JP2020177360A - Q&a extraction device, method, program, and answering system - Google Patents

Abstract

To provide a Q&A extraction device, method and program and an answering system for reducing a load of creating a scenario for teacher data or automatic answering.SOLUTION: A Q&A extraction device includes: a setting unit for setting indexes for identifying the start and the end of a question or an answer on the basis of conversation between a questioner and a respondent and a web page sentence in order to identify at least either the question or the answer; and an extraction unit for extracting a sentence from the start to the end of at least one of the question and the answer from data on the basis of the indexes set by the setting unit.SELECTED DRAWING: Figure 5

Description

The present invention relates to Q & A extraction devices, methods, programs, and response systems.

Conventionally, there is known an automatic response system (also called a chatbot) in which a computer can answer a question from a human by voice or text (Patent Document 1 and the like).

In such an automatic response system, it is necessary to collect data of a large number of questions and answers (hereinafter, also referred to as Q & A) in advance. For example, in an automatic response system, the collected question and answer data are manually created as teacher data, and machine learning is performed to generate a trained model for automatic response (machine learning type automatic response system). (In the case), or manually create a scenario for automatic response based on the collected question and answer data (in the case of a rule-based automatic response system).

However, collecting a large number of questions and answers and manually creating scenarios for teacher data and automatic responses is laborious, time-consuming, and not easy.

Therefore, one embodiment of the present invention aims to reduce the load of creating a scenario for teacher data or automatic response.

In order to solve the above-mentioned problems, one embodiment of the present invention includes a setting unit for setting an index for identifying at least one of a question and an answer, and the data from the data based on the index. It includes an extraction unit that extracts at least one of a question and the answer.

According to one embodiment of the present invention, the load of creating a scenario for teacher data or automatic response can be reduced.

It is a block diagram of the whole response system including the Q & A extraction apparatus which concerns on one Embodiment of this invention. It is a hardware block diagram of the Q & A extraction device and the response device which concerns on one Embodiment of this invention. It is a hardware block diagram of the questioning apparatus which concerns on one Embodiment of this invention. It is a figure for demonstrating the collection example of the question and answer which concerns on one Embodiment of this invention. It is a functional block diagram of the Q & A extraction apparatus which concerns on one Embodiment of this invention. It is a figure for demonstrating the index in voice of the conversation which concerns on one Embodiment of this invention. It is a figure for demonstrating the index by the format in the web page which concerns on one Embodiment of this invention. This is an example of data stored in the Q & A storage unit according to the embodiment of the present invention. It is a flowchart of the process of Q & A extraction which concerns on one Embodiment of this invention. It is a flowchart of the process of Q & A extraction which concerns on one Embodiment of this invention.

Hereinafter, each embodiment will be described with reference to the accompanying drawings. In the present specification and the drawings, components having substantially the same functional configuration are designated by the same reference numerals, so that duplicate description will be omitted.

<System configuration>
FIG. 1 is an overall configuration diagram of a response system 1 including a Q & A extraction device 10 (an example of an information processing device) according to an embodiment of the present invention. As shown in FIG. 1, the response system 1 includes a Q & A extraction device (hereinafter, also simply referred to as an extraction device) 10, a response device 20, and a questioning device 30. The response device 20 is communicably connected to the questioning device 30 and the extraction device 10 by an arbitrary network 40. Each will be described below.

Although the extraction device 10 and the response device 20 are described as separate devices in FIG. 1, the extraction device 10 and the response device 20 are mounted as one device (for example, in the existing response device 20). The extraction device 10 may be installed).

The Q & A extraction device 10 is a device that extracts at least one of the question data and the answer data used for the automatic response service provided by the response device 20. Specifically, the extraction device 10 can extract at least one of the question data and the answer data from the voice data acquired by the microphone 51 (described later with reference to FIG. 4). Further, the extraction device 10 can extract at least one of the question data and the answer data from the web page. In the latter part, the Q & A extraction device 10 will be described in detail with reference to FIG.

The response device 20 is a device that responds to a question from the question device 30. Specifically, the response device 20 receives a question from the question device 30. Further, the response device 20 transmits an answer to the question device 30.

The response device 20 may be configured to accept questions by voice or may be configured to accept questions by text. Further, the response device 20 may be configured to respond by voice or may be configured to respond by text.

The response device 20 can be configured to respond using the answer output by inputting a question into the trained model generated by machine learning. Alternatively, the response device 20 may be configured to respond according to a predetermined scenario. That is, the response device 20 performs machine learning using the question or answer data extracted by the extraction device 10 as teacher data to generate a trained model for automatic response (in the case of the machine learning type), or extracts. A scenario for automatic response can be generated (in the case of the rule-based type) based on the question and answer data extracted by the device 10.

The question device 30 is a device used by a person who wants to ask a question about the automatic response service provided by the response device 20. The questioning device 30 is composed of, for example, a digital signage 31, a controller 32, a microphone 33, and a speaker 34 as described in FIG. The questioning device 30 may be any computer such as a personal computer, a tablet, or a smartphone. For example, the questioning device 30 is a device for guiding a tourist spot installed in a tourist spot, and receives a question from a visitor of the tourist spot.

<Hardware configuration>
FIG. 2 is a hardware configuration diagram of the Q & A extraction device 10 and the response device 20 according to the embodiment of the present invention. The extraction device 10 and the response device 20 are composed of one or a plurality of computers.

The extraction device 10 and the response device 20 include a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, and a RAM (Random Access Memory) 13. The CPU 11, ROM 12, and RAM 13 form a so-called computer.

Further, the extraction device 10 and the response device 20 include an auxiliary storage device 14, a display device 15, an operation device 16, an I / F (Interface) device 17, and a drive device 18. The hardware of the extraction device 10 and the response device 20 are connected to each other via the bus 19.

The CPU 11 is an arithmetic device that executes various programs installed in the auxiliary storage device 14.

The ROM 12 is a non-volatile memory. The ROM 12 functions as a main storage device for storing various programs, data, and the like necessary for the CPU 11 to execute various programs installed in the auxiliary storage device 14. Specifically, the ROM 12 functions as a main storage device for storing boot programs such as BIOS (Basic Input / Output System) and EFI (Extensible Firmware Interface).

The RAM 13 is a volatile memory such as a DRAM (Dynamic Random Access Memory) or a SRAM (Static Random Access Memory). The RAM 13 functions as a main storage device that provides a work area that is expanded when various programs installed in the auxiliary storage device 14 are executed by the CPU 11.

The auxiliary storage device 14 is an auxiliary storage device that stores various programs and information used when various programs are executed.

The display device 15 is a display device that displays the internal state of the extraction device 10 and the response device 20.

The operation device 16 is an input device in which the manager of the extraction device 10 and the response device 20 inputs various instructions to the extraction device 10 and the response device 20.

The I / F device 17 is a communication device for connecting to the network 40 and communicating with the extraction device 10, the response device 20, and the questioning device 30.

The drive device 18 is a device for setting the storage medium 21. The storage medium 21 referred to here includes a medium such as a CD-ROM, a flexible disk, a magneto-optical disk, or the like that optically, electrically, or magnetically records information. Further, the storage medium 21 may include a semiconductor memory for electrically recording information such as an EPROM (Erasable Programmable Read Only Memory) and a flash memory.

The various programs installed in the auxiliary storage device 14 are installed, for example, by setting the distributed storage medium 21 in the drive device 18 and reading the various programs recorded in the storage medium 21 by the drive device 18. Will be done. Alternatively, the various programs installed in the auxiliary storage device 14 may be installed by being downloaded from another network different from the network 40 via the I / F device 17.

FIG. 3 is a hardware configuration diagram of the questioning device 30 according to the embodiment of the present invention. As shown in FIG. 3, the questioning device 30 can include a digital signage 31, a controller 32, a microphone 33, and a speaker 34.

The digital signage 31 is, for example, a touch panel type signage. The digital signage 31 can provide an automatic response service to a user, for example, via a web browser. Specifically, the digital signage 31 can display a screen prompting the microphone 33 to ask a question. In addition, the digital signage 31 can display a screen prompting the user to input a question using the touch panel. Further, the digital signage 31 can display the answer transmitted from the response device 20.

The controller 32 is a device for controlling the digital signage 31.

The microphone 33 acquires a voice (question) uttered by a person who wants to ask a question to the automatic response service provided by the response device 20.

The speaker 34 reproduces the voice data (answer) transmitted from the response device 20.

FIG. 4 is a diagram for explaining a collection example of questions and answers according to an embodiment of the present invention. As shown in FIG. 4, the conversation between the questioner 60 (for example, a visitor who visited a tourist spot) and the respondent 50 (for example, a guide who guides the tourist spot) is a microphone worn by the respondent 50. Recorded by 51. The Q & A extraction device 10 can extract at least one of the question data and the answer data from the voice data of the conversation acquired in this way.

<Functional block>
FIG. 5 is a functional block diagram of the Q & A extraction device 10 according to the embodiment of the present invention. As shown in FIG. 5, the extraction device 10 includes a setting unit 101, a voice acquisition unit 102, a web page search unit 103, an extraction unit 104, and a Q & A storage unit 105. Further, the extraction device 10 functions as a setting unit 101, a voice acquisition unit 102, a web page search unit 103, and an extraction unit 104 by executing a program. Each will be described below.

The setting unit 101 sets an index for identifying a question sentence or an answer sentence. Specifically, the setting unit 101 accepts the settings input by the operation device 16 of the extraction device 10 or another computer or the like. Further, the setting unit 101 stores the received setting in a memory such as in the extraction device 10 so that the extraction unit 104 can refer to it.

Here, an index for identifying a question sentence or an answer sentence will be described. Hereinafter, three examples of <index in voice of conversation>, <index by format in web page>, and <index by natural language analysis in web page> will be described.

<Indicator in conversation voice>
The setting unit 101 should be issued at the beginning and end of the question sentence and at the beginning and end of the answer sentence in the conversation between the questioner and the respondent (for example, the conversation between the guide and the visitor of the tourist spot). A wording (hereinafter, also referred to as a keyword) can be set as an index for identifying a question sentence or an answer sentence. Hereinafter, <index in the voice of conversation> will be described in detail with reference to FIG.

FIG. 6 is a diagram for explaining an index in the voice of conversation according to an embodiment of the present invention. In FIG. 6, time elapses from left to right. For example, the keyword at the beginning of the question sentence is "Yes, you have a question", the keyword at the end of the question sentence is "Your question is over", and the beginning of the answer is "The answer to that". End the answer with "that's it". By issuing these keywords by the respondent (or the person asking the question), the keywords at the beginning of the question ("Yes, you have a question") to the keywords at the end of the question ("Your question is over." The voice uttered up to that point will be identified as an interrogative sentence. Also, when the respondent (or the person asking the question) issues these keywords, the keyword at the beginning of the answer sentence (“the answer to it”) to the keyword at the end of the answer sentence (“more than”) The voice uttered in is identified as the answer sentence.

By setting a keyword at the end of the conversation, it may be possible to identify that the conversation has ended. Alternatively, by setting the start and end keywords of the conversation, the voice emitted between the start keyword of the conversation and the end keyword of the conversation may be identified as one conversation.

<Indicator by format in web page>
The setting unit 101 can set a predetermined format in the web page as an index for identifying a question sentence or an answer sentence. Hereinafter, <index by format in the web page> will be described in detail with reference to FIG. 7.

FIG. 7 is a diagram for explaining an index in a format in a web page according to an embodiment of the present invention. FIG. 7 shows a FAQ (Frequently Asked Questions and Answers) web page. When the FAQ web page is created in a predetermined format, the text written in the question text format is identified as the question text, and the text written in the answer text format is identified as the answer text. It will be. Two format examples will be described below. In addition, << format example 1 >> and << format example 2 >> may be combined.

<< Format Example 1 >>
For example, the setting unit 101 can set a predetermined attribute of HTML (HyperText Markup Language) (for example, a hidden attribute <hidden>) as an index for identifying a question sentence or an answer sentence. Therefore, the creator of the FAQ web page can use the hidden attribute <hidden> to specify the beginning and end of the question sentence and the beginning and end of the answer sentence. Since the start and end of the question sentence and the designation of the start and end of the answer sentence are hidden attributes <hidden>, they are not displayed on the user's web browser as shown in FIG.

By setting the hidden attribute <hidden> at the end of a series of sentences, it may be possible to identify that the series of sentences has ended. Alternatively, by setting the hidden attribute <hidden> at the beginning and end of a series of sentences, it is described between the hidden attribute <hidden> at the beginning of a series of sentences and the hidden attribute <hidden> at the end of a series of sentences. The text may be identified as a pair of FAQs.

<< Format Example 2 >>
For example, the setting unit 101 can set the position where the sentence is arranged in the web page as an index for identifying the question sentence and the answer sentence. Therefore, the creator of the FAQ web page can specify, for example, a sentence arranged in the left column as a question sentence and a sentence arranged in the right column as an answer sentence, as shown in FIG. 7.

<Indicator by natural language analysis in web page>
The setting unit 101 can set the question or a part of the wording in the question as an index for identifying the answer sentence. For example, the setting unit 101 inputs a question or a part of the wording in the question input to the questioning device 30 by the questioner, or a question or a part of the wording in the question specified by the system administrator of the response system 1. , Can be set as an index.

Returning to the description of FIG. The voice acquisition unit 102 acquires voice data of the conversation between the questioner and the respondent. For example, the voice acquisition unit 102 acquires voice data of a conversation between the questioner and the respondent, which is collected by the microphone 51 (see FIG. 4) worn by the respondent. In addition, the voice acquisition unit 102 converts the acquired voice data into text to generate document data. Further, the voice acquisition unit 102 stores the generated document data in a memory such as in the extraction device 10 so that the extraction unit 104 can refer to it.

In one embodiment of the present invention, the Q & A extraction device 10 processes the voice data as it is without converting the voice data into text and generating the document data (that is, the question text and the answer text are processed from the voice data. It can also be configured (specifically extracted).

The web page search unit 103 acquires a web page (HTML). For example, the web page search unit 103 collects (crawls) information from a specified range or all web pages. In addition, the web page search unit 103 converts the collected information into text and generates document data. Further, the web page search unit 103 stores the generated document data in a memory such as in the extraction device 10 so that the extraction unit 104 can refer to it.

The extraction unit 104 asks at least one of a question and an answer from the document data generated by the voice acquisition unit 102 and the document data generated by the web page search unit 103 based on the index set by the setting unit 101. Extract. In addition, the extraction unit 104 stores the extracted questions and answers in the Q & A storage unit 105. Below, it is divided into three examples: <extracting based on the index in the voice of conversation>, <extracting based on the index based on the format in the web page>, and <extracting based on the index based on natural language analysis in the web page>. I will explain.

<Extracted based on the index in the voice of conversation>
The extraction unit 104 searches the document data generated by the voice acquisition unit 102 for the start and end of the question sentence set by the setting unit 101 and the keywords to be issued at the beginning and end of the answer sentence. Further, the extraction unit 104 extracts sentences between the keywords at the beginning of the question sentence and the keywords at the end of the question sentence as questions. Further, the extraction unit 104 extracts a sentence between the keyword at the beginning of the answer sentence and the keyword at the end of the answer sentence as an answer.

In this way, in <extracting based on the index in the voice of the conversation>, the person who answers (or the person who asks the question) can register the question and the answer in the conversation as Q & A simply by issuing a predetermined keyword. it can. Further, since the question and the answer are not registered unless a predetermined keyword is issued, unnecessary conversation (for example, information not required by the response device 20) can be prevented from being registered.

<Extracted based on the index by the format in the web page>
The extraction unit 104 extracts a sentence described in the format set by the setting unit 101 from the document data generated by the web page search unit 103. Hereinafter, the above two format examples will be described separately.

<< Extraction from the web page of format example 1 >>
For example, the extraction unit 104 searches the document data generated by the web page search unit 103 for a predetermined HTML attribute (for example, a hidden attribute <hidden>) set by the setting unit 101. In addition, the extraction unit 104 searches for the start and end of the question sentence and the start and end of the answer sentence specified by using a predetermined attribute of HTML (for example, the hidden attribute <hidden>). Further, the extraction unit 104 extracts sentences between the hidden attribute <hidden> at the beginning of the question sentence and the hidden attribute <hidden> at the end of the question sentence as a question. Further, the extraction unit 104 extracts the sentences between the hidden attribute <hidden> at the beginning of the answer sentence and the hidden attribute <hidden> at the end of the answer sentence as the answer.

<< Extraction from the web page of format example 2 >>
For example, the extraction unit 104 searches the document data generated by the web page search unit 103 for a sentence arranged at a position set by the setting unit 101. Further, the extraction unit 104 extracts a sentence at a position set by the setting unit 101 that the question sentence should be arranged as a question. Further, the extraction unit 104 extracts a sentence at a position set by the setting unit 101 that the answer sentence should be arranged as an answer.

In this way, in <extracting based on the index by the format in the web page>, the creator of the web page can register the question and answer in the FAQ as Q & A simply by describing the FAQ in the predetermined format. it can. In addition, when the FAQ in the web page is updated, the questions and answers in the updated FAQ are automatically registered as Q & A.

<Extracted based on indicators by natural language analysis in web pages>
The extraction unit 104 analyzes the document data generated by the web page search unit 103 in natural language, and extracts the question set by the setting unit 101 or the answer to a part of the wording in the question.

Here, an example of the timing at which <extraction based on an index by natural language analysis in a web page> is performed will be described.

<Timing 1>
It is assumed that the response device 20 is operating using the questions and answers collected in advance by <extracting based on the index in the voice of the conversation> and <extracting based on the index in the format in the web page>. When the response device 20 cannot find the answer to the question from the question device 30, the extraction unit 104 can be configured to find the answer by <extraction based on the index by the natural language analysis in the web page>.

<Timing 2>
The extraction unit 104 <extracts based on the index by natural language analysis in the web page> in the same manner as <extracts based on the index in the voice of conversation> and <extracts based on the index by the format in the web page>. It can be configured to collect answers by. The response device 20 is a question collected by <extracting based on an index in the voice of conversation><extracting based on an index based on a format in a web page><extracting based on an index based on natural language analysis in a web page>. And can be operated using the answer.

The Q & A storage unit 105 stores the questions and answers extracted by the extraction unit 104. Hereinafter, the data stored in the Q & A storage unit 105 will be described in detail with reference to FIG.

FIG. 8 is an example of data stored in the Q & A storage unit 105 according to the embodiment of the present invention. As shown in FIG. 8, the Q & A storage unit 105 stores the question data and the answer data in association with each other. In the above-mentioned <extracting based on the index in the voice of conversation> <extracting based on the index in the format in the web page>, the question and the answer extracted by the extraction unit 104 are stored in association with each other. Further, in the above-mentioned <extracting based on the index by natural language analysis in the web page>, the question as the index and the answer extracted by the extraction unit 104 are stored in association with each other.

As shown in FIG. 8, the question data may store not only the question text but also the search keyword (keyword that will be specified by the question device 30) included in the question text. .. Further, the answer data may store not only the answer sentence but also the voice, action, and transition URL of the character when the response device 20 responds.

FIG. 9 shows a Q & A extraction according to an embodiment of the present invention <extraction based on an index in the voice of conversation>.
It is a flowchart of the process of.

In step 11 (S11), the voice acquisition unit 102 acquires voice data of the conversation between the questioner and the respondent.

In step 12 (S12), the voice acquisition unit 102 analyzes the voice data acquired in S11. Specifically, the voice acquisition unit 102 converts the voice data acquired in S11 into text to generate document data. Then, the voice acquisition unit 102 stores the generated document data in a memory such as in the extraction device 10 so that the extraction unit 104 can refer to it.

In step 13 (S13), the extraction unit 104 searches the document data generated in S12 for the keyword at the end of the conversation set by the setting unit 101 over time. If the keyword at the end of the conversation is detected, the process ends, and if it is not detected, the process proceeds to step 14.

In step 14 (S14), the extraction unit 104 searches the document data generated in S12 for the keyword at the beginning of the question sentence set by the setting unit 101 over time. If the keyword at the beginning of the question sentence is detected, the process proceeds to step 15, and if not detected, the process proceeds to step 16.

In step 15 (S15), the extraction unit 104 searches the document data generated in S12 for the keyword at the end of the question sentence set by the setting unit 101 over time. Then, the extraction unit 104 extracts the sentence between the keyword at the beginning of the question sentence and the keyword at the end of the question sentence as a question and stores it in the Q & A storage unit 105.

In step 16 (S16), the extraction unit 104 searches the document data generated in S12 for the keyword at the beginning of the answer sentence set by the setting unit 101 over time. If the keyword at the beginning of the answer sentence is detected, the process proceeds to step 17, and if it is not detected, the process returns to step 11.

In step 17 (S17), the extraction unit 104 searches the document data generated in S12 for the keyword at the end of the answer sentence set by the setting unit 101 over time. Then, the extraction unit 104 extracts the sentence between the keyword at the beginning of the answer sentence and the keyword at the end of the answer sentence as an answer, and stores it in the Q & A storage unit 105.

FIG. 10 is a flowchart of a process of Q & A extraction <extraction based on an index in a format in a web page> according to an embodiment of the present invention.

In step 21 (S21), the web page search unit 103 acquires the web page (HTML). Specifically, the web page search unit 103 collects (crawls) information from a designated range or all web pages.

In step 22 (S22), the web page search unit 103 analyzes the web page acquired in S21. Specifically, the web page search unit 103 converts the collected information into text and generates document data. Then, the web page search unit 103 stores the generated document data in a memory such as in the extraction device 10 so that the extraction unit 104 can refer to it.

In step 23 (S23), the extraction unit 104 searches the document data generated in S22 for the hidden attribute <hidden> at the end of a series of sentences set by the setting unit 101 in order from the beginning. If the hidden attribute <hidden> at the end of a series of sentences is detected, the process ends, and if it is not detected, the process proceeds to step 24.

In step 24 (S24), the extraction unit 104 searches the document data generated in S22 for the hidden attribute <hidden> at the beginning of the question sentence set by the setting unit 101 in order from the beginning. If the hidden attribute <hidden> at the beginning of the question sentence is detected, the process proceeds to step 25, and if it is not detected, the process proceeds to step 26.

In step 25 (S25), the extraction unit 104 searches the document data generated in S22 for the hidden attribute <hidden> at the end of the question sentence set by the setting unit 101 in order from the beginning. Then, the extraction unit 104 extracts the sentences between the hidden attribute <hidden> at the beginning of the question sentence and the hidden attribute <hidden> at the end of the question sentence as a question and stores them in the Q & A storage unit 105.

In step 26 (S26), the extraction unit 104 searches the document data generated in S22 for the hidden attribute <hidden> at the beginning of the answer sentence set by the setting unit 101 in order from the beginning. If the hidden attribute <hidden> at the beginning of the answer sentence is detected, the process proceeds to step 27, and if it is not detected, the process returns to step 21.

In step 27 (S27), the extraction unit 104 searches the document data generated in S22 for the hidden attribute <hidden> at the end of the answer sentence set by the setting unit 101 in order from the beginning. Then, the extraction unit 104 extracts the sentence from the hidden attribute <hidden> at the beginning of the answer sentence to the hidden attribute <hidden> at the end of the answer sentence as an answer and stores it in the Q & A storage unit 105.

Each function of the embodiment described above can be realized by one or more processing circuits. Here, the "processing circuit" in the present specification is a processor programmed to execute each function by software such as a processor implemented by an electronic circuit, or a processor designed to execute each function described above. It shall include devices such as ASIC (Application Specific Integrated Circuit), DSP (digital signal processor), FPGA (field programmable gate array) and conventional circuit modules.

The present invention is not limited to the configurations shown here, such as combinations with other elements in the configurations and the like described in the above embodiments. These points can be changed without departing from the spirit of the present invention, and can be appropriately determined according to the application form thereof.

1 Response system 10 Q & A extraction device 20 Response device 30 Question device 40 Network 31 Digital signage 32 Controller 33 Microphone 34 Speaker 50 Respondent 51 Microphone 60 Questioner 101 Setting unit 102 Voice acquisition unit 103 Web page search unit 104 Extraction unit 105 Q & A Memory

Japanese Unexamined Patent Publication No. 2001-256036

Claims (10)

A setting unit that sets an index to identify at least one of a question and an answer,
An information processing device including an extraction unit that extracts at least one of the question and the answer from the data based on the index. Further provided with a voice acquisition unit that acquires voice data of a conversation between a questioner and a respondent, converts the voice data into text, and generates the data.
The information processing apparatus according to claim 1, wherein the index is a keyword issued at the beginning and end of the question and the beginning and end of the answer. It further includes a web page search unit that collects information from a specified range of web pages or the entire web page, converts the information into text, and generates the data.
The information processing apparatus according to claim 1, wherein the index is in a predetermined format in the web page. The predetermined format is a hidden attribute and
The text between the hidden attribute at the beginning of the question and the hidden attribute at the end of the question is extracted as the question, and the text between the hidden attribute at the beginning of the answer and the hidden attribute at the end of the answer is extracted as the answer. , The information processing apparatus according to claim 3. The predetermined format is a position in the web page where the text is arranged.
The information processing apparatus according to claim 3, wherein the sentence at the position where the question should be arranged is extracted as the question, and the sentence at the position where the answer should be arranged is extracted as the answer. It further includes a web page search unit that collects information from a specified range of web pages or the entire web page, converts the information into text, and generates the data.
The indicator is a question or some wording within the question.
The information processing apparatus according to claim 1, wherein the extraction unit analyzes the data in natural language and extracts an answer to the question or a part of the wording in the question. The extraction unit
Get the question or some wording in the question and
The information processing apparatus according to claim 2 or 3, wherein a specified range of a web page or the entire web page is analyzed in natural language, and an answer to the question or a part of the wording in the question is extracted. The way the computer does
Steps to set indicators to identify at least one of the question and answer,
A method comprising the step of extracting at least one of the question and the answer from the data based on the index. A setting unit that sets an index to identify at least one of a question and an answer on a computer.
A program for functioning as an extraction unit that extracts at least one of the question and the answer from the data based on the index. A response system including an information processing device, a response device, and a question device.
The information processing device
A setting unit that sets an index to identify at least one of a question and an answer,
An extraction unit that extracts at least one of the question and the answer from the data based on the index is provided.
The response device transmits an answer to the question received from the question device based on the question and the answer extracted by the extraction unit.
The questioning device sends a question to the answering device and receives an answer from the answering device.
Response system.

Images