JP7257010B2 - SEARCH SUPPORT SERVER, SEARCH SUPPORT METHOD, AND COMPUTER PROGRAM - Google Patents

Description

According to the present invention, a user, who is the subject of a search, can quickly and reliably display desired information or contents by speaking out of a large amount of information or contents provided by a content provider on a website. The present invention relates to a possible search support server, search support method and computer program.

Various information exists on the website. Currently, it is possible to find the content that the user desires by clicking on the icon on the top page of the website, using a description using words that can be clicked, or using a search window using a search engine such as Google (R) or Yahoo (R). An address to a Web page, or information or content itself is extracted and displayed on a display device that can be visually recognized by the user.

Also, when receiving a search key input by user's utterance, a speech recognition engine provided by Google (R), Microsoft (R), etc. is used to recognize the user's utterance. These commercial speech recognition engines are provided as general-purpose speech recognition engines generated by extensively collecting speech data and machine learning or deep learning.

Therefore, when recognizing the user's utterance by voice, there is a possibility that the search key that the content provider intends to provide to the user cannot be correctly recognized. Therefore, for example, Patent Literature 1 discloses a speech understanding device that uses a plurality of language models to improve speech recognition accuracy.

JP 2010-170137

However, in Patent Document 1, processing is performed in parallel using a plurality of speech recognition engines, and the results of the language understanding engine based on the plurality of recognition results are integrated. However, not only does it take a considerable amount of time to integrate multiple recognition results and language understanding processing results, but the accuracy of recognition results after the integration processing cannot be guaranteed, which is not realistic. Therefore, there is a problem that there is no guarantee that the correct search key can be acquired in a short time by the user's utterance.

SUMMARY OF THE INVENTION The present invention has been made in view of the above circumstances. By using a speech window that can be narrowed down and displayed by the user, as long as the user speaks or selects according to the guide data narrowed down according to his/her purpose, the content provider can quickly access the specific part of the content that the content provider wants to provide to the user. It is an object of the present invention to provide a search support server, a search support method, and a computer program that can guide users accurately and reliably.

In order to achieve the above object, a search support server according to the present invention is a search support server that supports a search for content published on a website, and includes an utterance for selecting tag information indicating a specific portion of content. and a speech window capable of narrowing down the guide data to be displayed from among the guide data by user's utterance. receive the user's utterance of the tag information displayed on the signage window and/or the speech window, narrow down the tag information to be displayed, and select the narrowed down tag information. By receiving the selection, the specific part of the content corresponding to the selected tag information is displayed on the terminal device.

In the search support server according to the present invention, the tag information includes at least a tag name identifying a specific part of the content, text data consisting of a simple sentence or a group of words describing the content of the tag name, and attributes of the tag name. is preferably configured with attribute information indicating

Further, the search support server according to the present invention comprises search data generation means for generating search data that is the basis of guide data to be displayed on the signage window and/or the speech window, and the search data generation means generates the content Extraction/display means for extracting a simple sentence or a group of words indicating the contents based on the contents of and displaying the content corresponding to the selection of the extracted simple sentence or the group of words; and searching for pointer information indicating a specific portion of the content based on the displayed content. pointer search means; pointer allocation reception means for receiving allocation of the searched pointer information for each tag name whose input is received; and attribute information that can be spoken at the time of retrieval for each tag name whose input is received. It is preferable to have attribute information receiving means for receiving an input.

Further, in the search support server according to the present invention, the extraction/display means scrutinizes the contents, extracts the contained character strings as simple sentences or word groups, and outputs them as text data. and content display means for displaying the content corresponding to the selected text data when the selection of the output text data is accepted.

Further, in the search support server according to the present invention, it is preferable that the guide data is generated by associating the tag information included in the search data with search data identification information for identifying the search data.

Further, the search support server according to the present invention further includes speech accepting means for accepting input of voice data uttered by the user in order to narrow down the tag names displayed in the speech window and capable of accepting selection by the user. , converts the received voice data into text data, calculates the degree of matching of the converted text data with the tag name of the guide data and a simple sentence or word group indicating the content of the tag name, and calculates the calculated degree of matching Metatag speech recognition means for identifying the tag name with the maximum value and displaying the specific part of the content corresponding to the tag name when the degree of matching is greater than a predetermined value; extracting tag information that matches the attribute information from among the tag information of the guide data, narrowing down the guide data by logical AND of one or more pieces of attribute information that match; It is preferable to provide a narrowing speech recognition means for outputting the tag name and the attribute information, and to execute the metatag speech recognition means and the narrowing speech recognition means in parallel.

Further, in the search support server according to the present invention, the metatag speech recognition means converts the text data, all the tag names and the first data formed of simple sentences or word groups indicating the contents of the tag names The degree of matching with the set is calculated, and the narrowing speech recognition means extracts the second data set formed by the attribute information associated with all the tag names as a recognition result using as a speech recognition filter. Preferably, the guide data are narrowed down by logical product of the attribute information.

Further, in the search support server according to the present invention, it is preferable that the narrowed-down speech recognition means updates the second data set with the attribute information of the narrowed-down guide data.

Next, in order to achieve the above object, a search support method according to the present invention is a search support method that can be executed by a search support server that supports searches for content published on a website, comprising: A search support server provides a signage window for selectively displaying guide data for guiding an utterance for selecting tag information indicating a specific part of a content, and a user uttering the guide data to be displayed from among the guide data. a step of displaying a speech window that can be narrowed down by a terminal device connected to be capable of data communication; receiving and narrowing down the tag information to be displayed, and receiving a selection of the narrowed-down tag information, thereby causing the terminal device to display a specific portion of the content corresponding to the selected tag information. characterized by executing

Further, in the search support method according to the present invention, the tag information includes at least a tag name identifying a specific part of the content, text data consisting of a simple sentence or a group of words describing the contents of the tag name, and attributes of the tag name. is preferably configured with attribute information indicating

Further, in the search support method according to the present invention, when the search support server extracts a simple sentence or word group indicating the content based on the content of the content and accepts selection of the extracted simple sentence or word group receiving an input of the tag name and a simple sentence or word group indicating the content of the tag name based on the displayed content; and based on the displayed content a step of searching for pointer information indicating a specific portion of said content; a step of receiving allocation of said searched pointer information for each of said tag name whose input is received; and a step of searching for each of said tag name whose input is received. receiving the input of the attribute information that can be spoken at times, and generating search data serving as a basis for the guide data to be displayed in the signage window and/or the speech window.

Further, in the search support method according to the present invention, the search support server scrutinizes the content, extracts the included character string as a simple sentence or word group, and outputs it as text data; and displaying the content corresponding to the selected text data when the selection of the text data is accepted.

Further, in the search support method according to the present invention, it is preferable that the guide data is generated by associating the tag information included in the search data with search data identification information for identifying the search data.

Further, in the search support method according to the present invention, in order to select the tag name displayed in the speech window and capable of accepting selection by the user, the search support server uses voice data uttered by the user. performing a step of accepting an input, wherein the search support server converts the received voice data into text data, and a simple sentence or word indicating the tag name of the guide data in the converted text data and the content of the tag name calculating the degree of matching with respect to the group, identifying the tag name with the highest calculated degree of matching, and displaying a specific portion of the content corresponding to the tag name when the degree of matching is greater than a predetermined value; and extracting the tag information that matches the attribute information from among the tag information of the guide data based on the received voice data, and narrowing down the guide data by logical AND of one or more pieces of attribute information that match. It is preferable that the step of outputting the tag name and the attribute information of the narrowed down guide data is executed in parallel.

Further, in the search support method according to the present invention, the search support server provides a first data set formed of converted text data, simple sentences or word groups indicating all the tag names and the contents of the tag names. and the logic of the attribute information extracted as a recognition result using a second data set formed of the attribute information associated with all tag names as a speech recognition filter and narrowing down the guide data by product.

Further, in the search support method according to the present invention, it is preferable that the search support server updates the second data set with the attribute information of the narrowed down guide data.

Next, in order to achieve the above object, a computer program according to the present invention is a computer program that can be executed by a search support server that supports a search for content published on a website, comprising: A signage window for selectively displaying guide data for guiding a server to select tag information indicating a specific part of contents, and narrowing down the guide data to be displayed from among the guide data by the user's utterance. means for displaying a speech window capable of data communication on a terminal device connected to be capable of data communication; and narrowing down the tag information to be displayed, and receiving selection of the narrowed-down tag information, thereby functioning as a means for displaying on the terminal device a specific part of the content corresponding to the selected tag information. characterized by

Further, in the computer program according to the present invention, the tag information includes at least a tag name that identifies a specific part of the content, text data consisting of a simple sentence or a group of words describing the contents of the tag name, and attributes of the tag name. It is preferable that the attribute information is composed of the attribute information shown.

Further, the computer program according to the present invention causes the search support server to function as search data generating means for generating search data serving as a basis for guide data displayed in the signage window and/or the speech window, and extracting and displaying means for extracting a simple sentence or word group indicating content based on the content of the content, and displaying the content corresponding to the selection of the extracted simple sentence or word group; tag input receiving means for receiving an input of the tag name and a simple sentence or word group indicating the content of the tag name based on the displayed content; and a pointer indicating a specific portion of the content based on the displayed content. Pointer searching means for searching for information, pointer allocation receiving means for receiving allocation of the searched pointer information for each of the tag names that have received inputs, and said utterable at the time of searching for each of the tag names that have received inputs. It is preferable to function as attribute information receiving means for receiving input of attribute information.

Further, the computer program according to the present invention further comprises content transcription means for scrutinizing the content, extracting the contained character strings as simple sentences or word groups, and outputting them as text data; And, when receiving the selection of the output text data, it is preferable to function as content display means for displaying the content corresponding to the selected text data.

Further, in the computer program according to the present invention, it is preferable that the guide data is generated by associating the tag information included in the search data with search data identification information for identifying the search data.

Further, the computer program according to the present invention causes the search support server to input voice data uttered by the user in order to narrow down the tag names displayed in the speech window and capable of accepting selection by the user. Functioning as an utterance accepting means for accepting input, the search support server converts the input accepted voice data into text data, and a simple sentence or word indicating the tag name of the guide data of the converted text data and the content of the tag name Meta tag voice for calculating the degree of matching with respect to the group, identifying the tag name with the highest degree of matching calculated, and displaying a specific portion of the content corresponding to the tag name when the degree of matching is greater than a predetermined value. extracting tag information that matches the attribute information from among the tag information of the guide data based on the recognition means and the voice data that has been input; and outputting the tag name and attribute information of the guide data narrowed down.

Further, the computer program according to the present invention provides a first data set formed of converted text data, simple sentences or word groups indicating all of the tag names and the content of the tag names. The second data set formed by the attribute information associated with all the tag names is functioned as a means for calculating the degree of matching, and the second data set formed by the attribute information associated with all tag names is used as a speech recognition result. It is preferable to function as means for narrowing down the guide data by logical product of the attribute information extracted as .

Also, the computer program according to the present invention preferably causes the narrowed-down speech recognition means to function as means for updating the second data set with the attribute information of the narrowed-down guide data.

According to the present invention, a content provider who provides content among a large number of content published on a website displays guide data on a speech window or a signage window to easily guide a user to his/her own content. In addition, since the guide data to be selected can be narrowed down by the user's own intention, the content provider can be easily and quickly provided by each user speaking or keying in according to the displayed guide data. It is possible to display the content provided by according to the user's intention.

1 is a block diagram schematically showing the configuration of a voice search system according to an embodiment of the present invention; FIG. 1 is a block diagram schematically showing the configuration of a search support server according to an embodiment of the present invention; FIG. 1 is a block diagram schematically showing the configuration of a terminal device according to an embodiment of the present invention; FIG. 4 is a functional block diagram of search data generation processing of the search support server according to the embodiment of the present invention; FIG. FIG. 4 is an exemplary diagram of an input acceptance screen displayed on a terminal device used by a content provider by the search support server according to the embodiment of the present invention; FIG. 4 is an exemplary diagram of search data and guide data generated by the search support server according to the embodiment of the present invention; FIG. 4 is an exemplary diagram of search data and guide data of the search support server according to the embodiment of the present invention; FIG. 8 is another exemplary diagram of search data and guide data of the search support server according to the embodiment of the present invention; FIG. 4 is an exemplary diagram of a speech window displayed on a terminal device used by a user by the search support server according to the embodiment of the present invention; 9 is a flow chart showing a content provider setting processing procedure of the CPU of the search support server according to the embodiment of the present invention; FIG. 4 is a functional block diagram of voice recognition processing of an utterance by a user of the search support server according to the embodiment of the present invention; 4 is a flow chart showing the procedure of speech recognition processing of the CPU of the search support server according to the embodiment of the present invention;

A search support server according to an embodiment of the present invention will be described below with reference to the drawings. The following embodiments do not limit the invention described in the claims, and all combinations of characteristic items described in the embodiments are essential items for the solution. It goes without saying that there is no limit.

Moreover, the present invention can be implemented in many different modes and should not be construed as being limited to the description of the embodiments. The same reference numerals are given to the same elements throughout the embodiments.

In the following embodiments, a voice search system in which a computer program is installed in a computer system will be described. can be implemented. Therefore, the present invention can provide a signage window that selectively displays guide data that guides utterances for selecting tag information indicating a specific portion of content, and can display all guide data by narrowing down the information by utterances. By using the speech window, as long as the user speaks or selects according to the guide data narrowed down according to the user's purpose, it is a search support that can surely guide the content provider to the specific part of the content that the content provider wants to provide to the user. It can be implemented as a hardware called server, implemented as software, or implemented as a combination of software and hardware. The computer program can be recorded on any computer-readable recording medium such as a hard disk, DVD, CD, optical storage device, or magnetic storage device.

According to the embodiment of the present invention, a content provider who provides content from among a large number of content published on a website can set guide data to easily guide users to his or her own content in a speech window or signage. Since it is possible to display it in a window and to narrow down the guide data to be selected at the user's own will, individual users can speak or key-in according to the displayed guide data, which is easy and quick. It is possible to display the content provided by the content provider in accordance with the user's intention.

FIG. 1 is a block diagram schematically showing the configuration of a voice search system according to an embodiment of the invention. The voice search system according to the present embodiment can perform data communication with a terminal device 1a used by a content provider, a terminal device 1b used by a user searching for content or video content, and the terminal devices 1a and 1b. and a search support server 3 connected via a network 2 such as the Internet. The terminal devices 1a and 1b are not limited to stationary PCs to which microphones and speakers are connected, and may be portable terminals such as smart phones and tablets that have built-in microphones and speakers.

FIG. 2 is a block diagram schematically showing the configuration of the search support server 3 according to the embodiment of the invention. The search support server 3 according to this embodiment includes at least a CPU (Central Processing Unit) 31, a memory 32, a storage device 33, an I/O interface 34, a video interface 35, a portable memory drive 36, a communication interface 37, and the above-described It consists of an internal bus 38 that connects hardware.

The CPU 31 is connected to the above-described hardware units of the search support server 3 via the internal bus 38, controls the operation of the above-described hardware units, and executes the computer program 100 stored in the storage device 33. perform various software-like functions according to The memory 32 is composed of a volatile memory such as SRAM, SDRAM, etc. A load module is expanded when the computer program 100 is executed, and temporary data generated when the computer program 100 is executed is stored.

The storage device 33 is composed of a built-in fixed storage device (hard disk), ROM, and the like. The computer program 100 stored in the storage device 33 is downloaded by the portable memory drive 36 from a portable recording medium 90 such as a DVD, CD-ROM, USB memory, SD card, etc. that records information such as programs and data, When executed, it is expanded from the storage device 33 to the memory 32 and executed. Of course, it may be a computer program downloaded from an external computer connected via the communication interface 37 .

The storage device 33 has a search data storage section 331 and a guide data storage section 332 . The search data storage unit 331 stores search data (tag names that identify specific parts of the content, simple sentences or word groups that describe the content of the tag names) that can be accessed by users for content provided by content providers. attribute information indicating attributes of data and tag names) is stored in association with pointer information (and time stamp information in the case of video content). Note that the pointer information broadly means information indicating the location of content that can be accessed. If the content is a web page, the URL corresponds to the pointer information, and if the content is video content, not only the reproducible URL, but also time stamp information to start playing the video content and time stamp to end the playing. information is also included in the pointer information.

The guide data storage unit 332 stores guide data generated based on the search data in order to prompt the user to make an utterance for searching or a selection operation such as a click or a touch. The guide data extracts from the search data a tag name that identifies a specific part of the content, text data consisting of a simple sentence or a group of words describing the content of the tag name, and attribute information that indicates the attribute of the tag name. It is generated in association with the identification information of the search data to be searched. The guide data to be displayed is narrowed down by the user's utterance while the list is displayed in the speech window displaying the guide data. When the user selects one of the tag names from the guide data that has been narrowed down and displayed, the content (including video content) specified by the search data associated with the selected tag name is displayed ( or playback).

The communication interface 37 is connected to an internal bus 38, and by connecting to an external network 2 such as the Internet, LAN, WAN, etc., it is possible to transmit and receive data to and from an external computer or the like.

The I/O interface 34 is connected to a keyboard 41 and a mouse 42, which are input devices, to input data. In the present embodiment, it is the terminal device 1a, 1b (smartphone, tablet, etc.) used by the content provider or the user that actually inputs the voice, and the input voice data is transmitted via the communication interface 37. to receive. Of course, the search support server 3 may be directly connected with a microphone, a speaker, or the like.

The video interface 35 is connected to a display device 43 such as a CRT display or liquid crystal display. In the present embodiment, it is the terminal devices 1a and 1b (smartphones, tablets, etc.) used by the content provider or the user that actually output and display the images. 1b via the communication interface 37, such as image data (including audio data).

FIG. 3 is a block diagram schematically showing the configuration of the terminal device 1 (common to 1a and 1b) according to the embodiment of the present invention. The terminal device 1 according to the present embodiment includes at least a CPU (Central Processing Unit) 11, a memory 12, a storage device 13, an I/O interface 14, a video interface 15, a portable memory drive 16, a communication interface 17, and the hardware described above. It consists of an internal bus 18 that connects hardware.

The CPU 11 is connected to the above-described hardware units of the terminal device 1 via the internal bus 18 , controls the operation of the above-described hardware units, and executes computer programs 101 stored in the storage device 13 . , which performs various software-like functions. The memory 12 is composed of a volatile memory such as SRAM, SDRAM, etc. A load module is expanded when the computer program 101 is executed, and temporary data generated when the computer program 101 is executed is stored.

The storage device 13 is composed of a built-in fixed storage device (hard disk), ROM, and the like. A computer program 101 stored in the storage device 13 is downloaded from an external computer connected via the communication interface 17, and expanded from the storage device 13 to the memory 12 at the time of execution. Of course, it may be a computer program downloaded via the portable memory drive 16 from a portable recording medium 91 such as an SD card recording information such as programs and data.

The communication interface 17 is connected to an internal bus 18, and by connecting to an external network 2 such as the Internet, LAN, WAN, etc., it is possible to transmit and receive data to and from an external computer or the like.

The I/O interface 14 is connected to an input device such as the keyboard 203 and the mouse 204, as well as an audio input device such as the microphone 201 and an audio output device such as the speaker 202, and inputs and outputs data. A smart phone or the like may be connected via the communication interface 17 to substitute the voice input device and the voice output device.

The video interface 15 is connected to the display device 205 and displays images for input/output transmitted from the search support server 3 on a browser or the like. The retrieved content and moving image content may be displayed on the display device 205, or may be displayed on an external computer that is separately connected to enable data communication via a network.

The operation of the search support server 3 configured as described above will be described below.

FIG. 4 is a functional block diagram of search data generation processing of the search support server 3 according to the embodiment of the present invention. FIG. 4 describes a procedure in which a content provider generates search data, generates guide data, and displays them on a signage window and/or a speech window.

In FIG. 4, a search data generation unit 401 generates search data that serves as a basis for guide data displayed in a signage window and/or a speech window. The search data generation unit 401 includes an extraction/display unit 402 , a tag input reception unit 403 , a pointer search unit 404 , a pointer allocation reception unit 405 and an attribute information reception unit 406 .

The extraction/display unit 402 extracts a simple sentence or word group indicating the content based on the content, and displays the corresponding content when the selected simple sentence or word group is selected. More specifically, the extraction/display section 402 includes a content transcription section 4021 and a content display section 4022 .

The content transcription unit 4021 carefully examines the target content, extracts the contained character strings as simple sentences or word groups, and outputs them as text data. The output text data can be used as a reference for generating tag names based on character strings contained in the website. A website URL is associated with each extracted simple sentence or word group.

When the content is video content, a voice transcription section (not shown) is provided to extract the audio portion of the video content and output it as text data associated with appropriate time stamp information for each phrase. The output text data can be used as a reference for generating tag names based on the audio included in the video content.

The content display unit 4022 displays a specific part of the content corresponding to the selected text data when the selection of the output text data is accepted. When the selection of the tag name candidate is received, it is possible to confirm whether or not the correct content is associated.

Input of tag information associated with a specific part of the content is then accepted. In this embodiment, the tag information is composed of at least a tag name that identifies a specific part of the content, text data consisting of a simple sentence or a group of words describing the content of the tag name, and attribute information that indicates the attribute of the tag name. there is

The tag input reception unit 403 receives the input of a tag name and a simple sentence or word group indicating the content of the tag name based on the displayed text data or, in the case of video content, a specific portion of the video content being reproduced. accept.

A pointer search unit 404 searches for pointer information indicating a specific portion of the content based on the specific portion of the displayed content. The searched pointer information outputs as a result the already associated URL when the content is a website. When the content is video content, not only the already associated URL but also time stamp information regarding the reproduction start timing and the reproduction end timing are searched.

The pointer allocation reception unit 405 receives allocation of searched pointer information for each tag name whose input is received. Accordingly, by receiving the selection of the tag name, it is possible to display the specific portion of the content according to the allocated pointer information.

The attribute information reception unit 406 receives input of attribute information that can be spoken at the time of searching for each tag name whose input is received. The tag names can be narrowed down by the user uttering the attribute information as will be described later.

Here, the attribute information is information useful for narrowing down the content, such as "new", "video", "how to use", "price", "medical", etc., and if it is words that are easy to speak, it is particularly limited. not to be

In this way, input of a tag name, a simple sentence or a group of words indicating the content of the tag name, and attribute information is received, and search data is generated in association with pointer information as tag information. The guide data generation unit 407 generates guide data by associating the tag information included in the generated search data with the search data identification information for identifying the search data.

FIG. 5 is an exemplary diagram of an input reception screen displayed on the terminal device 1a used by the content provider by the search support server 3 according to the embodiment of the present invention. FIG. 5(a) is an exemplary diagram of the initial screen of the input reception screen, FIG. 5(b) is an exemplary diagram of the input reception screen of the tag information, and FIG. 5(c) is a display of the signage window. It is an illustration figure.

An initial screen 50 shown in FIG. 5(a) is popped up on the terminal device 1a used by the content provider along with the top page of the website. The text data output by the content transcription unit 4021 is displayed on the common window 60 of the initial screen 50. FIG.

Based on the text data displayed in shared window 60, the content provider searches for a Web page containing information that he or she wishes to provide to the user, and selects the text data displayed in shared window 60. , to the web page associated with the selected text data. The content provider selects the "input tag" button 51 on the initial screen 50 displayed as a popup.

When the selection of the "input tag" button 51 by the content provider is accepted, a tag information input acceptance screen 52 shown in FIG. 5(b) is displayed in another window. The content provider inputs a tag name that is easy for the user to select in the tag name input area 53 and a simple sentence or word group indicating the content in the meta tag input area 54 . The content provider selects attribute information using the attribute information selection button 55 . Of course, an area for optionally inputting attribute information may be provided for key-in.

When the content provider selects the "mark" button 56, the search support server 3 searches for the URL of the website displayed on the browser of the terminal device 1a as pointer information, and associates it with the input tag name. are stored in the search data storage unit 331 as search data.

Guide data is generated by associating identification information for identifying search data with the tag information portion of search data. FIG. 6 is an exemplary diagram of search data and guide data generated by the search support server 3 according to the embodiment of the present invention. FIG. 6(a) is an exemplary diagram of generated guide data, and FIG. 6(b) is an exemplary diagram of search data that is the basis of the generated guide data.

As shown in FIG. 6(a), the guide data are associated with a "number" indicating the order of the tag name of the search data, and common search data that is identification information for identifying a website. ID (search data ID=YYYYYY in FIG. 6). The generated guide data is stored in the guide data storage unit 332 .

Only the search data is stored as "pointer information" in association with the URL and the tag name. Therefore, by selecting the "guide data" button in FIG. 5(a), the tag information of the search data is copied, and the search data ID (YYYYYY) and the tag name indicating the number of the tag in the search data are displayed. The guide data is generated by associating with the "number". The fact that the "search data ID" column of the guide data shown in FIG. 6(a) is blank means that the guide data is generated based on only one search data, and is generated based on other search data. If the guide data is the guide data that has been written, the search data ID of the other search data is described.

The signage window 58 of FIG. 5(c) preferably displays the tag name of the guide data selected by the content provider as the data that the content provider wants to display to the user from among the generated guide data. In this case, when the selection of the "Guide Data" button is accepted, an "Edit" button and an "Expand" button (not shown) are displayed. When the selection of the “edit” button is accepted, the tag name displayed on the signage window 58 can be set by keying in the common window 60 . When the selection of the "expand" button is accepted, guide data generated from other search data can be added.

That is, in the present embodiment, search data may be generated for each content provider, or search data may be generated for each of multiple websites launched by the same content provider. For example, even if it is a website of the same company, there is a rapid increase in the number of opportunities to set up a website for each business division or product/service, and the amount of content is increasing day by day. Being able to find this content quickly and reliably is also important for customer service.

In the present embodiment, all guide data are generated based on basic search data. Therefore, if only guide data generated based on other search data is added to one guide data, content (including video content) associated with other search data, that is, content of another website can be obtained. can be quickly and reliably displayed and reproduced.

FIG. 7 is an exemplary diagram of search data and guide data of the search support server 3 according to the embodiment of the present invention. FIG. 7(a) is an exemplary diagram of guide data based on one search data, and FIG. 7(b) is an exemplary diagram of other search data that is the basis of added guide data.

The difference from the search data and guide data shown in FIG. 6 is that guide data with different search data IDs are included instead of common search data IDs. That is, when the search data and the guide data are generated, the search data ID is common. In this embodiment, the two-stage structure of search data and guide data makes it possible to set guide data with different search data IDs. As a result, it is possible to read the search data that is the basis of the guide data with different search data IDs, and display the search data with different search data IDs, that is, the specific parts of the content provided on different websites. .

For example, let the common search data ID be "YYYYYY". Guide data based on other search data IDs can be added at any time to the guide data generated based on the common search data ID "YYYYYY". In the example of FIG. 7A, guide data with a search data ID of "PPPPPP" is added. That is, it becomes possible to read the search data that is the basis of the guide data whose search data ID is "PPPPPP", and to display the content associated with the search data.

That is, it is possible to refer to the search data whose search data ID is "PPPPPP" shown in FIG. 7B, and to obtain the pointer information of the desired content. Therefore, by simply adding guide data, desired content can be displayed regardless of the content of any website. In FIG. 7B, blanks in the "search data ID" column of the guide data indicate that the guide data is not the added guide data but the guide data whose search data ID is the common search data ID "YYYYYY". means.

FIG. 8 is another exemplary diagram of search data and guide data of the search support server 3 according to the embodiment of the present invention. FIG. 8(a) is an exemplary diagram of guide data based on one piece of search data, and FIG. 8(b) is an exemplary diagram of other search data that is the basis of added guide data.

The difference between the search data and the guide data shown in FIG. 7 is that the pointer information of the search data includes not only the URL of the video content, but also the start time stamp indicating the playback start point and the end time stamp (time stamp) indicating the playback end point of the video content. stamp information). That is, in moving image content, only scenes that the content provider wants to show to the user can be reproduced.

For example, let the common search data ID be "YYYYYY". In addition, to the guide data generated based on the search data whose search data ID is "YYYYYY", guide data of moving image content based on other search data IDs can be added at any time. In FIG. 8A, guide data whose search data ID is "TTTTTT" is added. That is, it becomes possible to read out the search data that is the basis of the guide data whose search data ID is "TTTTTT", and to reproduce the video content associated with the search data.

That is, it is possible to refer to the search data whose search data ID is "TTTTTT" shown in FIG. Therefore, by simply adding guide data, only desired scenes of desired moving image content can be reproduced regardless of the moving image content of any website. Note that in FIG. 8B, blanks in the "search data ID" column of the guide data mean that the guide data is not added guide data but guide data whose search data ID is "YYYYYY".

Returning to FIG. 4, the signage window display unit 408 causes the content provider to select the tag name and the like of the generated guide data and display it on the signage window on the terminal device 1b used by the user. As a result, the user can be guided to utter an utterance for selecting the tag information indicating the specific portion of the content. The signage window is preferably displayed by the search support server on the terminal device 1a used by the content provider in such a manner that a character string as shown in FIG. 5(c) is displayed.

Specifically, the tag name of the selected guide data is displayed in the guidance display area 58 of the signage window 57 shown in FIG. 5(c). In FIG. 5(c), an arrow means that the character string is displayed while flowing in that direction. A user who sees the display of the signage window can know what kind of information the content provider intends to provide, and can devise the content of the utterance. The content provider can guide the user to the content to be provided by displaying the tag name or character string indicating the content to be provided to the user on the signage window 57 .

Also, the speech window display unit 409 displays all tag names of the generated guide data in a speech window on the terminal device 1b used by the user. The tag names of the displayed guide data can be narrowed down by the user's speech. By narrowing down, the tag names can be displayed in the speech window without operations such as scrolling. can be displayed.

FIG. 9 is an exemplary diagram of a speech window displayed on the terminal device 1b used by the user by the search support server 3 according to the embodiment of the present invention. FIG. 9(a) is an exemplary diagram of a speech window displayed on the terminal device 1b used by the user by the search support server 3 according to the present embodiment, and FIG. 9(b) is a speech window according to the present embodiment. 3 is an exemplary diagram of a terminal device 1b on which a search support server 3 displays a specific portion of content; FIG. As shown in FIG. 9A, all tag names of generated guide data are displayed in a guide data display area 902 in a selectable state on a speech window 901 .

Even if all the tag names cannot be displayed in the guide data display area 902, the tag names of all the guide data can be browsed by operating the scroll button 903. FIG. Selection of one tag name from tag names of the guide data displayed in the guide data display area 902 is accepted. As a result, it is possible to specify the search data that is the basis of the guide data of the tag name whose selection has been accepted, so that the specified portion of the content associated with the search data can be displayed.

In the example of FIG. 9, when the selection of the tag name displayed in the speech window 901 is accepted, the specific part of the corresponding content is displayed on the terminal device 1b so that it can be confirmed. The terminal device 1b used by the user may be a desktop type as shown in FIG. 9, or may be a mobile terminal such as a smartphone or a tablet.

FIG. 10 is a flow chart showing the content provider setting processing procedure of the CPU 31 of the search support server 3 according to the embodiment of the present invention. In FIG. 10, the CPU 31 of the search support server 3 extracts a simple sentence or word group indicating the content based on the content (step S1001), and accepts selection of the extracted simple sentence or word group. The content to be displayed is displayed (step S1002).

The CPU 31 accepts input of a tag name and a simple sentence or word group indicating the contents of the tag name based on the displayed text data or, in the case of moving image content, a specific portion of the moving image content being reproduced (step S1003). ).

Based on the displayed specific portion of the content, CPU 31 searches for pointer information indicating the specific portion of the content (step S1004). The CPU 31 accepts allocation of searched pointer information for each tag name whose input is accepted (step S1005). Accordingly, by receiving the selection of the tag name, it is possible to display the specific portion of the content according to the allocated pointer information.

The CPU 31 receives input of attribute information that can be spoken at the time of searching for each tag name whose input is received (step S1006). The tag names can be narrowed down by the user uttering the attribute information as will be described later.

The CPU 31 associates the input tag name, simple sentence or word group indicating the content of the tag name, and attribute information with the pointer information to generate search data (step S1007). CPU 31 associates the tag information included in the generated search data with the search data identification information for identifying the search data to generate guide data (step S1008).

The CPU 31 causes the content provider to select the tag name and the like of the generated guide data and displays it on the signage window on the terminal device 1b used by the user (step S1009), and displays the tag name of the generated guide data. All are displayed on the speech window on the terminal device 1b used by the user (step S1010).

A procedure for a user to execute search processing will be described below. This embodiment is characterized in that the content provider displays a signage window and a speech window for guiding the user's search. However, in addition to this, as a device for quickly and reliably searching and displaying desired content for the user, the user can speak the tag name of the displayed guide data to narrow down the search system, which is different from the conventional search system. This is a big difference.

FIG. 11 is a functional block diagram of speech recognition processing of an utterance by the user of the search support server 3 according to the embodiment of the present invention. In FIG. 11, the terminal device 1b used by the user accepts the user's utterance as an input in a state in which the signage window shown in FIG. 5(c) and the speech window shown in FIG. 9(a) are displayed. I will explain the procedure to do.

As shown in FIG. 11, an utterance accepting unit 1101 accepts an input of voice data uttered by a user. Specifically, the voice data is obtained by receiving the voice data uttered by the user in the terminal device 1b.

The preprocessing unit 1102 performs noise removal, speech period detection, and the like on the received voice data. The preprocessed speech data is passed to the metatag speech recognition unit 1103 and the narrowed speech recognition unit 1108, and the metatag speech recognition unit 1103 and the narrowed speech recognition unit 1108 are executed in parallel.

The metatag speech recognition unit 1103 recognizes the preprocessed speech data and converts it into text data, and based on the converted text data, matches the tag name of the guide data and a simple sentence or word group indicating the content of the tag name. The degree of matching is calculated, the tag name with the highest calculated degree of matching is identified, and when the degree of matching is greater than a predetermined value, a specific portion of the content corresponding to the tag name is displayed. In other words, when the metatag speech recognition unit 1103 correctly recognizes the speech, the recognition result can be immediately output without executing extra processing such as integration with other speech recognition processing.

That is, the metatag speech recognition unit 1103 includes a text data conversion unit 1104 , a match calculation unit 1105 , a tag name specification unit 1106 and a content display/playback unit 1107 . A text data conversion unit 1104 recognizes the preprocessed voice data based on a so-called dictation grammar and converts it into text data.

The degree-of-match calculation unit 1105 inquires the first data set formed of simple sentences or word groups indicating the contents of all tag names and tag names, and calculates the degree of match with the recognition result of the received voice data. calculate.

The tag name identification unit 1106 identifies the tag name with the highest calculated degree of matching. This is because the tag name with the highest degree of matching is the most probable recognition result. However, if the degree of matching is equal to or less than a predetermined threshold, there is a high possibility of erroneous recognition.

Content display/playback unit 1107 determines whether or not the calculated degree of matching is greater than a predetermined threshold. If it is determined that the value is equal to or less than the predetermined threshold value, it is determined that the recognition accuracy is insufficient, and the system waits for another utterance. When it is determined that the value is larger than the predetermined threshold, it is possible to display or reproduce the specific portion of the content associated with the specified tag name and determine whether or not the recognition result is correct.

Based on the preprocessed speech data, the narrowed-down speech recognition unit 1108 uses a so-called rule grammar to extract tag information of the guide data that matches the attribute information, and extracts one or more pieces of attribute information that match. The guide data is narrowed down by the logical product of , and the tag name and attribute information of the narrowed down guide data are output. As a result, even if the metatag speech recognition unit 1103 cannot obtain an appropriate recognition result, it is possible to further narrow down what the user will say next, so that the success rate of speech recognition can be increased. As a result, correct recognition results can be obtained in a short time. Specifically, the narrowed-down speech recognition unit 1108 includes an attribute extraction unit 1109 and a narrowing-down unit 1110 .

The attribute extraction unit 1109 extracts attribute information as a recognition result using the second data set formed of attribute information associated with all tag information as a speech recognition filter. The narrowing down unit 1110 narrows down the tag name (guide data) as a logical product of the extracted attribute information.

The narrowed-down speech recognition unit 1108 preferably includes an updating unit 1111 that updates the second data set, which is the speech recognition filter, according to the attribute information of the narrowed-down guide data. Since the second data set is updated each time the guide data is narrowed down by the attribute information, the user's next utterance is further restricted, and the degree of certainty that correct recognition results are obtained can be increased, and the recognition result can be obtained more quickly. A correct recognition result can be obtained.

FIG. 12 is a flow chart showing the procedure of voice recognition processing of the CPU 31 of the search support server 3 according to the embodiment of the present invention. The CPU 31 of the search support server 3 receives input of voice data uttered by the user (step S1201). Specifically, the voice data is obtained by receiving the voice data uttered by the user in the terminal device 1b.

CPU 31 performs noise removal, detection of a speech period, and the like on the received voice data (step S1202). Based on the preprocessed audio data, the following two processes (metatag speech recognition and refined speech recognition) are performed in parallel.

First, the CPU 31 recognizes the preprocessed voice data based on the so-called dictation grammar and converts it into text data (step S1203). The CPU 31 refers to all tag names and the first data set formed of simple sentences or word groups indicating the contents of the tag names, and calculates the degree of matching with the received voice data (step S1204). The CPU 31 identifies the tag name with the highest calculated degree of matching (step S1205).

CPU 31 determines whether or not the calculated degree of matching is greater than a predetermined threshold (step S1206). When the CPU 31 determines that it is equal to or less than the predetermined threshold (step S1206: NO), the CPU 31 returns the process to step S1201 and waits for another utterance.

When the CPU 31 determines that the value is greater than the predetermined threshold (step S1206: YES), the CPU 31 displays or reproduces the specified portion of the content associated with the specified tag name (step S1207).

On the other hand, based on the preprocessed voice data, the CPU 31 extracts the attribute information of the recognition result using the second data set formed of attribute information associated with all tag names as a voice recognition filter. (step S1208). The CPU 31 narrows down the tag name (guide data) as a logical product of the extracted attribute information (step S1209). The CPU 31 updates the second data set, which is the voice recognition filter, with the narrowed down tag name and attribute information of the guide data (step S1210), returns the process to step S1201, and waits for another utterance. .

As described above, according to the present embodiment, a content provider who provides content from among a large number of content published on a website can display guide data that easily guides the user to his/her own content in the speech window. or signage window, and the user can narrow down the guide data to be selected at his or her own will. Moreover, it is possible to reliably display the content provided by the content provider to the user.

Further, according to the present embodiment, by executing the metatag speech recognition process and the narrowing speech recognition process in parallel, if the metatag speech recognition is correct, the recognition result can be output as it is. If the certainty of the result is insufficient, tag names based on more refined guide data can be displayed in the speech window. Therefore, the user can display or reproduce desired content in a relatively short period of time.

The present invention is not limited to the above-described embodiments, and various modifications and improvements are possible within the scope of the present invention. For example, the method of inputting tag information including attribute information is not limited to the method described above, and is not particularly limited as long as it is possible to input a character string that can narrow down the tag name.

Further, when recognizing an utterance by a user, the recognition processing that executes processing in parallel with the metatag speech recognition unit 1103 is not limited to the narrowed-down speech recognition unit 1108, and may limit the next utterance by the user. It is not particularly limited as long as it can be processed.

Also, the dictation grammar and rule grammar used for speech recognition are desirably generated by learning text data, guide data, and externally acquired text data as teacher data. As a learning method, AI such as so-called machine learning or deep learning may be used, or a conventional method such as expanding a correspondence table may be used, and is not particularly limited.

Reference Signs List 1, 1a, 1b terminal device 2 network network 3 search support server 11, 31 CPU
12, 32 memory 13, 33 storage device 14, 34 I/O interface 15, 35 video interface 16, 36 portable disk drive 17, 37 communication interface 18, 38 internal bus 90, 91 storage medium 100, 101 computer program 331 search Data storage unit 332 Guide data storage unit

Claims (10)

A search support server that supports searches for content published on a website,
a signage window selectively displaying guide data for guiding speech for selecting tag information indicating a specific portion of content;
causing a terminal device connected to be capable of data communication to display a speech window capable of narrowing down guide data to be displayed from among the guide data by user's utterance,
Accepting the selection by accepting the user's speech of the tag information displayed in the signage window and/or the speech window, narrowing down the displayed tag information, and accepting the selection of the narrowed down tag information. A search support server that causes the terminal device to display a specific portion of the content corresponding to the tag information. The tag information is composed of at least a tag name identifying a specific part of the content, text data consisting of a simple sentence or a group of words describing the content of the tag name, and attribute information indicating attributes of the tag name. 2. The search support server according to claim 1. Search data generation means for generating search data that is the basis of guide data to be displayed in the signage window and/or the speech window;
The search data generation means is
Extraction/display means for extracting a simple sentence or word group indicating the content based on the content of the content and displaying the corresponding content when selection of the extracted simple sentence or word group is accepted;
tag input receiving means for receiving an input of the tag name and a simple sentence or word group indicating the content of the tag name based on the displayed content;
pointer searching means for searching for pointer information indicating a specific portion of the content based on the displayed content;
Pointer allocation receiving means for receiving allocation of the searched pointer information for each tag name whose input is received;
3. The search support server according to claim 2, further comprising attribute information receiving means for receiving an input of said attribute information that can be spoken at the time of searching for each said tag name whose input has been received. The extraction/display means is
a content transcription means for scrutinizing the content, extracting contained character strings as simple sentences or word groups, and outputting them as text data;
4. The search support server according to claim 3, further comprising content display means for, when receiving a selection of output text data, displaying the content corresponding to the selected text data. 5. The search support server according to claim 3, wherein the guide data is generated by associating the tag information included in the search data with search data identification information for identifying the search data. To narrow down the tag names that are displayed in the speech window and that can be selected by a user,
an utterance receiving means for receiving an input of voice data uttered by a user;
converts voice data received as input into text data, calculates the degree of matching of the converted text data with the tag name of the guide data and a simple sentence or word group indicating the content of the tag name, and the calculated degree of matching is metatag speech recognition means for identifying a maximum tag name and displaying a specific portion of the content corresponding to the tag name when the degree of matching is greater than a predetermined value;
Based on the received voice data, the tag information of the guide data that matches the attribute information is extracted, and the guide data is narrowed down by logical product of one or more pieces of attribute information that match. and narrowing-down speech recognition means for outputting the tag name and the attribute information of the narrowed-down guide data, wherein the meta-tag speech recognition means and the narrowing-down speech recognition means are executed in parallel. 6. The search support server according to any one of 3 to 5. The metatag speech recognition means calculates the degree of matching between the converted text data and a first data set formed of simple sentences or word groups indicating all the tag names and the contents of the tag names,
The narrowed-down speech recognition means uses a logical product of the attribute information extracted as a recognition result using a second data set formed of the attribute information associated with all tag names as a speech recognition filter to perform the 7. The search support server according to claim 6, wherein the guide data is narrowed down. 8. The search support server according to claim 7, wherein said narrowed-down speech recognition means updates said second data set with said attribute information of the narrowed-down guide data. A search support method that can be executed by a search support server that supports searches for content published on a website, comprising:
The search support server
a signage window selectively displaying guide data for guiding speech for selecting tag information indicating a specific portion of content;
a step of displaying a speech window capable of narrowing down the guide data to be displayed from among the guide data by user's utterance on a terminal device connected so as to be capable of data communication;
Accepting the selection by accepting the user's speech of the tag information displayed in the signage window and/or the speech window, narrowing down the displayed tag information, and accepting the selection of the narrowed down tag information. and causing the terminal device to display a specific portion of the content corresponding to the tag information. A computer program that can be executed by a search support server that supports searches for content published on a website,
the search support server,
a signage window selectively displaying guide data for guiding speech for selecting tag information indicating a specific portion of content;
a speech window capable of narrowing down the guide data to be displayed from the guide data by user's utterance;
means for displaying on a terminal device connected for data communication, and
Accepting the selection by accepting the user's speech of the tag information displayed in the signage window and/or the speech window, narrowing down the displayed tag information, and accepting the selection of the narrowed down tag information. means for displaying a specific portion of the content corresponding to the tag information on the terminal device;
A computer program characterized by functioning as

Images