JP2007164378A - Related term extraction device and related term extraction method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a device for extracting and storing related knowledge and a related term of information related to advertisement stored in a storage device connected via a communication line so as to analyze a vocabulary frequently used in a specific business category or industry by a provider of a Web document. <P>SOLUTION: This related term extraction device 500 associating vocabulary data related to each other from a plurality of pieces of vocabulary data receives the Web documents stored in Web servers 100a-c connected via a communication line network 30, stores the received Web documents, receives input of first advertisement vocabulary data related to the vocabulary data to be extracted, extracts the Web documents vocabulary including the inputted first advertisement data, extracts second vocabulary data included in the Web documents in common, generates a domain 410 wherein the extracted second advertisement vocabulary data are associated with the first advertisement vocabulary data, and stores the generated domain. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a related word extracting apparatus and method for extracting related words. In particular, the present invention relates to an apparatus and method for extracting vocabulary frequently included in many documents as related words.

Conventionally, as a method for extracting related words, for example, related word search is known in which language analysis such as morphological analysis is performed on a predetermined sentence to extract related words. Patent Document 1 describes an extraction device that can extract words related to a predetermined theme in consideration of the details about the theme related to the predetermined keyword and the reliability of the user reflecting the level of interest. .
Japanese Patent Laid-Open No. 11-053375

  However, in Patent Document 1, as described below, no consideration is given to an apparatus and a method related to extraction of related terms between information provided by a plurality of servers actually connected via a communication network. .

  In general, when providing information by a Web document via the Internet, a Web document related to this vocabulary or a link to the Web document is displayed on the user's terminal based on the vocabulary input by the user for searching or the like. To be done.

  That is, effective advertisement is performed by displaying advertisement information related to the search keyword in the search result of the search keyword input by the user. In such an aspect, in order to improve the advertising effect on the user, it becomes a problem what kind of advertisement content is displayed together with the search result according to the type of the search keyword input from the user.

  For example, when a life insurance company displays an advertisement, if the vocabulary entered by the user is “insurance”, it is considered appropriate to display the advertisement of this life insurance company, but “guarantor” It is difficult to determine whether it is appropriate to display a life insurance company advertisement. Therefore, it is desirable to have an index that can determine whether “insurance” and “guarantor” are related words for the life insurance company.

  On the other hand, companies such as life insurance companies set up a Web server for their advertisements and promotions, and provide Web documents to customers. In this way, web document providers who have launched a web server have competitors other than their own in order to organize and analyze industry knowledge such as vocabulary, technical terms, and product information often used in specific industries and industries. It is desirable to collect information from Web documents provided by organizations and organizations related to this industry.

  The present inventors have noted that it is desirable for Web document providers to provide an apparatus and method that can extract related vocabulary between information stored in storage devices connected via a communication network. .

  An object of the present invention is to provide information related to advertisements and advertisements stored in a storage device connected via a communication line in order to analyze a vocabulary frequently used by a Web document provider in a specific industry / industry. It is an object of the present invention to provide an apparatus and a method by which a Web document provider can collect and organize industry knowledge by extracting and storing (industry knowledge).

(1) A related word extracting device (for example, a related word described later) that associates advertising vocabulary data with each other from a plurality of advertising vocabulary data in order for a Web document provider to analyze a vocabulary frequently used in a specific industry / industry. An extraction device 500),
A receiving unit (for example, a communication unit 510 described later) for receiving a Web document stored in a storage device connected via a communication line;
A Web document storage unit (for example, a Web document storage unit 530 described later) that stores the Web document received by the reception unit;
An input unit (for example, an input unit 550 described later) that receives input of first advertising vocabulary data related to the extracted advertising vocabulary data;
A Web document extraction unit (for example, an extraction unit 525 described later) that extracts a Web document containing the first advertising vocabulary data input via the input unit from the Web document storage unit;
An extraction unit (for example, an extraction unit 525 described later) that extracts second advertising vocabulary data commonly included in the Web document;
A domain generating unit (for example, a domain generating unit 527) that generates a domain in which the second advertising vocabulary data extracted by the extracting unit is associated with the first advertising vocabulary data;
A domain storage unit (for example, a domain storage unit 540) that stores the domain generated by the domain generation unit;
A related word extraction device.

According to the invention described in (1), the related word extracting device receives a Web document stored in a storage device connected via a communication line, stores the received Web document, and extracts an advertisement. Accepting the input of the first advertising vocabulary data related to the vocabulary data, extracting the web document including the first advertising vocabulary data, and extracting the second advertising vocabulary data commonly included in the extracted web document. The extracted second advertisement vocabulary data is extracted, a domain associated with the first advertisement vocabulary data is generated, and the generated domain is stored.
Therefore, the second advertisement vocabulary data is extracted from the Web document including the first advertisement vocabulary data input by the user, and a domain that is data in which the first advertisement vocabulary data and the second advertisement vocabulary data are associated is generated. It is possible.
Therefore, according to the related word extracting device, related words of information (Web document) related to advertisements and advertisements stored in the storage device connected via the communication line based on the input first advertisement vocabulary data. Can be extracted as the second advertisement vocabulary data, and data in which the first advertisement vocabulary data and the second advertisement vocabulary data are associated with each other can be generated.

  (2) The domain generation unit includes a first advertisement vocabulary data different from the first advertisement vocabulary data, and a domain generated from the second advertisement vocabulary data extracted from the other first advertisement vocabulary data; The related word extracting device according to (1), wherein the domain storage unit associates a domain that is already stored.

According to the invention described in (2), the related word extracting device includes other first advertisement vocabulary data different from the first advertisement vocabulary data, and second advertisement vocabulary data extracted from the other first advertisement vocabulary data. Associate the domain generated from the domain with the already stored domain.
Therefore, by extracting the second advertisement vocabulary data from the first advertisement vocabulary data and associating the generated domain with the newly generated domain, it is possible to generate data indicating the relationship between the domains. It is.

  (3) When extracting the second advertisement vocabulary data commonly included in the Web document, the extraction unit preferentially extracts the second advertisement vocabulary data having a high frequency (1) or (2 ) Related word extraction device.

According to the invention described in (3), the related word extraction device preferentially extracts the second advertisement vocabulary data having a high frequency when extracting the second advertisement vocabulary data commonly included in the Web document. To do.
Therefore, since the second advertisement vocabulary data is extracted from the vocabulary included in the Web document based on the frequency of occurrence, it is possible to extract the second advertisement vocabulary data appropriate as the association degree of the related words.

(4) A related word extraction method for associating advertisement vocabulary data related to each other from a plurality of advertisement vocabulary data in order for a web document provider to analyze vocabulary frequently used in a specific industry / industry.
Receiving a Web document stored in a storage device connected via a communication line;
Storing the web document received in the receiving step;
An input step for receiving input of first advertising vocabulary data related to the advertising vocabulary data to be extracted;
An extraction step of extracting a Web document including the first advertising vocabulary data input in the input step;
A second advertising vocabulary data extracting step for extracting second advertising vocabulary data commonly included in the Web document;
A domain generating step of generating a domain in which the second advertising vocabulary data extracted by the second advertising vocabulary data extracting step is associated with the first advertising vocabulary data;
A domain storage step of storing the domain generated by the domain generation step;
A related word extraction method comprising:

  (5) The domain generation step includes a first advertisement vocabulary data different from the first advertisement vocabulary data, and a domain generated from the second advertisement vocabulary data extracted from the other first advertisement vocabulary data; The related word extraction method according to (4), wherein in the domain storing step, a domain already stored is associated.

  According to the present invention, by extracting and storing vocabulary frequently used in a specific type of business / industry such as advertisements and promotions stored in a storage device connected via a communication line, this stored information can be stored on the Web. Document providers can use it for analysis of industry knowledge.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
FIG. 1 is a configuration diagram of a related word extraction system 1 which is a preferred embodiment of the present invention. The related word extraction system 1 includes a related word extraction device 500 and Web servers 100a to 100c. The related word extraction device 500 and the Web servers 100a to 100c are connected to be communicable via the communication line network 30.

  The related word extraction device 500 is an information processing device that extracts related words, and may be a computer including a CPU (Central Processing Unit) and a memory. The related word extraction device 500 generates a communication unit 510 that connects to the communication line network 30 to realize communication, a control unit 520 that controls information and data, a Web document storage unit 530 that stores Web documents, The domain storage unit 540 that stores the domain, the input unit 550 that receives input from the user, and the output unit 560 that outputs information and data.

  The control unit 520 controls information and data processed by the related word extraction device 500, and may be a CPU, for example. The control unit 520 includes an extraction unit 525 that extracts a Web document and second advertisement vocabulary data, and a domain generation unit 527 that generates a domain to be described later. The control unit 520 executes main processing (flowchart in FIG. 2) and domain association processing (flowchart in FIG. 8) described later.

  The web document storage unit 530 is a device in which the control unit 520 stores web documents received from the web servers 100a to 100c via the communication unit 510. Here, the web document is electronic data in which sentences and images that can be browsed via a communication line are described, and may be a web page. Here, the Web document may be electronic data including advertisement vocabulary data, and includes information related to corporate promotion and advertisement, that is, vocabulary, technical terms, product information, and industry knowledge that are often used in a specific industry / industry. It may be. The advertising vocabulary data is data relating to vocabulary information related to company promotion and advertisement.

  The Web document storage unit 530 may be in a form that the related word extraction apparatus 500 does not have. That is, the Web document storage unit 530 is provided in a device such as a server or a computer connected to the communication line network 30, and the related word extraction device 500 appropriately transmits the Web document and data related to the Web document via the communication unit 510. It is possible to read from the Web document storage unit 530.

  The domain storage unit 540 is a device in which the control unit 100 stores the domain generated by the domain generation unit 527. Here, the domain is data in which a plurality of advertising vocabulary data is associated as will be described later with reference to FIGS.

  The input unit 550 is a device that receives input from the user, and may be, for example, a keyboard or a mouse. The output unit 560 is a device that outputs information and data. For example, the output unit 560 may be an output device such as a monitor, a liquid crystal display, or a printer.

  Each of the Web servers 100a to 100c is a computer that provides content to be browsed by a web browser, stores a web document, and transmits the web document to a requested computer in response to a web document browse request. That is, the Web servers 100 a to 100 c transmit Web documents to the related word extraction device 500.

  The user terminal 200 is a computer for operating the related word extraction device 500 from a remote location. The user terminal 200 includes an input unit 250 that receives information and data, a communication unit 210 that is connected to the communication line network 30 to realize communication, a control unit 220 that controls the information and data, and outputs information and data. Output unit 260. Instead of the user inputting information and data via the input unit 550 of the related word extraction device 500, information and data are input from the input unit 250 of the user terminal 200.

  Next, main processing executed by the control unit 520 of the related word extraction device 500 will be described with reference to FIG. The control unit 520 receives input of first advertising vocabulary data (extracted keyword) via the input unit 550 or the input unit 250 of the user terminal 200 (step S01). The first advertisement vocabulary data (extracted keyword) is advertisement vocabulary data that is the basis of the vocabulary extracted as related words.

  Next, the extraction unit 525 extracts a Web document including the first advertising vocabulary data from the Web document storage unit 530 (step S02).

  Here, the mode in which the extraction unit 525 extracts the Web document including the first advertisement vocabulary data from the Web document storage unit 530 has been described in Step S02. However, the Web document storage unit 530 stores the Web document summary data. , A part of data of a Web document including a keyword that characterizes the Web document, such as lead sentence information and index information, may be stored, and the extracting unit 525 A mode of extracting data including the first advertisement vocabulary data may be used.

  In any of the above aspects, the Web document or a part of the Web document is stored in advance in the Web document storage unit 530 before the main processing flow is executed.

  FIG. 3 is a Web document correspondence table diagram in which extracted keywords input by the user, Web documents related to the extracted keywords, and link data are associated with each other. The user inputs “keyword 01” as an extraction keyword, and a Web document corresponding to the keyword is extracted from the text portion of the Web document. As a result, the extracted data becomes Web documents A to D. Link data (for example, link destination URL data) corresponding to these Web documents A to D is associated. Note that the Web document correspondence table is stored in the Web document storage unit 530 and read out to the control unit 520 as appropriate.

  FIG. 4 shows a screen image diagram in which the input of the first advertising vocabulary data is received and the result of extracting the Web document is output to the output units 260 and 560. First, input of the first advertisement vocabulary data (search keyword) is received from the input window 305. Here, it is assumed that the user inputs “loan” and extracts a Web document or partial data of the Web document. The extraction unit 525 extracts Web documents (for example, corresponding to Web documents A and B) of XX trading 310 and loan XX trading 320 with respect to the extracted keyword. Then, as illustrated in FIG. 4, the control unit 620 may display link data 311 and 321 corresponding to these Web documents on the output units 260 and 560.

  Next, the second advertisement vocabulary data is extracted from the Web document extracted by the extraction unit 525 (step S03). The second advertisement vocabulary data is advertisement vocabulary data included in common with one or more Web documents extracted by the first advertisement vocabulary data. In the extraction of the second advertising vocabulary data, the vocabulary extracted as the second advertising vocabulary data may be selected based on the co-occurrence frequency. That is, the vocabulary frequently included in the Web document may be preferentially extracted by the extraction unit 525 as the second advertising vocabulary data.

  FIG. 5 shows an example of Web documents 600 and 605 extracted when a Web document is extracted using the extracted keyword shown in FIG. The extraction unit 525 extracts the vocabulary commonly included in the Web documents 600 and 605 as the second advertisement vocabulary data. Here, as the second advertisement vocabulary data, the extraction unit 525 extracts “same day” 610 a and 610 b and “real annual rate” 615 a and 615 b common to the Web documents 600 and 605.

  Next, the domain generation unit 527 generates a domain by associating the extracted keyword (first advertisement vocabulary data) with the extracted second advertisement vocabulary data (step S04). Here, the domain is data in which a plurality of advertising vocabulary data is associated. An example of the domain 410 is shown in FIG. In this example, as in the examples of FIGS. 4 and 5 described above, “loan” is input from the user as the first advertising vocabulary data, and the extraction unit 525 extracts the second advertising vocabulary related to the first advertising vocabulary data. The case where “same day” and “real annual rate” are extracted as data is shown. Therefore, the domain generation unit 527 generates a domain in which these are associated, and stores the generated domain in the domain storage unit 540 (step S05).

  The extraction unit 525 may extract a plurality of second advertisement vocabulary data to generate a domain, or extract only one vocabulary having the highest appearance frequency (co-occurrence rate) as the second advertisement vocabulary data. A domain may be generated. In the above description, the extracted Web documents are only the two Web documents 600 and 605. However, as the Web document, the control unit can be selected from three or more Web documents as Web documents A to D. 520 may calculate the appearance frequency and the like, extract the second advertising vocabulary data, and the domain generation unit 527 may generate the domain.

  FIG. 7 shows a domain 420 when a plurality of second advertisement vocabulary data is extracted. Here, the relationship between the second advertisement vocabulary data may be arbitrary, but for example, the vocabulary included in the Web documents A to D is “finance” such as “same day” and “interest rate”. ”May be positioned as a vocabulary strongly related to. On the other hand, vocabularies ("repayment", "collateral", "examination") that are included in Web documents A and B but not in Web documents C and D are associated as a group. Vocabularies ("guarantor", "no interest", "real annual rate") that are not included in A and B but are included in Web documents C and D may be associated as another group.

  Next, a process of associating domains after a plurality of domains generated by the domain generation unit 527 are generated will be described with reference to FIG. First, the domain generation unit 527 determines that a new domain has been stored in the domain storage unit 540 (step S10). When the domain generation unit 527 determines that the new domain has been stored in the domain storage unit 540, the domain generation unit 527 has the same first advertisement vocabulary as the second advertisement vocabulary data included in the generated new domain. It is determined whether a domain including data is stored in domain storage unit 540 (step S11). When the domain generation unit 527 determines that the domain having the same first advertisement vocabulary data as the second advertisement vocabulary data included in the generated new domain is stored in the domain storage unit 540, the corresponding The domain and the new domain are associated (step S12), and a new domain is generated (step S13).

  FIG. 9 shows a conceptual diagram of the domain for explaining the domain association process. Assume that the domain A is already stored in the domain storage unit 540. Domain A uses the first advertising vocabulary data as “loan”. The domain generation unit 527 generates a domain B in which the first advertisement vocabulary data is “guarantor”. Since the first advertising vocabulary data “guarantor” of domain B is the same as “guarantor” of the second advertising vocabulary data of domain A, “guarantor” of domain B and “guarantor” of domain A are associated with each other. . Similarly, domain C and domain D are associated with domain A, and a new domain associated with a plurality of domains is stored in domain storage unit 540.

  On the other hand, in the domain E, the same advertisement vocabulary data does not exist in all of the domains A to D. That is, there is no domain that includes the first advertising vocabulary data “digital camera” of domain E in the second advertising vocabulary data, and there is no domain that includes the second advertising vocabulary data of domain E in the first advertising vocabulary data. Therefore, the domain E is a domain that is less relevant to the domains A to D. As described above, it is possible to generate data capable of grasping the correlation between domains by newly generating domains between domains.

  The related term extraction device 500 described above provides a search service provided via a communication line such as the Internet (receives input of specific advertising vocabulary data and provides link data of a Web document related to the inputted advertising vocabulary data. Applied to the service), the vocabulary related to the word to be searched may be used to present to the searching user.

  Furthermore, the information extracted by the related word extraction apparatus 500 and stored as a domain in the domain storage unit 540 is not limited to advertisements and advertisement information of companies and the like, and may be arbitrary technical terms. That is, instead of the first advertising vocabulary data, the first specialized vocabulary data related to a predetermined specialized field is input and the second vocabulary data is extracted, whereby a domain related to this specialized field is generated and stored in the domain storage unit 540. Remembered. For example, as the first specialized vocabulary data, data related to cooking is input, and a domain related to cooking is generated, whereby a domain storage unit 540 that stores related words related to cooking can be provided.

  By using the related term extraction device 500 described above, it is possible to extract vocabulary related to a Web document for the promotion or advertisement of one product or service, and generate data indicating the relationship between the extracted related terms. . That is, it is possible to extract a vocabulary commonly included in a Web document provided by each of a plurality of companies that provide one product and service. For example, when the user inputs “insurance” as a search keyword, it is assumed that insurance company A wants to include a link of its own Web document in the search result. In this case, in addition to the vocabulary “insurance”, an index as to what vocabulary (for example, “guarantor”, etc.) should be included in the search result for the Web document of insurance company A is necessary. Become. In such a case, the domain generated by the related word extraction device 500 becomes an index as a vocabulary related to the search keyword. Therefore, the related word extraction apparatus 500 may be applied as a service that provides the vocabulary related to the search keyword in this way.

  As an embodiment, the present invention can be realized by a computer program of each computer operating in the related word extraction system 1. The storage medium storing the program can be an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system (or apparatus or device) or a propagation medium. Examples of computer readable media include semiconductors, magnetic tapes, examples of removable computer readable media include semiconductors, magnetic tapes, removable computer diskettes, random access memory (RAM), read only Memory (ROM), rigid magnetic disk and optical disk are included. Current examples of optical disks include compact disk-read only memory (CD-ROM), compact disk-read / write (CD-R / W) and DVD.

  As mentioned above, although embodiment of this invention was described, it only showed the specific example and does not specifically limit this invention. Further, the effects described in the embodiments of the present invention only list the most preferable effects resulting from the present invention, and the effects of the present invention are not limited to the effects described in the embodiments of the present invention.

1 is a configuration diagram of a related word extraction system 1. FIG. It is a flowchart figure of a main process. It is a Web document correspondence table figure. It is a screen image figure which outputted the result of extracting a Web document. It is a figure which shows a Web document. It is a figure which shows the produced | generated domain. It is a figure which shows the produced | generated domain. It is a flowchart figure of a domain correlation process. It is a conceptual diagram which shows that a some domain is related.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Related word extraction system 30 Communication line network 100 Control part 100a, 100b, 100c Web server 200 User terminal 210 Communication part 220 Control part 250 Input part 260 Output part 305 Input window 311 Link data 321 Link data 410 Domain 420 Domain 500 Related word Extraction device 510 Communication unit 520 Control unit 525 Extraction unit 527 Domain generation unit 530 Document storage unit 540 Domain storage unit 550 Input unit 560 Output unit 600 Document 605 Document 620 Control unit

Claims (5)

In order to analyze a vocabulary frequently used in a specific industry / industry, a web document provider is a related word extraction device that associates advertisement vocabulary data related to each other from a plurality of advertisement vocabulary data,
A receiving unit that receives a Web document stored in a storage device connected via a communication line;
A Web document storage unit for storing the Web document received by the receiving unit;
An input unit for receiving input of first advertising vocabulary data related to the advertising vocabulary data to be extracted;
A Web document extraction unit that extracts a Web document including the first advertising vocabulary data input via the input unit from the Web document storage unit;
An extraction unit for extracting second advertising vocabulary data commonly included in the Web document;
A domain generating unit that generates a domain in which the second advertising vocabulary data extracted by the extracting unit is associated with the first advertising vocabulary data;
A domain storage unit for storing the domain generated by the domain generation unit;
A related word extraction device.   The domain generating unit includes a first advertisement vocabulary data different from the first advertisement vocabulary data, a domain generated from second advertisement vocabulary data extracted from the other first advertisement vocabulary data, and the domain The related word extraction device according to claim 1, wherein the storage unit associates a domain that is already stored with the storage unit.   The extraction unit preferentially extracts second advertising vocabulary data having a high frequency when extracting second advertising vocabulary data commonly included in the Web document. The related word extraction device described. A related word extraction method for associating advertisement vocabulary data related to each other from a plurality of advertisement vocabulary data in order for a web document provider to analyze vocabulary frequently used in a specific industry / industry,
Receiving a Web document stored in a storage device connected via a communication line;
Storing the web document received in the receiving step;
An input step for receiving input of first advertising vocabulary data related to the advertising vocabulary data to be extracted;
An extraction step of extracting a Web document including the first advertising vocabulary data input in the input step;
A second advertising vocabulary data extracting step for extracting second advertising vocabulary data commonly included in the Web document;
A domain generating step for generating a domain in which the second advertising vocabulary data extracted in the second advertising vocabulary data extracting step is associated with the first advertising vocabulary data;
A domain storage step of storing the domain generated by the domain generation step;
A related word extraction method comprising:   The domain generation step includes a first advertisement vocabulary data different from the first advertisement vocabulary data, a domain generated from second advertisement vocabulary data extracted from the other first advertisement vocabulary data, and the domain The related word extraction method according to claim 4, wherein in the storing step, the already stored domain is associated.

Images