposes the text included in the LP for each word. Then, the generation unit 132 generates a word vector corresponding to the LP based on the type of word included in the LP and the number of appearances of the word in the LP. For example, the generating unit 132 generates a word vector corresponding to LP by giving an element “1” for a word that appears in LP and giving an element “0” for a word that does not appear.

  The generation unit 132 also assigns a weight to each word and generates a word vector corresponding to LP. Here, with respect to the weighting of the word vector, for example, a method of extracting a feature word by a known feature extraction algorithm such as tf-idf (Term Frequency-Inverse Document Frequency) can be considered. In this case, for a word extracted as a feature word that characterizes LP, the generation unit 132 generates a word vector with a higher weight. Thereby, the production | generation part 132 can produce | generate the word vector reflecting the characteristic of LP.

  The generation unit 132 may use a technique called tf-ilf (Term Frequency-Inverse Label Frequency) instead of the above-described tf-idf. In the above-described tf-idf, df (Document Frequency) is used as a frequency indicating how many documents are commonly used in all documents (in this case, it corresponds to “document”). What you do is LP). On the other hand, in tf-ilf, a frequency (lf, Label Frequency) indicating how much a word to be processed included in LP is linked to a keyword (that is, a label) is used.

  In this case, for the word vector corresponding to LP, the generation unit 132 is based on the total number of appearances of all words in the LP, the number of occurrences of the predetermined word in the LP, and the frequency with which the predetermined word is associated with the keyword. Thus, a word vector weighted for each predetermined word is generated. Specifically, the generation unit 132 calculates a ratio between the total number of appearances of all words in the LP and the number of occurrences of the predetermined word in the LP (that is, tf. Appearance frequency of the predetermined word in the LP). ) And the frequency with which the word is associated with the keyword (ie, lf), a word vector corresponding to LP is generated. As a result, the generation unit 132 generates a word vector that is appropriately weighted so that the weight of an ordinary word that is associated with a keyword very frequently is not excessively weighted. For this reason, the generation unit 132 can generate a word vector indicating the characteristics of the LP more appropriately.

  In addition, the generation unit 132 generates a word vector corresponding to the bid keyword. When there are a plurality of LPs associated with the bid keyword, the generation unit 132 generates a centroid vector of each word vector corresponding to the plurality of LPs associated with the bid keyword as a word vector corresponding to the bid keyword. This point will be described with reference to FIG.

  FIG. 6 is a conceptual diagram illustrating an example of a generation process according to the embodiment. As shown in FIG. 6, LP01, LP02, LP03,... Are associated with the bid keyword K001. The generating unit 132 generates word vectors V01, V02, V03,... Corresponding to LP01, LP02, LP03,. Then, the generation unit 132 generates a centroid vector of the generated word vector of each LP as a word vector corresponding to the bid keyword K001. That is, the generation unit 132 generates the centroid vector of the word vector of each LP as the word vector G01 of the bid keyword K001. As described above, the generation unit 132 generates a word vector corresponding to the bid keyword by using the centroid vector of each word vector of the plurality of LPs linked to each other. Can be generated.

(About the calculation unit 133)
The calculation unit 133 calculates the similarity between the existing LP feature information associated with the keyword and the new LP feature information including an arbitrary word. Here, since the advertising device 100 accepts bids for a plurality of keywords and submission of a plurality of advertisement contents in the advertisement distribution service, there are a plurality of combinations of bid keywords and existing LPs. At this time, the calculation unit 133 calculates the similarity of the feature information for each keyword. Taking FIG. 5 as an example, when the bid keyword K001 is a processing target, the calculation unit 133 includes the feature information of each of LP01, LP02, and LP03, which are existing LPs associated with the bid keyword K001, and new The similarity with the LP feature information is calculated. That is, the calculation unit 133 is similar to the feature information of all the existing LPs associated with the keyword and the feature information of the new LP for each keyword for a plurality of combinations in which the keyword and the existing LP are associated. Calculate the degree.

  The calculation unit 133 uses a word vector as the feature information. For example, the calculation unit 133 calculates the similarity between the word vector corresponding to the bid keyword and the word vector corresponding to the new LP, which is generated based on the existing LP feature information. Specifically, the calculation unit 133 calculates the cosine similarity between the word vectors as the similarity between the word vector corresponding to the keyword associated with the existing LP and the word vector corresponding to the new LP. That is, the calculation unit 133 calculates the cosine similarity between two word vectors that are targets of calculation processing. As described above, the generation unit 132 gives an element “1” for a word that appears in LP, and gives an element “0” for a word that does not appear, and generates a word vector corresponding to LP. The unit 133 calculates a predetermined numerical value indicated in the range of “0 to 1” as the cosine similarity of two word vectors.

  Note that, when calculating the similarity, the calculation unit 133 may extract word vectors that are assumed to have a relatively high similarity, and calculate the similarity between the extracted word vectors. . For example, the calculation unit 133 extracts a processing target using an algorithm such as Top-k retry (Top-k search). That is, the calculation unit 133 extracts the top k word vectors (k is an arbitrary number) of word vectors ordered by a score given by a predetermined method (for example, tf-ilf described above), and the extracted word vectors The similarity between each other is calculated. Thereby, since the calculation part 133 skips a process about the process target with a low score, it can calculate a similarity at high speed.

  In addition, when there are a plurality of existing LPs linked to the keyword, the calculation unit 133 corresponds to the centroid vector of each word vector corresponding to the plurality of existing LPs linked to the keyword and the new LP. The similarity with the word vector is calculated. The extraction unit 134 also extracts keywords associated with a plurality of existing LPs whose similarity calculated by the calculation unit 133 is equal to or greater than a predetermined threshold. Thus, the calculation unit 133 averages and captures the characteristics of a plurality of existing LPs that other advertisers use as bid keywords. For this reason, the calculation unit 133 can calculate the degree of similarity in consideration of a situation where a plurality of existing LPs are associated with keywords.

  As described above, the calculation unit 133 calculates the similarity between the feature information of the keyword associated with the existing LP and the feature information of the new LP, and thus indicates the relationship between the keyword and the new LP by a numerical value. Can do. Then, the calculation unit 133 sends the calculated result to the extraction unit 134 described later.

(About the extraction unit 134)
The extraction unit 134 extracts keywords corresponding to feature information whose feature information similarity calculated by the calculation unit 133 is equal to or greater than a predetermined threshold. For example, based on the cosine similarity of the word vector calculated by the calculation unit 133, the extraction unit 134 extracts a keyword corresponding to a word vector whose cosine similarity is greater than or equal to a predetermined threshold.

  For example, as the cosine similarity between word vectors is higher (that is, closer to the value of “1”), the extraction unit 134 has a higher relevance between the keyword corresponding to the word vector to be calculated and LP. Is determined. Therefore, the extraction unit 134 sets a predetermined threshold for the similarity, and extracts a keyword corresponding to a word vector related to the similarity equal to or higher than the predetermined threshold. The extraction unit 134 can arbitrarily set a predetermined threshold value. For example, the extraction unit 134 sets a predetermined threshold according to the number of keywords to be extracted. That is, the extraction unit 134 may set the predetermined threshold value low when extracting a large number of keywords.

(About the presentation unit 135)
The presentation unit 135 presents the keyword extracted by the extraction unit 134 to the advertiser who is the provider of the new LP. Specifically, the presentation unit 135 is extracted by the extraction unit 134 as a keyword that triggers the display of the new LP when selected by the user in the search service (that is, a bid keyword associated with the new LP). Present keywords to advertisers.

  Note that the presentation unit 135 may not present the keyword as a bid keyword to the advertiser even if it is the keyword extracted by the extraction unit 134. For example, the presentation unit 135 may not present the excluded keyword that can be set for the new LP. The exclusion keyword that can be set for the new LP corresponds to, for example, the name of a company competing with the advertiser of the new LP or the name of a product handled by the competing company.

(About the request receiving unit 136)
The request reception unit 136 receives a request regarding distribution of advertisement content. For example, the request reception unit 136 receives an HTTP (Hypertext Transfer Protocol) request included in a web page displayed by the user terminal 10 as an advertisement content acquisition request. Further, the request receiving unit 136 receives information related to the search query from the search server 30.

(About distribution unit 137)
The distribution unit 137 distributes advertising content. For example, the distribution unit 137 distributes the advertising content to the user terminal 10 in accordance with the HTTP request received by the request reception unit 136. When the request reception unit 136 receives an advertisement content acquisition request, the distribution unit 137 selects the distribution candidate advertisement content from the LP information storage unit 121.

  Here, when the search query received by the request reception unit 136 and the bid keyword are the same or similar, the distribution unit 137 selects the advertisement content associated with the bid keyword as a distribution candidate. Then, the distribution unit 137 distributes the selected advertisement content to the user terminal 10 that has transmitted the search query to the search server 30.

[3. (Extraction procedure)
Next, the procedure of the extraction process executed by the advertising device 100 according to the embodiment will be described with reference to FIG. FIG. 7 is a flowchart illustrating an extraction processing procedure by the advertising device 100 according to the embodiment.

  As shown in FIG. 7, the advertising device 100 generates a word vector of a bid keyword associated with an existing LP (step S101). Then, the advertising device 100 determines whether or not new advertisement content submission has been received from a new advertiser using the advertisement distribution service (step S102). When the advertising device 100 has not acquired the advertising content (step S102; No), the advertising device 100 stands by until it is acquired.

  On the other hand, when the advertising device 100 newly acquires advertising content (step S102; Yes), the advertising device 100 generates a new LP word vector that is a link destination of the advertising content (step S103). Then, the advertising device 100 calculates the similarity between the bid keyword and the new LP (step S104).

  Subsequently, the advertising device 100 extracts keywords related to the similarity equal to or higher than a predetermined threshold based on the calculated similarity (step S105). Then, the advertising device 100 presents the extracted keyword as a bid keyword associated with the new LP (step S106).

  FIG. 7 illustrates an example in which the advertising device 100 acquires a new LP after generating a word vector of a bid keyword associated with an existing LP. However, without being limited to this example, the advertising device 100 may generate a word vector of a bid keyword associated with an existing LP after acquiring a new LP. In this case, the advertising device 100 stores the generated new LP word vector in the predetermined storage unit 120. In addition, after presenting the bid keyword associated with the new LP to the advertiser in step S106, the advertising device 100 may extract a different keyword. That is, the advertising device 100 may repeatedly execute step S105 and step S106 depending on the correspondence of the advertiser to the extracted keyword. For example, when the advertiser desires a keyword other than the keyword presented, the advertising device 100 changes the threshold value and extracts a different keyword. Then, the advertising device 100 presents the newly extracted keyword to the advertiser.

[4. (Modification)
The embodiment described above may be implemented in various different forms other than the above embodiment. Accordingly, other embodiments will be described below.

[4-1. Advertising content)
In the above embodiment, the advertising device 100 generates an existing LP, a bid keyword associated with the existing LP, a word vector corresponding to the new LP, and calculates a similarity between the generated word vectors. An example of extracting keywords to be associated with a new LP has been shown. Here, the advertising device 100 may extract a keyword associated with the new LP using other information, not limited to the LP. For example, the advertising device 100 can use advertising content that links to LP instead of LP. This point will be described with reference to FIG.

  FIG. 8 is a diagram illustrating an example of the LP information storage unit 121 according to the modification. The advertising device 100 according to the modification includes the LP information storage unit 121 illustrated in FIG. 8 instead of the LP information storage unit 121 illustrated in FIG. 4. As illustrated in FIG. 8, the LP information storage unit 121 according to the modification includes an item “CTR (Click Through Rate)” in addition to the LP information storage unit 121 illustrated in FIG. 4.

  “CTR” indicates the CTR of the advertising content. Here, CTR indicates a value obtained by dividing the number of times the advertisement content is selected by the user (for example, the number of times the user terminal 10 is clicked) by the number of times the advertisement content is displayed. The CTR of advertising content that has not been distributed at the present time includes a predetermined fixed value, an average value of CTR of all advertising content, or CTR of all advertising content belonging to the same advertising category. The average value is stored. Further, “CTR” may store a predicted CTR predicted from a CTR prediction model or the like. Such a predicted CTR is predicted based on, for example, the type of advertising content, the type of web page on which the advertising content is displayed, and the like.

  For example, FIG. 8 shows an example in which the advertiser CL01 has submitted the advertising content “AD01” with “LP01” as the link destination, and the CTR of the advertising content AD01 is “0.01”. Further, the advertiser CL01 is bidding on bid keywords “K001”, “K002”, “K003”, “K004”,... As keywords for displaying the advertisement content AD01 along with the search result. Is shown.

  Subsequently, the generation unit 132 according to the modification will be described. As described above, the generation unit 132 according to the embodiment generates the word vector corresponding to the existing LP, but the generation unit 132 according to the modification generates a word vector corresponding to the advertising content.

  For example, the generation unit 132 performs morphological analysis on text data included in the advertising content, as in the embodiment. Then, the generation unit 132 generates a word vector corresponding to the advertising content. In addition, the generation unit 132 generates a word vector corresponding to a bid keyword associated with the advertisement content (that is, a keyword associated with LP) based on the word vector corresponding to the advertisement content.

  In addition, the generation unit 132 generates a word vector corresponding to the advertising content with the new LP as a link destination. Then, the calculation unit 133 calculates the similarity between the word vector corresponding to the bid keyword generated by the generation unit 132 and the word vector corresponding to the advertising content with the new LP as a link destination. Then, based on the similarity calculated by the calculation unit 133, the extraction unit 134 extracts a keyword corresponding to the word vector related to the similarity greater than or equal to a predetermined threshold. As described above, according to the advertising device 100 according to the modification, it is possible to extract a keyword associated with the new LP using the advertising content.

  Further, the advertising device 100 according to the modification may perform the above extraction process using an index indicating the advertising effect of the advertising content. For example, in the generation processing of the word vector corresponding to the bid keyword associated with the advertising content, the generation unit 132 generates the barycentric vector of the word vector of each advertising content by increasing the weight of the word vector as the advertising content has a higher CTR. To do.

  Alternatively, the generation unit 132 selects an advertisement content having a high CTR from the advertisement content associated with the bid keyword in the generation processing of the word vector corresponding to the bid keyword associated with the advertisement content, and selects each advertisement content selected. The centroid vector may be generated using the word vector. As described above, the generation unit 132 according to the modification generates the word vector by reflecting the CTR of the advertising content, and thus can generate the word vector reflecting the characteristics of the advertising content with higher advertising effect. Further, the calculation unit 133 according to the modified example can calculate the similarity that reflects the advertising content with higher advertising effect.

  Note that the generation unit 132 may generate a word vector based on an index indicating an advertising effect other than the CTR illustrated in FIG. For example, the generation unit 132 may generate a word vector based on information such as CVR (Conversion Rate) indicating a percentage of conversion to the number of times advertisement content is displayed. In this case, the calculation unit 133 can calculate the similarity that reflects the advertising content with a high CVR.

  Note that conversion refers to a final result that can be acquired by the advertiser, and corresponds to user behavior that leads to profits for the advertiser. Examples of conversion include an action of purchasing a product on a web page, an action of registering a member, an action of requesting a material, an action of inquiring about a product and the like. In addition, the conversion may include an action in which a user who accesses the advertising content makes a purchase, member registration, document request, inquiry, etc. offline (telephone or mail) without going through the advertiser's web page. .

  In addition, the submission receiving unit 131 may acquire information related to the number of times a bid keyword is input from the user in the search service. Then, the generation unit 132 may generate the barycentric vector by increasing the weight of the advertising content with the larger number of times the bid keyword is input from the user. In this case, the calculation unit 133 calculates the similarity between the word vectors based on the number of times the bid keyword is input from the user.

  As described above, the calculation unit 133 according to the modified example displays the number of times the keyword associated with the advertising content (or existing LP) is input from the user, the number of times the advertising content is selected from the user, or the advertising content. Thus, the similarity can be calculated based on at least one of the information on the profit obtained by the advertiser.

  As a result, the advertising device 100 can calculate the similarity according to the actual situation based on the current search service and the status of the advertisement distribution, so that an appropriate bidding keyword can be selected for the advertiser of the new LP. Can be extracted.

  Note that the calculation unit 133 may calculate the degree of similarity using information related to different advertisement evaluations other than the example shown above. For example, the calculation unit 133 displays the actual display count (impressions) of the advertising content on the user terminal 10, CPC (Cost Per Click) that is a billing amount generated when the advertising content is clicked, Information related to advertising content such as eCPM (effective cost per mill), which is the billing amount, may be used.

  Moreover, although the example which calculates the similarity between advertising content was shown in the said example, the similarity may be calculated with the feature information of advertisement content, and the feature information of LP.

[4-2. Learning)
Further, after the extracted keyword is associated with the new LP, the advertising device 100 may acquire a result such as how many times the advertising content with the new LP as a link destination is displayed. That is, the advertising device 100 acquires the number of times the advertising content is displayed, the number of times the advertising content is clicked by the user, and the like. In other words, the advertising device 100 measures the effect when the advertiser uses the extracted keyword as a bid keyword. Then, the advertising device 100 may further use the acquired index, such as CTR or CVR, for the extraction process. As described above, the advertising device 100 may measure a result such as an advertisement effect by the extracted keyword, and may extract the keyword by a learning model in which the result is reflected. Thereby, the advertising device 100 can extract keywords that are more easily searched by the user and can improve the advertising effect by repeating the processing. For this reason, the advertising device 100 can further enhance the advertising effect of the advertising content.

[4-3. Advertiser)
In the above embodiment, the advertiser has shown an example in which the advertising content is submitted to the advertising device 100 using the advertiser terminal 20. However, the advertiser may request such processing from the agency without submitting the advertising content to the advertising device 100 itself. In this case, it is an agency that performs the processing related to the advertising device 100. That is, the notation “advertiser” is a concept including not only the advertiser but also the agency, and the notation “advertiser terminal” includes not only the advertiser terminal 20 but also the agency terminal used by the agency. It is assumed to be a concept.

[4-4. (Advertisement delivery)
In addition, when there are a plurality of advertisement contents associated with the same bid keyword in the distribution of the advertisement content, the distribution unit 137 displays an advertiser with a high bid price of the bid keyword, an advertisement content with a high CTR, a bid price and a CTR. It is also possible to preferentially select advertising content that is both high. Then, the distribution unit 137 distributes the selected advertisement content to the user terminal 10.

[4-5. Other uses)
In the above embodiment, the advertising device 100 has shown an example in which bid keywords are extracted in a search-linked advertisement distribution service. However, the advertising device 100 may be used for advertisement distribution other than the search-linked type as described above. For example, the advertising device 100 may be used for a so-called advertising content matching type advertising service that accepts bids for keywords and distributes the advertising content to web pages having high affinity with the keywords.

  Moreover, in the said embodiment, LP or advertisement content was mentioned as an example of 1st content and 2nd content, and the bid keyword was mentioned as an example as an example of a keyword. However, the advertising device 100 may be used not only for extracting bid keywords associated with LPs or advertising content but also for other purposes. For example, the advertising device 100 calculates keyword similarities between text documents based on feature information of a predetermined text document, and extracts keywords related to the text document. It may be a device that executes That is, the first content and the second content may be content such as a document including an arbitrary word.

  When the extraction apparatus according to the present application corresponds to the search server 30, the search server 30 may display a keyword that assists the search query on the search result page 50. That is, the search server 30 calculates the similarity between the feature information of a predetermined keyword and the feature information of a predetermined web page. Then, the search server 30 extracts a keyword corresponding to the feature information whose calculated similarity is greater than or equal to a predetermined threshold as a keyword for searching a predetermined web page. In this case, the search server 30 may present the extracted keyword on the web page that provides the search service. The user may be able to efficiently search for a target web page by executing a search based on the presented keyword. In this case, the search server 30 includes processing units corresponding to the generation unit 132, the calculation unit 133, the extraction unit 134, and the presentation unit 135 according to the above embodiment, in addition to the control unit that executes the search service.

[4-6. Others]
In addition, among the processes described in the above embodiment, all or part of the processes described as being automatically performed can be performed manually, or the processes described as being performed manually can be performed. All or a part can be automatically performed by a known method. In addition, the processing procedures, specific names, and information including various data and parameters shown in the document and drawings can be arbitrarily changed unless otherwise specified. For example, the various types of information illustrated in each drawing is not limited to the illustrated information.

  Further, each component of each illustrated apparatus is functionally conceptual, and does not necessarily need to be physically configured as illustrated. In other words, the specific form of distribution / integration of each device is not limited to that shown in the figure, and all or a part thereof may be functionally or physically distributed or arbitrarily distributed in arbitrary units according to various loads or usage conditions. Can be integrated and configured.

  For example, the LP information storage unit 121 and the keyword information storage unit 122 illustrated in FIG. 3 may be held by a storage server or the like (not shown) without being held by the advertising device 100. In this case, the advertising device 100 acquires LP information and the like from the storage server.

  Further, for example, in the above-described embodiment, an example in which the advertising device 100 generates a word vector has been described. However, the entity that generates the word vector may be a predetermined server device different from the advertising device 100. In this case, the advertising device 100 acquires a word vector generated by a predetermined server device and holds the acquired word vector. Further, for example, the advertising device 100 described above includes a word vector generation device having a submission receiving unit 131 and a generation unit 132, an extraction device having a calculation unit 133 and an extraction unit 134, and a presentation device having a presentation unit 135. And a distribution apparatus having a request reception unit 136 and a distribution unit 137.

[5. Hardware configuration)
Further, the advertising device 100 according to the embodiment described above is realized by a computer 1000 having a configuration as shown in FIG. 9, for example. Hereinafter, the advertisement apparatus 100 will be described as an example. FIG. 9 is a hardware configuration diagram illustrating an example of a computer 1000 that implements the functions of the advertising device 100. The computer 1000 includes a CPU 1100, RAM 1200, ROM 1300, HDD 1400, communication interface (I / F) 1500, input / output interface (I / F) 1600, and media interface (I / F) 1700.

  The CPU 1100 operates based on a program stored in the ROM 1300 or the HDD 1400 and controls each unit. The ROM 1300 stores a boot program executed by the CPU 1100 when the computer 1000 is started up, a program depending on the hardware of the computer 1000, and the like.

  The HDD 1400 stores a program executed by the CPU 1100, data used by the program, and the like. The communication interface 1500 receives data from other devices via the communication network 500 (corresponding to the network N) and sends the data to the CPU 1100, and transmits data generated by the CPU 1100 to other devices via the communication network 500. .

  The CPU 1100 controls an output device such as a display and a printer and an input device such as a keyboard and a mouse via the input / output interface 1600. The CPU 1100 acquires data from the input device via the input / output interface 1600. Further, the CPU 1100 outputs the data generated via the input / output interface 1600 to the output device.

  The media interface 1700 reads a program or data stored in the recording medium 1800 and provides it to the CPU 1100 via the RAM 1200. The CPU 1100 loads such a program from the recording medium 1800 onto the RAM 1200 via the media interface 1700, and executes the loaded program. The recording medium 1800 is, for example, an optical recording medium such as a DVD (Digital Versatile Disc) or PD (Phase change rewritable disk), a magneto-optical recording medium such as an MO (Magneto-Optical disk), a tape medium, a magnetic recording medium, or a semiconductor memory. Etc.

  For example, when the computer 1000 functions as the advertising device 100 according to the embodiment, the CPU 1100 of the computer 1000 implements the function of the control unit 130 by executing a program loaded on the RAM 1200. The HDD 1400 stores data in the storage unit 120. The CPU 1100 of the computer 1000 reads these programs from the recording medium 1800 and executes them, but as another example, these programs may be acquired from other devices via the communication network 500.

[6. effect〕
As described above, the advertising device 100 according to the embodiment includes the calculation unit 133 and the extraction unit 134. The calculation unit 133 calculates the similarity between the existing LP feature information associated with the keyword and the new LP feature information including an arbitrary word. Further, the extraction unit 134 extracts a keyword corresponding to feature information whose similarity calculated by the calculation unit 133 is equal to or greater than a predetermined threshold.

  As described above, the advertising device 100 according to the embodiment is a keyword that is expected to be highly effective because other advertisers are currently bidding, and is associated with an existing LP that is similar in characteristics to the new LP. Extract keywords. Thereby, according to the advertising device 100, when the user uses the search service, it is possible to easily display the advertising content associated with the bid keyword, so that the advertising effect of the advertising content can be improved.

  In addition, the calculation unit 133 includes, as the feature information for each keyword among the plurality of combinations in which the keyword and the existing LP are associated, the feature information of all the existing LPs associated with the keyword, and the new LP Similarity with the feature information is calculated.

  As described above, the advertising device 100 according to the embodiment calculates the degree of similarity for each keyword among the combinations of the plurality of keywords and LP. Thereby, the advertising device 100 can extract an appropriate keyword for the new LP.

  In addition, the calculation unit 133 calculates the similarity between the word vector corresponding to the existing LP linked to the keyword and the word vector corresponding to the new LP as the feature information.

  Thereby, since the advertising device 100 according to the embodiment can express the feature of the content such as LP including an arbitrary word, the feature of the new LP can be accurately captured. For this reason, the advertising device 100 can extract an appropriate keyword for the new LP.

  In addition, the calculation unit 133 calculates the cosine similarity between the word vector corresponding to the existing LP linked to the keyword and the word vector corresponding to the new LP. The extraction unit 134 extracts a keyword corresponding to a word vector whose cosine similarity calculated by the calculation unit 133 is equal to or greater than a predetermined threshold.

  Thereby, since the advertising device 100 according to the embodiment can express the similarity of the tendency of the text included in the LP or the like with a numerical value, the similarity between the word vectors can be accurately calculated. For this reason, the advertising device 100 can extract an appropriate keyword for the new LP.

  In addition, when there are a plurality of existing LPs linked to the keyword, the calculation unit 133 corresponds to the centroid vector of each word vector corresponding to the plurality of existing LPs linked to the keyword and the new LP. The similarity with the word vector is calculated. Further, the extraction unit 134 extracts a keyword corresponding to a centroid vector whose similarity calculated by the calculation unit 133 is equal to or greater than a predetermined threshold.

  Thereby, the advertising device 100 according to the embodiment can average and grasp the characteristics of a plurality of existing LPs that other advertisers have set as bid keywords. For this reason, since the advertising device 100 can calculate the degree of similarity in consideration of a situation where a plurality of existing LPs are associated with keywords, it is possible to extract an appropriate keyword for the new LP.

  In addition, the advertising device 100 according to the embodiment includes a word vector corresponding to an existing LP linked to a keyword, a centroid vector of each word vector corresponding to a plurality of existing LPs linked to the keyword, and a new LP. A generation unit 132 that generates at least one of the corresponding word vectors is further provided. The calculation unit 133 calculates the similarity using at least one of the word vectors generated by the generation unit 132.

  Thereby, the advertising device 100 according to the embodiment can calculate the similarity of each content using the feature information such as LP that captures the feature of the content.

  In addition, the generation unit 132 adds, for a word vector corresponding to an existing LP or a new LP, the total number of appearances of all the words in the existing LP or the new LP, the number of occurrences of the predetermined word, and the predetermined word A word vector weighted for each predetermined word is generated based on the frequency associated with the keyword.

  Thereby, the advertising device 100 according to the embodiment can generate a word vector to which an appropriate weight is given. For this reason, according to the advertising device 100, since a highly reliable similarity can be calculated, an appropriate keyword for a new LP can be extracted.

  In addition, the calculation unit 133 uses the advertisement content submitted by the advertiser or the web page (that is, the existing LP) linked to the advertisement content as the first content, and the search query that the advertiser bids as the keyword. The bid keyword is used, and the degree of similarity is calculated using a content (for example, a new LP) different from the content corresponding to the first content as the second content.

  Thereby, the advertising device 100 according to the embodiment can appropriately extract the keyword associated with the new LP in the search-linked advertisement distribution service.

  In addition, the calculation unit 133 displays the number of times a keyword associated with an existing LP is input from the user, the number of times that the advertisement content that links to the existing LP is selected by the user, or the advertisement content is displayed. The similarity is calculated on the basis of at least one of the information on the profit obtained.

  Thereby, the advertising device 100 according to the embodiment can calculate the similarity between the feature information of the existing LP and the new LP based on the current advertising effect. For this reason, the advertising device 100 can extract an appropriate keyword in accordance with the current situation.

  In addition, the advertising device 100 according to the embodiment further includes a presentation unit 135 that presents the keyword extracted by the extraction unit 134 to the provider of the new LP as a keyword associated with the new LP.

  Thereby, the advertising device 100 according to the embodiment can present an appropriate keyword to be bid to an advertiser who is a provider of a new LP. Therefore, according to the advertising device 100, the advertiser can know an appropriate keyword to be bid. Further, the manager of the advertisement distribution service can promote bidding by increasing the number of keywords bid by the advertiser, and can activate the service.

  As described above, some of the embodiments of the present application have been described in detail based on the drawings. It is possible to implement the present invention in other forms with improvements.

  In addition, the “section (module, unit)” described above can be read as “means” or “circuit”. For example, the calculation unit can be read as calculation means or a calculation circuit.

DESCRIPTION OF SYMBOLS 1 Extraction system 10 User terminal 20 Advertiser terminal 30 Search server 100 Advertising apparatus 110 Communication part 120 Storage part 121 LP information storage part 122 Keyword information storage part 130 Control part 131 Submission acceptance part 132 Generation part 133 Calculation part 134 Extraction part 135 Presentation section 136 Request reception section 137 Distribution section

Claims (12)

A calculation unit that calculates the similarity between the feature information of the first content linked to the keyword and the feature information of the second content including an arbitrary word;
An extraction unit that extracts a keyword corresponding to feature information having a similarity calculated by the calculation unit equal to or greater than a predetermined threshold among the keywords;
An extraction device comprising: The calculation unit includes:
Of the plurality of combinations in which the keyword and the first content are linked, for each keyword, the feature information of all the first content linked to the keyword and the feature information of the second content The similarity of
The extraction unit includes:
Extracting a keyword corresponding to feature information whose similarity between the feature information of the keyword calculated by the calculation unit and the feature information of the second content is equal to or greater than a predetermined threshold;
The extraction device according to claim 1. The calculation unit includes:
The similarity is calculated using the word vector corresponding to the first content associated with the keyword and the word vector corresponding to the second content as the feature information,
The extraction unit includes:
Extracting a keyword corresponding to the word vector whose similarity is equal to or greater than a predetermined threshold;
The extraction apparatus according to claim 1 or 2, wherein The calculation unit includes:
Calculating a cosine similarity between a word vector corresponding to the first content linked to the keyword and a word vector corresponding to the second content;
The extraction unit includes:
Extracting a keyword corresponding to the word vector whose cosine similarity is greater than or equal to a predetermined threshold;
The extraction device according to claim 3. The calculation unit includes:
When there are a plurality of first contents associated with the keyword, a centroid vector of each word vector corresponding to the plurality of first contents associated with the keyword, and a word vector corresponding to the second content, The similarity of
The extraction unit includes:
Extracting a keyword corresponding to the centroid vector whose similarity is equal to or greater than a predetermined threshold;
The extraction device according to claim 3 or 4, characterized in that. At least one of a word vector corresponding to the first content linked to the keyword, a centroid vector of each word vector corresponding to the plurality of first contents linked to the keyword, and a word vector corresponding to the second content Generating unit for generating
Further comprising
The calculation unit includes:
Calculating the similarity using at least one of the word vectors generated by the generation unit;
The extraction device according to any one of claims 3 to 5, wherein The generator is
For a word vector corresponding to the first content or the second content, the sum of the number of appearances of all words in the first content or the second content, the number of times a predetermined word appears, and the predetermined word are Generating the word vector weighted for each predetermined word based on the frequency associated with the keyword;
The extraction apparatus according to claim 6. The calculation unit includes:
As the first content, an advertising content submitted by an advertiser or a web page linked to the advertising content is used, and as the keyword, a bid keyword which is a search query bid by the advertiser is used, and the second content is used. And calculating the similarity using content different from the content corresponding to the first content,
The extraction device according to any one of claims 1 to 7, wherein The calculation unit includes:
The number of times a keyword associated with the first content is input by the user, the number of times the first content is selected by the user, or the profit obtained by the provider of the first content by displaying the first content Calculating the similarity based on at least one of the information;
The extraction device according to any one of claims 1 to 8, wherein A presenting unit that presents the keyword extracted by the extracting unit to a provider of the second content as a keyword associated with the second content;
The extraction device according to claim 1, further comprising: An extraction method performed by a computer,
A calculation step of calculating the similarity between the feature information of the first content linked to the keyword and the feature information of the second content including an arbitrary word;
An extraction step of extracting a keyword corresponding to the feature information whose similarity calculated by the calculation step is equal to or greater than a predetermined threshold;
The extraction method characterized by including. A calculation procedure for calculating the similarity between the feature information of the first content linked to the keyword and the feature information of the second content including an arbitrary word;
An extraction procedure for extracting a keyword corresponding to the feature information whose similarity calculated by the calculation procedure is equal to or greater than a predetermined threshold;
An extraction program characterized by causing a computer to execute.

Images