JP2014074992A - Information extraction device and method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the accuracy of address recognition.SOLUTION: A television device 30 searches for a Web page including a predetermined keyword on the Internet 49, and as for the extracted Web page and a related Web page related to the Web page, the television device 30 collects at least URL and the image featured values of moving images included in the Web page and the related Web page, and registers at least the collected URL and image featured values of the Web page group in association in a URL dictionary 66 stored in a storage part 36. Also, when the moving image with the displayed URL as the object of recognition is designated from the moving images of television broadcasting, the television device 30 extracts the image featured values from the designated moving image, and collates the image featured values extracted from the moving image with the image featured values registered in the URL dictionary 66, and recognizes, as the URL as the object of recognition, the URL registered in the URL dictionary 66 in association with the image featured values which are similar to the image featured values extracted from the designated moving image.

Description

  The disclosed technology relates to an information extraction apparatus, an information extraction method, and an information extraction program.

  For CF (Commercial Film), which is an advertisement inserted into a television broadcast, in order to guide viewers interested in the advertising target to browse the corresponding web page (for example, URL (Uniform Resource Locator)) may be displayed. When the viewer desires to obtain more information by viewing the CF or the like, the address such as a URL is usually recorded on paper or the like, and manually entered on a browser software or the like started on a PC (Personal Computer) or the like. Enter the address with, and browse the displayed web page. However, it is troublesome to record an address when watching a TV program and input it later to a PC, etc., and it may not be possible to record the address, making it difficult to obtain information in a timely manner. There are challenges.

  In relation to the above, one or more areas that can be character-coded from the image are extracted and converted into character codes, character code information indicating a character string including one or more characters is created, and the character code information indicates A first technique for extracting a communication character string such as a URL from a character string has been proposed.

  In addition, depending on the input recognition mode, the reading target extraction method is automatically switched to automatically extract a character string, a character line, or a word, and the extracted image content is displayed as a character string, a character line, or a word. A second technique for displaying has been proposed. In the second technique, a specific character string, character line, or word displayed on the display unit is recognized by a selection operation.

  In addition, when the input character image is recognized, and the recognized character string is a URL, the result of character recognition is obtained by connecting to the connection destination indicated by the URL via the Internet and obtaining the result. A third technique for verifying whether it is correct has been proposed.

  Further, in the fourth technique, if the URL is displayed on the hanging advertisement image taken by the user, the character recognition of the URL is performed, and if the URL is not displayed, the company name is collated. It has been proposed to specify a URL. In the fourth technique, it is also proposed to specify a URL by collating an image of a company trademark when no URL is displayed on an image taken by a user.

JP 2003-152829 A Japanese Patent No. 4431335 JP 2002-279352 A JP 2002-278857 A

  The first technique and the second technique recognize an address such as a URL by character recognition, and the processing accuracy depends on the accuracy of character recognition, and the recognized address may contain an error. .

  In the third technique, after an address such as a URL is recognized by character recognition, a web page of the recognized address is accessed, and the recognition result includes an error more than the first technique and the second technique. The probability of being lost can be reduced. However, in the third technique, even if an error is included in the recognized address, if there is a web page at the address including the error, the wrong web page is hit in the search of the web page, Correct recognition results may not be obtained.

  In the fourth technique, it is necessary to create a dictionary in which addresses such as URLs of web pages that may be accessed and company names or company trademark images are registered in association with each other. However, there are several methods for creating a dictionary.

  The first creation method is to create a dictionary manually. However, the first creation method has a problem that it takes much time to create a dictionary. The second creation method purchases information to be registered in a dictionary from a company. However, the second creation method has a problem of increasing costs.

  In the third creation method, information to be registered in the dictionary is collected by a web crawler (search robot). However, the third creation method has a problem that the amount of collected information becomes enormous, leading to a decrease in collation accuracy and a decrease in processing speed. In the fourth creation method, only information on a web page including a specific keyword is collected by the web crawler as information to be registered in the dictionary. However, since the web page at the address displayed on the CF of the television broadcast does not necessarily include a specific keyword, the fourth creation method may cause the address of the web page to be searched not to be registered in the dictionary. obtain.

  As described above, any of the existing dictionary creation methods has a problem. If the fourth technique is applied to the extraction of an address such as a URL displayed on a CF of a television broadcast, it may take time or cost, There was a problem that the accuracy of recognition was lowered.

  An object of the disclosed technique is to improve the accuracy of address recognition.

  The disclosed technology includes a collection unit and a recognition unit. The collection unit searches for a web page including a predetermined keyword. In addition, the collection unit, for the web page including the predetermined keyword extracted by the search and the related web page related to the web page, at least the address and the image feature amount of the moving image included in the web page and the related web page. collect. The collection unit stores at least the address of the collected web page group and the image feature amount in the storage unit in association with each other. The recognition unit is extracted from the designated moving image when a moving image whose designated address is to be recognized is designated from the broadcasted moving image and an image feature amount is extracted from the designated moving image. The obtained image feature amount is collated with the image feature amount stored in the storage unit. The recognizing unit recognizes an address registered in the storage unit in association with an image feature amount similar to the image feature amount extracted from the specified moving image as a recognition target address.

  The disclosed technology has an effect that the accuracy of address recognition can be improved as one aspect.

It is a functional block diagram of the information extraction device explained in the embodiment. It is a block diagram which shows schematic structure of the television apparatus which concerns on 1st Embodiment. It is a flowchart which shows an example of an information collection process. It is an image figure which shows an example of a web page. It is an image figure which shows an example of the image feature-value of a moving image. It is a flowchart which shows an example of a dictionary update process. It is a flowchart which shows an example of URL recognition processing. It is a block diagram which shows schematic structure of the television apparatus and server which concern on 2nd Embodiment. It is a sequence diagram which shows the flow of the URL recognition process in 2nd Embodiment.

  Hereinafter, an example of an embodiment of the disclosed technology will be described in detail with reference to the drawings.

[First Embodiment]
FIG. 1 shows an information extraction apparatus 10 according to this embodiment. The information extraction apparatus 10 has been registered in advance when a user viewing a television broadcast is instructed to acquire an address (URL in this embodiment) displayed in the CF inserted in the television broadcast. An apparatus that extracts (recognizes) the URL based on a URL dictionary. The information extraction apparatus 10 also generates a URL dictionary 66 (see FIG. 2). The URL dictionary 66 extracts a web page including a predetermined keyword and a related web page from web pages published on the Internet, and the URL of the extracted web page and a moving image included in the web page. A feature amount is registered in association with each other.

  The information extraction apparatus 10 includes an information collection unit 12, a confirmation unit 14, a deletion unit 16, a storage unit 18, a moving image extraction unit 20, a feature amount extraction unit 22, a recognition unit 24, and a page display unit 26.

  The information collection unit 12 generates a URL dictionary 66. That is, the information collection unit 12 searches a web page including a predetermined keyword from web pages published on the Internet, and relates to the web page including the predetermined keyword extracted by the search and the web page. Extract related web pages. A web page including a predetermined keyword is a web page presumed to correspond to a CF inserted in a television broadcast and displayed with a URL. Examples of the predetermined keyword include “CF”, “CM”, “campaign”, “Stakes” etc. Examples of the related web page include a parent web page including a link to a web page including a predetermined keyword, and a child web page that is a link destination of a link included in the web page including the predetermined keyword.

  In addition, the information collection unit 12 acquires each URL of the extracted web page group and extracts a feature amount of a moving image included in the extracted web page group. Then, the URL of the acquired web page group and the feature amount of the extracted moving image are registered in the URL dictionary 66 in association with the date and time. The URL dictionary 66 is stored in the storage unit 18.

  The confirmation unit 14 confirms whether the CF corresponding to the information registered in the URL dictionary 66 is still inserted and broadcast in the television broadcast. In general, the CF has a predetermined broadcast period, and the CF broadcast is stopped when the predetermined broadcast period elapses. The confirmation unit 14 determines whether or not the corresponding CF is broadcast for information older than a certain period of information registered in the URL dictionary 66. For the information on which the corresponding CF is broadcast, the date and time registered in the URL dictionary 66 is updated. When the confirmation unit 14 determines that the corresponding CF is not broadcast for information that is older than a certain period of information registered in the URL dictionary 66 by the confirmation unit 14, the deletion unit 16 stores the corresponding information in the URL dictionary. Delete from 66.

  When the user viewing the television broadcast is instructed to acquire the URL displayed in the CF inserted in the television broadcast, the moving image extraction unit 20 performs the URL extraction target CF from the television broadcast moving image. To extract the moving image. The feature amount extraction unit 22 extracts the feature amount of the CF moving image extracted by the moving image extraction unit 20.

  The recognizing unit 24 is registered in the URL dictionary 66 by comparing the feature amount of the CF moving image extracted by the feature amount extracting unit 22 with the feature amount of the moving image registered in the URL dictionary 66. Among the information, information corresponding to the CF from which the feature amount of the moving image is extracted is recognized. The page display unit 26 accesses the web page of the URL included in the information recognized by the recognition unit 24, and displays the accessed web page on the display.

  The information collection unit 12 is an example of a collection unit in the disclosed technology, and the recognition unit 24 is an example of a recognition unit in the disclosed technology. The confirmation unit 14 is an example of a confirmation unit in the disclosed technology, and the deletion unit 16 is an example of a deletion unit in the disclosed technology. The moving image extraction unit 20 is an example of a moving image extraction unit in the disclosed technology, the feature amount extraction unit 22 is an example of a feature amount extraction unit in the disclosed technology, and the page display unit 26 is a page in the disclosed technology. It is an example of a display part.

  The information extraction device 10 can be realized by, for example, the television device 30 shown in FIG. The television apparatus 30 includes a CPU 32, a memory 34, a non-volatile storage unit 36, a display 38, a tuner 40, a remote control reception unit 42, and a network control unit 44. The CPU 32, the memory 34, the storage unit 36, the display 38, the tuner 40, the remote control reception unit 42, and the network control unit 44 are connected to each other via a bus 46. An antenna 47 is connected to the tuner 40. Further, the remote control receiving unit 42 receives a signal (for example, an infrared signal) transmitted from the remote control 48. The remote controller 48 is provided with a plurality of buttons including a button for instructing acquisition of a URL (access to a web page of the URL). The network control unit 44 is connected to the Internet 49.

  The storage unit 36 can be realized by an HDD (Hard Disk Drive), a flash memory, or the like. The storage unit 36 as a recording medium stores a URL extraction program 50 and a URL dictionary 66 for causing the television device 30 to function as the information extraction device 10. The CPU 32 reads the URL extraction program 50 from the storage unit 36 and expands it in the memory 34, and sequentially executes the processes of the URL extraction program 50.

  The URL extraction program 50 includes an information collection process 52, a confirmation process 54, a deletion process 56, a moving image extraction process 58, a feature amount extraction process 60, a recognition process 62, and a page display process 64.

  The CPU 32 operates as the information collection unit 12 illustrated in FIG. 1 by executing the information collection process 52. The CPU 32 operates as the confirmation unit 14 illustrated in FIG. 1 by executing the confirmation process 54. The CPU 32 operates as the deletion unit 16 illustrated in FIG. 1 by executing the deletion process 56. The CPU 32 operates as the moving image extraction unit 20 shown in FIG. 1 by executing the moving image extraction process 58. The CPU 32 operates as the feature amount extraction unit 22 illustrated in FIG. 1 by executing the feature amount extraction process 60. The CPU 32 operates as the recognition unit 24 shown in FIG. 1 by executing the recognition process 62. The CPU 32 operates as the page display unit 26 shown in FIG. 1 by executing the page display process 64.

  As a result, the television apparatus 30 that has executed the URL extraction program 50 functions as the information extraction apparatus 10. The URL extraction program 50 is an example of an information extraction program in the disclosed technology.

  The information extraction apparatus 10 can be realized by, for example, a semiconductor integrated circuit, more specifically, an ASIC (Application Specific Integrated Circuit).

  Next, as an operation of the first embodiment, first, an information collection process periodically executed by the information extraction apparatus 10 (television apparatus 30) will be described with reference to FIG. In step 100 of the information collection process, the information collection unit 12 searches the Internet 49 for a web page that includes a predetermined keyword and is not registered in the URL dictionary 66, so that the corresponding web page is found. Determine if it exists. If the determination in step 100 is negative, the information collection process is terminated.

  On the other hand, if a corresponding web page is found in step 100, the determination in step 100 is affirmed and the process proceeds to step 102. In step 102, the information collecting unit 12 acquires information on a web page that includes a predetermined keyword and is not registered in the URL dictionary 66 from the Internet 49. In this embodiment, for example, “CF”, “CM”, “campaign”, “sweep prize”, etc. are used as the predetermined keywords, and the web page from which information is acquired in step 102 is inserted into the television broadcast and the URL is displayed. It is a web page presumed to correspond to the selected CF.

  An example of the web page acquired in step 102 and corresponding to a specific CF is shown in FIG. A web page 70 shown in FIG. 4A includes a character string 72 including “campaign” which is one of predetermined keywords, a character string 74 representing the contents of the campaign, and a CF moving image inserted in a television broadcast. The playback screen 76 is included.

  However, as shown in FIG. 4A, the web page corresponding to a specific CF does not necessarily contain a variety of character strings and CF moving image playback screens in a single web page. Another example of a web page corresponding to a specific CF is shown in FIG. The web pages 78 to 82 shown in FIG. 4B include a character string 72 including “campaign” and a character string 74 representing the content of the campaign, while the web page 80 includes a CF moving image. The playback screen 76 is included in the web page 82. Further, the web page 78 includes a link 84 to the web page 80, the web page 80 includes a link 86 to the web page 82, the web page 78 is the parent web page of the web page 80, and the web page 82 is the web page. 80 child web pages.

  Considering the case where various types of information are distributed and stored in a plurality of web pages as in the example of FIG. 4B, in the next step 104, the information collection unit 12 acquires the web acquired in step 102. Get related web pages related to the page (eg parent and child web pages). By acquiring related web pages, even if web pages containing a predetermined keyword differ from web pages containing a CF moving image, a web page containing a CF moving image is also acquired by a search based on the predetermined keyword. It becomes possible to do.

  In the next step 106, the information collection unit 12 extracts a moving image included in the web page group acquired in steps 102 and 104. In the next step 108, the information collection unit 12 extracts a feature amount from the moving image extracted in step 106. In this embodiment, as shown in FIG. 5 as an example, the telop added to the moving image is extracted by an existing telop extraction technique. Then, the display period (display start time and display end time) of the telop display area in the moving image, the range (coordinate value) of the telop display area, and the value of each pixel in the telop display area are extracted as the feature amount of the moving image.

  In the next step 110, the information collection unit 12 registers the URL of the web page group acquired in steps 102 and 104, the feature amount of the moving image extracted in step 108 and the current date and time in association with the URL dictionary 66. As a result, information as shown in Table 1 below is registered in the URL dictionary 66 as an example.

  Of the moving image feature quantities shown in Table 1, the first information (for example, 5 sec) represents the time (telop display start time) from the beginning of the moving image until the display of the telop display area is started. Yes. The second information (for example, (300,450)-(400,480) etc.) of the moving image feature amounts shown in Table 1 is the range of the telop display area in the moving image (the upper left corner point and the lower right corner point). Coordinate value). In addition, the third information (for example, (20, 4, 50,..., 49), etc.) of the moving image feature amounts shown in Table 1 represents the pixel value of each pixel in the telop display area.

  In Table 1, the URL character string and the feature amount of the moving image are registered in the URL dictionary 66 in a one-to-one correspondence. However, the present invention is not limited to this. For example, as shown in the example of FIG. 4B, when various types of information are distributed and stored in a plurality of web pages, the feature amount of the moving image is set to a web page including a predetermined keyword and the web. A plurality of URL character strings including related web pages related to the page may be associated with each other. In addition, when there are a plurality of moving images in a web page group including a web page including a predetermined keyword and a related web page, the feature amounts of the plurality of moving images may be associated with a plurality of URL character strings, respectively.

  Subsequently, a dictionary update process periodically executed by the information extraction apparatus 10 (television apparatus 30) in the same manner as the information collection process (FIG. 3) will be described with reference to FIG. In step 120 of the dictionary update process, the confirmation unit 14 determines whether information whose registration date is older than the current date by a certain period or more is registered in the URL dictionary 66. If the determination in step 120 is negative, the dictionary update process ends.

  On the other hand, if the determination at step 120 is affirmative, the routine proceeds to step 122. In step 122, the confirmation unit 14 causes the tuner 40 to receive a television broadcast for a certain period (for example, several hours to several days), and stores the received video image of the television broadcast in the storage unit 36. In the next step 124, the confirmation unit 14 extracts sections (CF sections) corresponding to the CF from the television broadcast moving images received by the tuner 40 and stored in the storage unit 36 using the existing CF extraction technique. .

  In step 126, the confirmation unit 14 extracts feature amounts from the individual CF sections extracted in step 124. As the feature quantity extracted in step 126, the same feature quantity as the feature quantity extracted from the moving image in the above-described information collection process (FIG. 3), that is, the telop display start time, the telop display area range, and the telop display area The pixel value of each pixel can be applied.

  In the next step 128, the confirmation unit 14 extracts one piece of information that is older than a certain period registered in the URL dictionary 66. In step 130, the confirmation unit 14 collates the feature amount included in the information extracted in step 128 with the feature amount extracted from each CF section. Note that the feature amount is collated by, for example, calculating an evaluation value for evaluating the degree of similarity for each of the telop display start time, the telop display area range, the pixel value of each pixel in the telop display area, and the like. This can be done by calculating a comprehensive evaluation value (for example, a weighted addition value). In the next step 132, the confirmation unit 14 matches the feature amount included in the information extracted in step 128 within the individual CF sections extracted in the previous step 124 (for example, the above-described comprehensive evaluation). It is determined whether or not there is a CF section whose value is equal to or greater than the threshold.

  If the determination in step 132 is affirmed, it can be determined that the CF corresponding to the information extracted in step 128 continues to be inserted into the television broadcast and being broadcast. Therefore, if the determination in step 132 is affirmed, the process proceeds to step 134. In step 134, the confirmation unit 14 updates the date included in the information extracted in step 128 to the current date, and stores it in the URL dictionary 66. Write back.

  If the determination in step 132 is negative, it can be determined that the CF corresponding to the information extracted in step 128 is highly likely to have been broadcast. Therefore, if the determination in step 132 is negative, the process proceeds to step 136, and in step 136, the deletion unit 16 deletes the information extracted in step 128 from the URL dictionary 66. As a result, it is possible to prevent the URL dictionary 66 from being unnecessarily enlarged because the information on the CF for which broadcasting has ended remains in the URL dictionary 66.

  If the process of step 134 or step 136 is performed, the process proceeds to step 138. In step 138, the confirmation unit 14 determines whether all information older than a certain period registered in the URL dictionary 66 has been extracted from the URL dictionary 66. . If the determination in step 138 is negative, the process returns to step 128, and steps 128 to 138 are repeated until the determination in step 138 is positive.

  As a result, it is confirmed whether or not the corresponding CF is being broadcasted for information older than a certain period registered in the URL dictionary 66, and the date is updated or deleted from the URL dictionary 66 according to the confirmation result. Done. If the determination in step 138 is affirmed, the dictionary update process is terminated.

  Next, URL recognition processing executed by the information extraction device 10 (television device 30) when reception of television broadcast is started by the information extraction device 10 (television device 30) will be described with reference to FIG. In step 140 of the URL recognition process, the moving image extraction unit 20 determines whether or not an instruction to stop receiving television broadcasts has been issued. If the determination in step 140 is affirmative, the URL recognition process is terminated.

  On the other hand, if the determination in step 140 is negative, the process proceeds to step 142, and in step 142, the moving image extraction unit 20 receives a television broadcast. In step 144, the moving image extraction unit 20 holds the received moving image of the television broadcast in the storage unit 18 for a predetermined time (for example, about several tens of seconds to several minutes). In the next step 146, the moving image extraction unit 20 instructs to acquire a URL (access to the web page of the URL) from the remote control 48 based on whether or not a predetermined signal is received by the remote control reception unit 42. It is determined whether the button is pressed. If the determination in step 146 is negative, the process returns to step 140, and step 140 to step 146 are repeated while the determination in step 140 is negative and the determination in step 146 is negative. Therefore, the storage unit 18 is maintained in a state in which the latest moving image for a certain time of the television broadcast is always held.

  By the way, when a user who is watching a television broadcast by the television apparatus 30 is interested in the advertisement target of the CF by the CF being broadcast, and the web page URL or the like is displayed on the CF, the remote controller 48 is used. The button for instructing acquisition of the URL is pressed. Note that the above operation is not limited to the case where the user is interested in the advertisement target of the CF, and the above operation may be performed when applying for a campaign that is being executed with respect to the advertisement object. In the present embodiment, when the above operation is performed by the user, a predetermined signal from the remote control 48 is received by the remote control reception unit 42, so that the determination in step 146 is affirmed and the process proceeds to step 148.

  In step 148, the moving image extraction unit 20 extracts a CF section from the latest moving image for a certain time of the television broadcast held in the storage unit 18 using an existing CF extraction technique. In the next step 150, the feature amount extraction unit 22 extracts a feature amount from the extracted CF section. Note that the feature quantity extracted in step 150 is the same as the feature quantity extracted from the moving image in the information collection process (FIG. 3), that is, the telop display start time, the telop display area range, and the telop display area. The pixel value of each of the pixels can be applied.

  In step 152, the recognition unit 24 collates the feature amount extracted from the CF section with the feature amount of each information registered in the URL dictionary 66. Note that the feature value collation in step 152 is also performed by calculating the evaluation value for evaluating the degree of similarity for each of the telop display start time, the telop display area range, the pixel value of each pixel in the telop display area, and the like. This can be realized by calculating a comprehensive evaluation value from the evaluation value. In step 154, the recognizing unit 24 selects, from the information registered in the URL dictionary 66, information whose feature amount is most similar to the feature amount extracted from the CF section (for example, the total evaluation value is the maximum). To do. Instead of selecting the most similar information, for example, information with a comprehensive evaluation value of similarity equal to or greater than a predetermined value, or a predetermined number of information in descending order of the comprehensive evaluation value of similarity is displayed as a list. The user may be allowed to select from the options.

  In step 156, the page display unit 26 acquires the URL character string included in the information selected in step 154 from the URL dictionary 66, and accesses the web page using the acquired URL character string as a URL via the Internet 49. When the feature amount of the moving image is registered in the URL dictionary 66 in association with each of a plurality of URL character strings, any one of the plurality of URL character strings can be acquired. The URL character string to be acquired may be the URL of a web page that includes a predetermined keyword, the URL of a web page that is the parent of a web page that includes the predetermined keyword, or a child web page of a web page that includes the predetermined keyword. But you can. Alternatively, the URL of each web page may be displayed in a list so that the user can select the URL of the web page to be accessed, or each web page is accessed and each web page is displayed on the display 38. You may do it.

  In the next step 158, the page display unit 26 displays the web page accessed through the Internet 49 in step 156 on the display 38. When displaying the web page on the display 38, the page display unit 26, for example, subdivides a television broadcast display area on the display surface of the display 38, and a blank area generated as the TV broadcast display area is reduced. The web page can be displayed on the screen. Accordingly, a user who has performed an operation of pressing a button for instructing acquisition of a URL on the remote controller 48 acquires more information about the advertising target that is interested, or a campaign that is being performed on the advertising target. It becomes possible to apply for.

  If the process of step 158 is performed, it will return to step 140, and step 140-146 will be repeated while determination of step 140 is denied and determination of step 146 is denied. Then, when the user performs the operation of pressing the button for instructing the URL acquisition of the remote controller 48 again, the processing from step 148 to step 158 is performed again.

  As described above, in the first embodiment, whether or not a web page including a predetermined keyword exists is searched on the Internet 49, and the URL is registered in the URL dictionary 66. Therefore, the URL contains an error. Can eliminate the possibility of being. In the first embodiment, in addition to the web page including the predetermined keyword, information on the parent and child web pages of the web page including the predetermined keyword is also collected and registered in the URL dictionary 66. For this reason, the probability that the URL displayed on the CF of the television broadcast is registered in the URL dictionary 66 is improved, and the URL in which the user operates the remote controller 48 to acquire the URL (access the web page) is correct. The probability of being acquired and displaying an appropriate web page is improved.

[Second Embodiment]
Next, a second embodiment of the disclosed technique will be described. In addition, the same code | symbol is attached | subjected to the part same as 1st Embodiment, and description is abbreviate | omitted. FIG. 8 shows an information extraction system 170 according to the second embodiment. The information extraction system 170 includes a television device 172 and a server 174 connected via the Internet 49.

  The television apparatus 172 according to the second embodiment is the television described in the first embodiment except that the URL extraction program 50A is stored in the storage unit 36 and the URL dictionary 66 is not stored in the storage unit 36. The configuration is the same as that of the device 30. The URL extraction program 50A according to the second embodiment is the URL extraction program described in the first embodiment except that the information extraction process 52, the confirmation process 54, the deletion process 56, and the recognition process 62 are not included. 50.

  The server 174 includes a CPU 176, a memory 178, a nonvolatile storage unit 188, a tuner 180, and a network control unit 184. The CPU 176, the memory 178, the storage unit 188, the tuner 180, and the network control unit 184 are connected to each other via a bus 186. An antenna 182 is connected to the tuner 180. The network control unit 184 is connected to the Internet 49.

  The storage unit 188 can be realized by an HDD (Hard Disk Drive), a flash memory, or the like. A storage unit 188 as a recording medium stores a dictionary operation program 190 and a URL dictionary 66. The CPU 176 reads the dictionary operation program 190 from the storage unit 188, expands it in the memory 178, and sequentially executes the processes included in the dictionary operation program 190. The dictionary operation program 190 includes an information collection process 52, a confirmation process 54, a deletion process 56, and a recognition process 62.

  The CPU 176 operates as the information collection unit 12 illustrated in FIG. 1 by executing the information collection process 52. The CPU 176 operates as the confirmation unit 14 illustrated in FIG. 1 by executing the confirmation process 54. The CPU 176 operates as the deletion unit 16 illustrated in FIG. 1 by executing the deletion process 56. The CPU 176 operates as the recognition unit 24 illustrated in FIG. 1 by executing the recognition process 62. As a result, the server 174 that has executed the dictionary operation program 190 functions as the information extraction device 10 shown in FIG. 1 together with the television device 172. The dictionary operation program 190 is an example of an information extraction program in the disclosed technology.

  Next, the operation of the second embodiment will be described. In the second embodiment, the server 174 operates as the information collection unit 12, the confirmation unit 14, and the deletion unit 16 illustrated in FIG. Accordingly, the information collection process shown in FIG. 3 and the dictionary update process shown in FIG. 6 are performed by the server 174, and the URL dictionary 66 is stored in the storage unit 188 of the server 174 by these processes. Updated.

  Next, the flow of URL recognition processing in the second embodiment will be described with reference to FIG. In the sequence shown in FIG. 9, a user who is watching a television broadcast by the television device 172 performs an operation of pressing a button for instructing acquisition of a URL on the remote controller 48 (“remote control operation” shown in FIG. 9). ).

  When the above operation is performed by the user, the moving image extraction unit 20 of the television device 172, from the latest moving image for a certain time of the television broadcast held in the storage unit 18, as in the first embodiment, A CF section is extracted by an existing CF extraction technique (step 200 in FIG. 9). Next, the feature amount extraction unit 22 of the television apparatus 172 extracts feature amounts from the CF sections extracted by the moving image extraction unit 20 (step 202 in FIG. 9). Further, the feature amount extraction unit 22 of the television device 172 notifies the server 174 of the feature amount extracted from the CF section (step 204 in FIG. 9).

  The recognizing unit 24 of the server 174 notified of the feature amount extracted from the CF section by the television device 172 collates the feature amount extracted from the CF section with the feature amount of each information registered in the URL dictionary 66 ( Step 206 in FIG. Next, the recognition unit 24 of the server 174 has information whose feature quantity is most similar to the feature quantity extracted from the CF section from information registered in the URL dictionary 66 (for example, the comprehensive evaluation value is the largest). Is selected (step 208 in FIG. 9).

  In addition, the recognition unit 24 of the server 174 acquires the URL character string included in the information selected in step 208 from the URL dictionary 66, and accesses the web page using the acquired URL character string as a URL via the Internet 49 (FIG. 9 step 210). Then, the recognition unit 24 of the server 174 transmits the information on the web page acquired by accessing the web page to the television device 172 (step 212 in FIG. 9).

  The page display unit 26 of the television device 172 that has received the web page information from the server 174 displays the web page represented by the received information on the display 38 (step 214 in FIG. 9). Accordingly, a user who has performed an operation of pressing a button for instructing acquisition of a URL on the remote controller 48 acquires more information about the advertising target that is interested, or a campaign that is being performed on the advertising target. It becomes possible to apply for.

  As described above, in the second embodiment, the function as the information extraction device 10 illustrated in FIG. 1 is realized by the cooperation of the television device 172 and the server 174. This eliminates the need for the TV device 172 to perform information collection processing and dictionary update processing, and eliminates the need to store the URL dictionary 66 or perform search processing on the URL dictionary 66, so that the TV device 172 can be configured at low cost. It becomes possible to do.

  In the second embodiment, the URL dictionary 66 is stored in the storage unit 188 of the server 174 and the server 174 also functions as the recognition unit 24. However, the present invention is not limited to this. For example, the server 174 may generate and update the URL dictionary 66, and the latest URL dictionary 66 may be transferred from the server 174 to the television device 172 and stored in the storage unit 36 of the television device 172. In this case, the television device 172 also functions as the recognition unit 24.

  Further, in the above description, a mode in which the information corresponding to the CF among the information registered in the URL dictionary 66 is selected by comparing the feature amount of the CF moving image with the feature amount of the moving image registered in the URL dictionary 66. Although described, the present invention is not limited to this. For example, in addition to the feature amount of the moving image, CF among the information registered in the URL dictionary 66 is used in combination with the result of character recognition performed on the telop display area in the moving image during the telop display period of the moving image. You may make it select the information corresponding to.

  Further, in the above description, a mode in which a part or all of the functions as the information extraction apparatus according to the disclosed technology is incorporated in the television apparatuses 30 and 172 has been described. However, the present invention is not limited thereto, and functions as an information extraction apparatus. A device to be operated may be provided separately from the television devices 30 and 172.

  Furthermore, although the aspect which recognizes URL displayed on the moving image of CF inserted in the television broadcast was demonstrated above, the technique of an indication is applied also when recognizing URL displayed on moving images other than CF. Is possible. In the above description, a URL is described as an address of a web page in the disclosed technique, but the present invention is not limited to this, and may be a shortened URL, another URI (Uniform Resource Identifier), or the like.

  In the first embodiment, the URL extraction program 50, which is an example of the information extraction program according to the disclosed technique, is described in advance in the storage unit 36 of the television device 30. Further, in the second embodiment, the aspect in which the dictionary operation program 190 that is an example of the information extraction program according to the disclosed technique is stored in the storage unit 188 of the server 174 in advance has been described. However, the present invention is not limited to this, and the information extraction program according to the disclosed technology can be provided in a form recorded on a recording medium such as a CD-ROM or a DVD-ROM.

  All documents, patent applications and technical standards mentioned in this specification are to the same extent as if each individual document, patent application and technical standard were specifically and individually stated to be incorporated by reference. Incorporated by reference in the book.

  Regarding the above embodiment, the following additional notes are disclosed.

(Appendix 1)
A web page including a predetermined keyword is searched, and a web page including the predetermined keyword extracted by the search and a related web page related to the web page are at least an address, a moving image included in the web page and the related web page A collecting unit that collects image feature values of an image and stores at least an address of the collected web page group and an image feature value in a storage unit in association with each other;
When a moving image for which the displayed address is to be recognized is specified from the broadcasted moving image and an image feature amount is extracted from the specified moving image, the image feature amount extracted from the specified moving image Is matched with an image feature amount stored in the storage unit, and an address stored in the storage unit in association with an image feature amount similar to the image feature amount extracted from the designated moving image is obtained. A recognition unit that recognizes as an address to be recognized;
An information extraction device.

(Appendix 2)
The collection unit includes a parent web page including a link to a web page including the predetermined keyword as the related web page, and a child web page that is a link destination of a link included in the web page including the predetermined keyword. The information extraction apparatus according to appendix 1, wherein the address and the image feature amount are collected for at least one of the above.

(Appendix 3)
A moving image extraction unit for extracting a designated moving image, which is a recognition target of a displayed address among the broadcasted moving images;
A feature amount extraction unit that extracts the image feature amount from the moving image extracted by the moving image extraction unit;
A page display unit for accessing the web page at the address recognized by the recognition unit and displaying the accessed web page on the display unit;
The information extraction device according to supplementary note 1 or supplementary note 2, further comprising:

(Appendix 4)
Broadcast video is a TV broadcast video,
The information extraction device according to any one of Supplementary Note 1 to Supplementary Note 3, wherein the moving image having the displayed address as a recognition target is a CF moving image included in the television broadcast.

(Appendix 5)
A confirmation unit for confirming whether or not the CF moving image corresponding to the address and the image feature amount stored in association with the storage unit is broadcast on a television broadcast;
A deletion unit that deletes from the storage unit an address and an image feature amount corresponding to a moving image of a CF that has been confirmed not to be broadcast by television broadcast by the confirmation unit;
The information extracting device according to appendix 4, further comprising:

(Appendix 6)
The collection unit stores the date and time in the storage unit in association with at least an address and an image feature amount of the collected web page group,
The confirmation unit includes a CF moving image corresponding to an address and an image feature amount stored in association with a date and time older than a predetermined period among addresses and image feature amounts stored in association with the storage unit. When it is confirmed whether or not the moving picture of the CF is being broadcast by television broadcasting, it is associated with an address and an image feature amount corresponding to the moving picture of the CF. The information extraction apparatus according to appendix 5, wherein the date and time stored in the storage unit is updated to the current date and time (FIG. 6: 134).

(Appendix 7)
A web page including a predetermined keyword is searched, and a web page including the predetermined keyword extracted by the search and a related web page related to the web page are at least an address, a moving image included in the web page and the related web page Collect image feature values of the image, store at least the address of the collected web page group and the image feature values in the storage unit,
When a moving image for which the displayed address is to be recognized is specified from the broadcasted moving image and an image feature amount is extracted from the specified moving image, the image feature amount extracted from the specified moving image Is matched with an image feature amount stored in the storage unit, and an address stored in the storage unit in association with an image feature amount similar to the image feature amount extracted from the designated moving image is obtained. An information extraction method including recognizing as an address to be recognized.

(Appendix 8)
As at least one of a parent web page including a link to a web page including the predetermined keyword and a child web page that is a link destination of a link included in the web page including the predetermined keyword as the related web page, The information extraction method according to appendix 7, wherein the address and the image feature amount are collected.

(Appendix 9)
Of the broadcasted moving images, the specified moving image whose target is the displayed address is extracted, the image feature amount is extracted from the extracted moving image, and the web page of the recognized address is accessed. The information extraction method according to appendix 7 or appendix 8, wherein the accessed web page is displayed on the display unit.

(Appendix 10)
The broadcasted moving image is a moving image of a television broadcast, and the moving image whose recognition target is a displayed address is a moving image of a CF included in the television broadcast, according to any one of appendix 7 to appendix 9. Information extraction method.

(Appendix 11)
It is confirmed whether or not the CF moving image corresponding to the address and the image feature quantity stored in association with the storage unit is broadcast on the television broadcast, and the CF of the CF confirmed to be not broadcast on the television broadcast. The information extraction method according to appendix 10, wherein an address and an image feature amount corresponding to a moving image are deleted from the storage unit.

(Appendix 12)
Store the date and time in the storage unit in association with at least the address and image feature quantity of the collected web page group, and the date and time older than a predetermined period among the addresses and image feature quantity stored in association with the storage unit To check whether the CF moving image corresponding to the address and the image feature value stored in association with is broadcast on TV broadcasting, and confirm that the CF moving image is broadcast on TV broadcasting In such a case, the information extraction method according to appendix 11, wherein the date and time stored in the storage unit is updated to the current date and time in association with an address and an image feature amount corresponding to the moving image of the CF.

(Appendix 13)
On the computer,
A web page including a predetermined keyword is searched, and a web page including the predetermined keyword extracted by the search and a related web page related to the web page are at least an address, a moving image included in the web page and the related web page Collect image feature values of the image, store at least the address of the collected web page group and the image feature values in the storage unit,
When a moving image for which the displayed address is to be recognized is specified from the broadcasted moving image and an image feature amount is extracted from the specified moving image, the image feature amount extracted from the specified moving image Is matched with an image feature amount stored in the storage unit, and an address stored in the storage unit in association with an image feature amount similar to the image feature amount extracted from the designated moving image is obtained. An information extraction program for performing processing including recognition as an address to be recognized.

(Appendix 14)
As at least one of a parent web page including a link to a web page including the predetermined keyword and a child web page that is a link destination of a link included in the web page including the predetermined keyword as the related web page, 14. The information extraction program according to appendix 13, which collects addresses and image feature amounts.

(Appendix 15)
Of the broadcasted moving images, the specified moving image whose target is the displayed address is extracted, the image feature amount is extracted from the extracted moving image, and the web page of the recognized address is accessed. The information extraction program according to supplementary note 13 or supplementary note 14, wherein the accessed web page is displayed on the display unit.

(Appendix 16)
16. The broadcast moving image is a television broadcast moving image, and the moving image whose recognition target is the displayed address is any one of appendix 13 to appendix 15 that is a CF moving image included in the television broadcast. Information extraction program.

(Appendix 17)
It is confirmed whether or not the CF moving image corresponding to the address and the image feature quantity stored in association with the storage unit is broadcast on the television broadcast, and the CF of the CF confirmed to be not broadcast on the television broadcast. The information extraction program according to supplementary note 16, wherein an address and an image feature amount corresponding to a moving image are deleted from the storage unit.

(Appendix 18)
Store the date and time in the storage unit in association with at least the address and image feature quantity of the collected web page group, and the date and time older than a predetermined period among the addresses and image feature quantity stored in association with the storage unit To check whether the CF moving image corresponding to the address and the image feature value stored in association with is broadcast on TV broadcasting, and confirm that the CF moving image is broadcast on TV broadcasting In such a case, the information extraction program according to appendix 17, wherein the date and time stored in the storage unit is updated to the current date and time in association with an address and an image feature amount corresponding to the moving image of the CF.

DESCRIPTION OF SYMBOLS 10 Information extraction apparatus 12 Information collection part 14 Confirmation part 16 Deletion part 18,36,188 Storage part 20 Moving image extraction part 22 Feature-value extraction part 24 Recognition part 26 Page display part 30,172 Television apparatus 32,176 CPU
34,178 Memory 38 Display 40,180 Tuner 42 Remote control receiver 44,184 Network controller 47,182 Antenna 48 Remote control 49 Internet 50,50A Extraction program 66 URL dictionary 170 Information extraction system 174 Server

Claims (8)

A web page including a predetermined keyword is searched, and a web page including the predetermined keyword extracted by the search and a related web page related to the web page are at least an address, a moving image included in the web page and the related web page A collecting unit that collects image feature values of an image and stores at least an address of the collected web page group and an image feature value in a storage unit in association with each other;
When a moving image for which the displayed address is to be recognized is specified from the broadcasted moving image and an image feature amount is extracted from the specified moving image, the image feature amount extracted from the specified moving image Is matched with an image feature amount stored in the storage unit, and an address stored in the storage unit in association with an image feature amount similar to the image feature amount extracted from the designated moving image is obtained. A recognition unit that recognizes as an address to be recognized;
An information extraction device.   The collection unit includes a parent web page including a link to a web page including the predetermined keyword as the related web page, and a child web page that is a link destination of a link included in the web page including the predetermined keyword. The information extraction apparatus according to claim 1, wherein addresses and image feature quantities are collected for at least one of the above. A moving image extraction unit for extracting a designated moving image, which is a recognition target of a displayed address among the broadcasted moving images;
A feature amount extraction unit that extracts the image feature amount from the moving image extracted by the moving image extraction unit;
A page display unit for accessing the web page at the address recognized by the recognition unit and displaying the accessed web page on the display unit;
The information extraction device according to claim 1 or 2, further comprising: Broadcast video is a TV broadcast video,
The information extraction device according to claim 1, wherein the moving image whose target is a displayed address is a CF moving image included in the television broadcast. A confirmation unit for confirming whether or not the CF moving image corresponding to the address and the image feature amount stored in association with the storage unit is broadcast on a television broadcast;
A deletion unit that deletes from the storage unit an address and an image feature amount corresponding to a moving image of a CF that has been confirmed not to be broadcast by television broadcast by the confirmation unit;
The information extraction device according to claim 4, further comprising: The collection unit stores the date and time in the storage unit in association with at least an address and an image feature amount of the collected web page group,
The confirmation unit includes a CF moving image corresponding to an address and an image feature amount stored in association with a date and time older than a predetermined period among addresses and image feature amounts stored in association with the storage unit. When it is confirmed whether or not the moving picture of the CF is being broadcast by television broadcasting, it is associated with an address and an image feature amount corresponding to the moving picture of the CF. The information extraction apparatus according to claim 5, wherein the date and time stored in the storage unit is updated to the current date and time. A web page including a predetermined keyword is searched, and a web page including the predetermined keyword extracted by the search and a related web page related to the web page are at least an address, a moving image included in the web page and the related web page Collect image feature values of the image, store at least the address of the collected web page group and the image feature values in the storage unit,
When a moving image for which the displayed address is to be recognized is specified from the broadcasted moving image and an image feature amount is extracted from the specified moving image, the image feature amount extracted from the specified moving image Is matched with an image feature amount stored in the storage unit, and an address stored in the storage unit in association with an image feature amount similar to the image feature amount extracted from the designated moving image is obtained. An information extraction method including recognizing as an address to be recognized. On the computer,
A web page including a predetermined keyword is searched, and a web page including the predetermined keyword extracted by the search and a related web page related to the web page are at least an address, a moving image included in the web page and the related web page Collect image feature values of the image, store at least the address of the collected web page group and the image feature values in the storage unit,
When a moving image for which the displayed address is to be recognized is specified from the broadcasted moving image and an image feature amount is extracted from the specified moving image, the image feature amount extracted from the specified moving image Is matched with an image feature amount stored in the storage unit, and an address stored in the storage unit in association with an image feature amount similar to the image feature amount extracted from the designated moving image is obtained. An information extraction program for performing processing including recognition as an address to be recognized.