JP2022080691A - Document creation assisting device, document creation assisting method and program - Google Patents

Abstract

To provide a technology that assists creation of a questioning and answering document.SOLUTION: A document creation assisting device 100 includes: a source sentence obtaining unit 121 that obtains a sentence from a manual; a dividing unit 122 that divides the sentence obtained by the source sentence obtaining unit 121 into a subject element, a within-sentence element, and a sentence-end element in accordance with a predetermined rule; a question creating unit 125 that creates a question sentence using the respective elements divided by the dividing unit 122; and an answer creating unit 126 that creates, using the respective elements divided by the dividing unit 122, an answer for the question sentence created by the question creating unit 125.SELECTED DRAWING: Figure 1

Description

The present invention relates to a document creation assisting technique.

In the past, people have created and prepared Q & A (Question and Answer: hereinafter referred to as "question and answer documents") such as FAQ (Frequently Asked Questions) with reference to response logs and manuals. Recently, a method of creating an FAQ using a response log has been proposed.

In the case of the method of creating an FAQ using a response log, it is not possible to create an FAQ without a contact center that does not have a response log or a response log immediately after a new product or service is released. In addition, it takes a relatively long time to manually create a question and answer document.

Japanese Unexamined Patent Publication No. 2020-077091 Japanese Unexamined Patent Publication No. 2020-071690

In view of the above circumstances, it is an object of the present invention to provide a technique for supporting the preparation of a question and answer document.

One aspect of the present invention includes an acquisition unit that acquires a sentence from a manual, a division unit that divides the sentence acquired by the acquisition unit into each element of the subject, the sentence, and the end of the sentence according to a predetermined rule, and the division. A question generation unit that generates a question sentence using elements divided by a unit, and an answer generation unit that generates an answer corresponding to the question sentence generated by the question generation unit using the elements divided by the division unit. It is a document creation support device equipped with a department.

One aspect of the present invention is the document creation support device, and the predetermined rule is a character string included in a sentence by using correspondence information in which a character string and a score corresponding to the character string are associated with each other. Of these, the rule is to divide a sentence using the character string with the highest score.

One aspect of the present invention is the document creation support device, which is a determination unit for determining the type of sentence end divided by the division unit, and a question template according to the type of sentence end determined by the determination unit. , With a selection section to select the answer template.

One aspect of the present invention is the document creation support device, in which the question generation unit generates a question sentence using a question template selected by the selection unit.

One aspect of the present invention is the document creation support device, in which the answer generation unit generates an answer sentence using the answer template selected by the selection unit.

One aspect of the present invention is an acquisition stage in which a document creation support device acquires a sentence from a manual, and a sentence acquired by the document creation support device in the acquisition stage, according to a predetermined rule, in the subject, in the sentence, and at the end of the sentence. A division stage for dividing into each element of the above, a question generation stage in which the document creation support device generates a question sentence using the elements divided by the division stage, and a document creation support device are generated by the question generation stage. It is a document creation support method including an answer generation step of generating an answer corresponding to a question sentence by using the elements divided by the division step.

One aspect of the present invention is a program for operating a computer as a document creation support device, in which an acquisition unit that acquires a sentence from a manual and a sentence acquired by the acquisition unit are subjected to a predetermined rule. Corresponds to a division unit that divides into each element of the subject, the sentence, and the end of a sentence, a question generation unit that generates a question sentence using the elements divided by the division unit, and a question sentence generated by the question generation unit. It is a program that functions as an answer generation unit that generates an answer by using the elements divided by the division unit.

INDUSTRIAL APPLICABILITY According to the present invention, it is possible to provide a technique for supporting the preparation of a question and answer document.

It is a functional block diagram which shows the structure of the document creation support apparatus 100. It is a figure which shows the subject, the sentence, and the end of a sentence. It is a figure which shows the specific example of the sentence end database 131. It is a figure which shows the specific example of the subject database 132. It is a figure which shows the specific example of the subject database 132. It is a figure which shows the specific example of the subject database 132. It is a figure which shows the specific example of the subject database 132. It is a figure which shows the specific example of the database 133 in a sentence. It is a figure which shows the specific example of the adaptation database. It is a flowchart which shows the processing flow of the document creation support apparatus 100.

FIG. 1 is a functional block diagram showing the configuration of the document creation support device 100. The document creation support device 100 includes a CPU (Central Processing Unit), a memory, an auxiliary storage device, and the like connected by a bus, and by executing a support program, the communication unit 110, the support information storage unit 130, and the control unit 120 are combined. Functions as a device to be equipped.

All or part of each function of the communication unit 110, the support information storage unit 130, and the control unit 120 may be an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), an FPGA (Field Programmable Gate Array), or the like. It may be realized by using hardware. The assistive program may be recorded on a computer-readable recording medium. The computer-readable recording medium is, for example, a flexible disk, a magneto-optical disk, a portable medium such as a ROM or a CD-ROM, or a storage device such as a hard disk built in a computer system. The management program may be transmitted over a telecommunication line.

The communication unit 110 is a network interface. The communication unit 110 communicates with another device. For example, the communication unit 110 performs communication for acquiring an electronic manual from another device. In this case, the acquired manual is stored in the holding storage device or the memory.

The support information storage unit 130 is configured by using a storage device such as a magnetic hard disk device or a semiconductor storage device. The support information storage unit 130 stores the sentence end database 131, the subject database 132, the sentence middle database 133, and the adaptation database 134.

Prior to the explanation of each of these databases and the control unit 120, the subject, the sentence, the sentence end, and the like in the present embodiment will be described. First, the document creation support device 100 acquires one sentence (also referred to as “target sentence”) from the manual. Manuals typically contain sentences such as chapters and chapter names. Therefore, in this embodiment, a manual provided with a tag indicating that it is a target sentence is used so that a target sentence in which a question sentence and an answer sentence can be created can be easily obtained.

For example, a tag such as "Chapter 2 VPN <p> When using VPN, you cannot log in twice using the same user ID as the already connected user ID </ p>". This is a manual with <p> and </ p>. The document creation support device 100 acquires a character string enclosed in tags <p> and </ p> as a target sentence. Document creation can be supported without this tag, but it takes more time than with the tag.

The document creation support device 100 divides the target sentence into each element of the subject, the sentence, and the end of the sentence according to a predetermined rule. FIG. 2 is a diagram showing a subject, a sentence, and a sentence end (sentence end 1, sentence end 2) in which the target sentence is divided. The document creation support device 100 first searches for the delimiter A between the sentence end 1 and the sentence end 2, and divides the sentence end 2 from the target sentence. Next, the document creation support device 100 searches for the delimiter B and divides the subject from the target sentence. The document creation support device 100 searches for the delimiter C and divides it into a sentence and a sentence end 1. The delimiters A, B, and C do not always exist.

Each database and the control unit 120 will be described based on the above-mentioned subject matter, sentence middle, sentence end, and the like. As shown in FIG. 1, the control unit 120 includes a target sentence acquisition unit 121, a division unit 122, a sentence end determination unit 123, a selection unit 124, a question generation unit 125, and an answer generation unit 126.

The target sentence acquisition unit 121 acquires a sentence surrounded by the above-mentioned tags as a target sentence from the manual. As will be described with reference to FIG. 2, the division unit 122 divides the target sentence into each element of the subject, the sentence, and the sentence end (sentence end 1, sentence end 2) according to a predetermined rule. The sentence end determination unit 123 determines the type of the sentence end of the target sentence. The selection unit 124 selects a question template and an answer template according to the determined type. The question generation unit 125 generates a question sentence using the divided elements and the selected question template. The answer generation unit 126 generates an answer corresponding to the question sentence generated by the question generation unit 125 by using the divided elements and the selected answer template.

FIG. 3 is a diagram showing a specific example of the sentence end database 131. The sentence end database is a database showing a character string for searching for the delimiter A. The sentence end database is composed of priority 1, priority 2, priority 3, and character strings corresponding to those priorities. In FIG. 3, "|" indicates "or". For example, A | B indicates A or B.

The division unit 122 searches for a character string or the like corresponding to the target sentence in the order of priority 1, priority 2, and priority 3. The "time" in priority 1 is a character string indicating a time such as a date or a date and time. For example, "September 30, 2020" can be mentioned as a character string indicating the time. Priority 1 indicates a character string having "kara", "yori", "de", "with", "no", and "ni" following this "time". For example, "from September 30, 2020", "with September 30, 2020", etc.

If there is a target sentence corresponding to the character string shown in priority 1, the division unit 122 divides the target sentence by setting immediately after the corresponding character string as the delimiter A. For example, when the target sentence is "○○ is the end of service on September 30, 2020", the delimiter A is immediately after "mote". If there is no character string indicated by priority 1 in the target sentence, the division unit 122 searches for a character string indicated by priority 2.

(Comma) in priority 2 and 3 indicates ",". The division unit 122 searches for "because", "to", "when", and "sometimes" shown in priority 2. If there is a target sentence corresponding to the character string shown in priority 2, the division unit 122 divides the target sentence by setting immediately after the corresponding character string as the delimiter A. For example, when the target sentence is "○○ is carried out in XX and is not carried out in YY", the delimiter A is immediately after "because". If there is no character string indicated by priority 2 in the target sentence, the division unit 122 searches for a character string indicated by priority 3.

The division unit 122 searches for "if," "if," and "possible." Or "necessary." In the target sentence. If there is a target sentence corresponding to the character string shown in priority 3, the division unit 122 divides the target sentence by setting the delimiter A immediately before the sentence following "possible" or "necessary". do. For example, when the target sentence is "XX is required for XX", "XX is required" is a sentence following "necessary", so the sentence immediately before "XX is required" is separated from A. do. If the target sentence does not correspond to the character string shown in priority 3, the division unit 122 does not have the delimiter A.

Using the sentence end database 131 described above, the division unit 122 searches for the division A. When the delimiter A is searched, the division unit 122 sets the sentence end 2 from the delimiter A to the end of the target sentence. Next, the division unit 122 searches for the division B. The subject database 132 is a database showing a character string for searching for the delimiter B. 4A, 4B, 4C, and 4D are diagrams showing specific examples of the subject database 132. In FIGS. 4A, 4B, 4C, and 4D, (comma) indicates "," as described above.

The subject database 132 is an example of correspondence information in which a character string and a score corresponding to the character string are associated with each other. In FIGS. 4A, 4B, 4C, and 4D, the character strings are arranged in descending order of score (from 120 to 1). For example, in FIG. 4A, the score of "by" is 120. In FIG. 4B, the score of "when" is 70. In FIG. 4C, the score of "ha" is 11. In FIG. 4D, the score of "to" is 7.

The division unit 122 searches for the character strings shown in FIGS. 4A, 4B, 4C, and 4D from the beginning of the target sentence to the delimiter A (until the end if there is no delimiter A), and the searched characters. Of the columns, the character string immediately after the character string with the highest score is defined as the delimiter B. As for the character strings having the same score, the higher-ranked character strings are given priority in the order shown in FIGS. 4A, 4B, 4C, and 4D. For example, when "kotode" and "de" with a score of 120 are included in the target sentence at the same time, "kotode" is higher in the order shown in FIG. 4A. , ”Is immediately followed by the delimiter B. If the target sentence does not correspond to the character strings shown in FIGS. 4A, 4B, 4C, and 4D, the division unit 122 does not have the division B.

In this way, the predetermined rule uses the correspondence information (subject database 132) in which the character string and the score corresponding to the character string are associated with each other, and the character having the highest score among the character strings included in one sentence. It is a rule to divide a sentence using columns.

Using the subject database 132 described above, the division unit 122 searches for the division B. When the delimiter B is searched, the division unit 122 has the subject from the beginning of the target sentence to the delimiter B. Next, the division unit 122 searches for the division C. The in-text database 133 is a database showing a character string for searching for the delimiter C. FIG. 5 is a diagram showing a specific example of the database 133 in the text. In FIG. 5, (comma) indicates "," as described above. Further, in FIG. 5, (hope) indicates a desired sentence. For example, (hope) includes "want" such as "want" and "want to do".

The in-text database 133 is composed of a score and a corresponding replacement and character string. FIG. 5 shows the replacements and character strings arranged in descending order of score (from 100 to 2). The replacement indicates a character string to be replaced instead of the character string when the question sentence is generated. The underscore also indicates replacement using the BERT (Bidirectional Encoder Representations from Transformers) technique.

The division unit 122 searches for the character string shown in FIG. 5 from the delimiter B to the delimiter A of the target sentence (from the end if there is no delimiter A, or from the beginning of the target sentence if there is no delimiter B). Then, among the searched character strings, the character string immediately after the character string having the highest score is set as the delimiter C. As for the character strings having the same score, the higher-ranked character strings are given priority in the order shown in FIG. When the delimiter C is searched, the division unit 122 sets the delimiter B to the delimiter C in the sentence. If the target sentence does not correspond to the character string shown in FIG. 5, the division unit 122 does not have the delimiter C.

For example, the target sentence is "In the case of XX, (separation B), in principle, XX cannot be set as YY (separation A)". At this time, since there is "as" in "in principle, XX is YY" from the delimiter B to the delimiter A, the delimiter C is immediately after the "as". And the replacement, which corresponds to the underscore, is the "principle". Is it possible to ask the question at this time, "In the case of XX, (separation B), in principle, (separation C) XX can be set as YY (separation A)"? Will be.

When the target sentence is divided by the division unit 122, the sentence end determination unit 123 determines which type of sentence end is the sentence end of the target sentence. FIG. 6 is a diagram showing a specific example of the adaptation database. The adaptation database 134 is composed of a sentence end, a question sentence template corresponding to the sentence end, and an answer sentence template score.

As shown in FIG. 6, the types of sentence endings include "because", "denial", "confirmation", "negative word", "case (necessary)", "necessary", "impossible", and "date and time". , "Case (possible)", "Other". The sentence end determination unit 123 determines which of the above types the sentence end of the target sentence is.

When the sentence end type is determined by the sentence end determination unit 123, the question generation unit 125 creates a question sentence using the question sentence template corresponding to the determined type. The question generation unit 125 generates a question sentence using the subject, the sentence, or the element at the end of the sentence (mainly the subject).

The answer generation unit 126 generates an answer sentence by using the answer sentence template corresponding to the determined type. The answer generation unit 126 generates a question sentence using the subject, the sentence, or the element at the end of the sentence (mainly in the sentence or the end of the sentence).

Next, specific examples of each type will be described. In the following description, the elements are shown using curly braces. Specifically, the divided target sentence is indicated as {subject} {in the sentence} {sentence end 1} {sentence end 2}. As mentioned above, since there is not always a delimiter, in that case, nothing is described in the curly braces. For example, in the case of {X} {} {Y} {Z}, since there was no delimiter C, the text was not found, so nothing was found in the second curly braces corresponding to the text. Not listed. In addition, any of the target sentences used below is an example.

First, "for" will be explained.
Target sentence: {Screen image related to construction operation acquisition} {Construction operation acquisition is carried out by the collaboration company} {Because} {In principle, it is not carried out at the collaboration center}
Since the end of the sentence is {because}, the sentence end determination unit 123 determines that the type of the end of the sentence is "because". At this time, is the question sentence template {subject} {why} {sentence end 2}? Is. Therefore, the question generation unit 125 generates the following as a question sentence.
{Screen image related to construction operation acquisition} {Why} {In principle, the collaboration center does not implement it}?
The answer sentence template is {in the sentence} {end of sentence 1} {end of sentence 2}. Therefore, the answer generation unit 126 generates the following as an answer sentence.
{Implemented by the collaboration company} {Because} {In principle, it is not implemented at the collaboration center}.

Explain "denial". "FLET'S" in the following description is a registered trademark.
Target text: {FLET'S Hikari Next Family Type / Family / High Speed Type / Family / Super High Speed Type Hayabusa Mansion Type 1 / Mansion / High Speed Type / Mansion / Super High Speed Type Hayabusa / Fretz / Hikari Premium Family Type Mansion Type 11 For the condominium type,} {wiring method} {} {regardless}
Since the end of the sentence is {regardless}, the sentence end determination unit 123 determines that the type of the sentence end is "negative". At this time, is the question sentence template {subject} {inside the sentence} {end of sentence 1} {end of sentence prototype}? Is. Therefore, the question generation unit 125 generates the following as a question sentence.
{FLET'S Hikari Next Family Type / Family / High Speed Type / Family / Super High Speed Type Hayabusa Mansion Type 1 / Mansion / High Speed Type / Mansion / Super High Speed Type Hayabusa / Fretz / Hikari Premium Family Type Mansion Type 11 Mansion Type Is} {wiring method} {} {question}?
The answer sentence template is {end of sentence 2}. Therefore, the answer generation unit 126 generates the following as an answer sentence.
{No matter}.

"Confirmation" will be described.
Target sentence: {In the case of ARENA} {Confirm the billing method and storage status of the monthly fee on the existing confirmation screen} {} {Confirm}
Since the end of the sentence is {confirm}, the sentence end determination unit 123 determines that the type of the sentence end is "confirmation". At this time, where can I check the question text template? }. Therefore, the question generation unit 125 generates the following as a question sentence.
{In the case of ARENA} {Where can I check? }
The answer sentence template is {in the sentence} {end of sentence 1} {end of sentence 2}. Therefore, the answer generation unit 126 generates the following as an answer sentence.
{Confirm the billing method and storage status of the monthly fee on the existing confirmation screen} {} {Confirm}.

Explain the "negative word".
Target sentence: {Processing such as article posting at the collaboration center} {} {} {unnecessary}
Since the sentence end is {unnecessary}, the sentence end determination unit 123 determines that the type of the sentence end is "negative word". At this time, is the question sentence template {subject} {in the sentence} {end of sentence 1} (denial of {end of sentence 2})? Is. Therefore, the question generation unit 125 generates the following as a question sentence.
{Processing such as article posting at the collaboration center} {necessary}?
The answer sentence template is {end of sentence 2}. Therefore, the answer generation unit 126 generates the following as an answer sentence.
{Unnecessary}.

The "case (necessary)" will be explained.
Target sentence: {If the customer wishes to abolish the discount service such as existing subscriber phones,} {} {} {It is necessary to abolish the product according to the FLET'S Hikari construction date}
Since the target sentence includes "case" and {necessary} at the end of the sentence, the sentence end determination unit 123 determines that the type of sentence end is "case (necessary)". At this time, is the question sentence template {subject} {inside the sentence} {end of sentence 1} {end of sentence 2}? Is. Therefore, the question generation unit 125 generates the following as a question sentence.
{If the customer wishes to abolish the discount service such as existing subscriber phones,} {} {} {It is necessary to abolish the product according to the FLET'S Hikari construction date}?
In addition, the answer sentence template requires {subject}. Is. Therefore, the answer generation unit 126 generates the following as an answer sentence.
{If the customer wishes to abolish the discount service such as existing subscriber phones}, it is necessary.

Explain "necessity".
Target sentence: {In the address registration of the equipment department, the address of the place you want to open is based on the Japanese name or map} {The location is specified and the supplementary information required for the registration process is input} {} {Necessary Become}
Since {necessary} is included at the end of the sentence, the sentence end determination unit 123 determines that the type of the sentence end is "necessary". At this time, is the question text template {subject} {} {} {what is needed}? Is. Therefore, the question generation unit 125 generates the following as a question sentence.
{In the address registration of the equipment department, the address of the place you want to open is based on the Japanese name or map} {} {} {What is required}?
The answer sentence template is {in the sentence} {end of sentence 1} {end of sentence 2}. Therefore, the answer generation unit 126 generates the following as an answer sentence.
{It is necessary to input supplementary information, etc. required when specifying the position and performing the registration process} {} {necessary}.

Explain "impossible".
Target sentence: {In principle, make an address registration request and make corrections if you cooperate with the equipment department} {} {Cannot be done}
Since {cannot} is included at the end of the sentence, the sentence end determination unit 123 determines that the type of the sentence end is "impossible". At this time, can the question sentence template be {subject} {in the sentence} {end of sentence 1} {end of sentence stem}? Is. Therefore, the question generation unit 125 generates the following as a question sentence.
{In principle, make an address registration request,} {can make corrections if you cooperate with the equipment department} {} {can do}?
The answer sentence template is {end of sentence 2}. Therefore, the answer generation unit 126 generates the following as an answer sentence.
{Cannot be implemented}.

The "date and time" will be described.
Target sentence: {1 FLET'S Hikari Premium} {On January 31, 2019} {} {End of service}
Since the target sentence includes the date and time and {end} at the end of the sentence, the sentence end determination unit 123 determines that the type of the end of the sentence is "date and time". At this time, when is the question sentence template {subject} and {end of sentence 2}? Is. Therefore, the question generation unit 125 generates the following as a question sentence.
When is {1 FLET'S Hikari Premium} {end of service}?
The answer sentence template is {in the sentence} {end of sentence 1} {end of sentence 2}. Therefore, the answer generation unit 126 generates the following as an answer sentence.
{On January 31, 2019} {} {End of service}.

The "case (possible)" will be described.
Target sentence: {Judgment In the case of a detached house} {} {} {It is possible to judge whether or not dispatch is possible only by the system judgment result}
Since the target sentence includes a case and {possible} is included at the end of the sentence, the sentence end determination unit 123 determines that the type of the sentence end is "case (possible)". At this time, is the question sentence template {subject} {inside the sentence} {end of sentence 1} {end of sentence 2}? Is. Therefore, the question generation unit 125 generates the following as a question sentence.
{Judgment In the case of a detached house} {} {} {It is possible to judge whether or not dispatch is possible only by the system judgment result}?
The answer sentence template is {in the sentence} {end of sentence 1} {end of sentence 2}. Therefore, the answer generation unit 126 generates the following as an answer sentence.
{} {} {It is possible to judge whether or not dispatch is possible only by the system judgment result}.

"Other" will be explained. If it does not correspond to any of the above types, it is judged as "other".
Target sentence: {End user's order} {By the collaboration company, from the order reception system to the collaboration center} {} {Intermediate}
Since it does not correspond to any of the types, the sentence end determination unit 123 determines that the sentence end type is "other". At this time, is the question sentence template {{subject} {interrogative word} {sentence end 1} {sentence end 2}? Is. The interrogative word contained in the {interrogative word} in this interrogative sentence template is automatically selected by the above BERT. The question generation unit 125 generates the following as a question sentence.
{End-user orders} {where} {} {can be relayed}?
The answer sentence template is {in the sentence} {end of sentence 1} {end of sentence 2}. Therefore, the answer generation unit 126 generates the following as an answer sentence.
{By the collaboration company, from the order reception system to the collaboration center} {} {Intermediary}.

The types at the end of sentences described above are examples, and it is possible to add types as appropriate. By preparing the question sentence template and the answer sentence template according to the type of the sentence end in this way, the question sentence and the answer sentence can be generated with high accuracy.

FIG. 7 is a flowchart showing a processing flow of the document creation support device 100. The target sentence acquisition unit 121 acquires the target sentence from the manual (step S101). The division unit 122 divides the target sentence acquired by the target sentence acquisition unit 121 into a subject, a sentence, a sentence end 1, and a sentence end 2 (step S102). As described above, if there is no delimiter, there may be no subject, sentence, sentence end 1, and sentence end 2.

The sentence end determination unit 123 determines the type of sentence end (step S103). The selection unit 124 selects a template according to the type of the sentence end determined by the sentence end determination unit 123 (step S104). The question generation unit 125 generates a question sentence by applying it to the question template selected by the selection unit 124 (step S105). The answer generation unit 126 generates a question sentence by applying the answer corresponding to the question sentence generated by the question generation unit 125 to the answer template selected by the selection unit 124 (step S106).

By doing so, it is possible to generate a question sentence and an answer sentence with high accuracy, so that it is possible to support the creation of a question and answer document. In fact, when an experiment was conducted using the "FLET'S Access Service Charge Business Manual" (page 314) as a manual, and whether or not there was a sense of discomfort in Japanese and whether or not it was appropriate as a question and answer document, 74% of the respondents. It was confirmed that the appropriate questions and answers were given in accurate and appropriate Japanese. Similarly, when an experiment was conducted using "FLET'S VPN Wide Service Knowledge Edition" (page 112) as a manual, and whether or not there was a sense of discomfort in Japanese and whether or not it was appropriate as a question and answer document was judged by humans, 72%. It was confirmed that the appropriate questions and answers were made in appropriate Japanese with the accuracy of.

Below are examples of successful output results from the above experiments. In the following output result, the target sentence, the question sentence, and the answer sentence are described in order for each paragraph. In the output result, there were cases where two punctuation marks were consecutive, but for the sake of clarity, one is used. In addition, the types (1) to (10) at the end of the sentence are described separately. As described above, since it is possible to add types to the end of sentences as appropriate, the end of sentences other than the above is also included. In the following, (1) is "comma (comma)", (2) is "possible", (3) is "confirmation", (4) is "case (necessary)", and (5) is "case (possible)". , (6) is "denial", (7) is "negative word", (8) is "impossible", (9) is "unnecessary", and (10) is "other". The output results are described in this order.

(1) The output result of "comma (comma)" is shown below.

{FLET'S Access Service} {Because it is a completely flat-rate service} {Because} {The use of this service is not displayed in the "Call Details Breakdown"}
{The FLET'S Access Service} {Why} {The use of the service is not displayed in the "Call Details Breakdown"}?
{Because it is a completely flat-rate service} {Because} {The use of the service is not displayed in the "Call details breakdown"}.

{When using center line connection and east-west connection at the same time} {Because the route information of the network address for the customer base registered in the "static route" is transferred to the East Japan area} {From the East Japan base to the customer base Communication to the network address is possible}
{When using center line connection and east-west connection at the same time} {Why} {Communication from the East Japan base to the customer base network address is possible}?
{Because the route information of the customer base network address registered in the "static route" is transferred to the East Japan area} {Communication from the East Japan base to the customer base network address is possible}.

(1) The output result of "comma (comma)" is over.

(2) The output result of "possible" is shown below. In addition, "NTT" in the following description is a registered trademark.

{For customers using access services} {It is possible to easily build a private network across 30 prefectures in the NTT West area} {} {without the need for new communication equipment} {} {
{If you are using the access service, what can you do} {} {} {what can you do}?
{It is possible to easily build a private network that spans the 30 prefectures of the NTT West area without the need for new communication equipment} {} {.

{By applying for this service by the VPN administrator of NTT West, NTT East's FLET'S VPN Wide basic service, or the subscriber of NTT East's FLET'S VPN prio service} {NTT East area Communication between the VPN and the corresponding VPN in the NTT West area is possible, and it is possible to build a wider private network} {} {is possible}
{By applying for this service by the VPN administrator of NTT West, NTT East's FLET'S VPN Wide basic service, or the contractor of NTT East's FLET'S VPN Prio service, respectively} {} {} {What can you do}?
{Communication between the corresponding VPNs in the NTT East area and NTT West area is possible, and it is possible to build a wider private network} {} {.

(2) The output result of "possible" is over.

(3) The output result of "confirmation" is shown below.

{For FLET'S Hikari Next My Town} {Monthly usage fee varies depending on each municipality} {Because each municipality's terms of use for My Town service} {Please check}
{For FLET'S Hikari Next My Town} {Where can I find out? }
{Monthly usage fee varies depending on each municipality} {Please check the terms and conditions of My Town Service of each municipality} {Please check}.

{For available IP addresses for VPN connection} {Refer to "Chapter 2, Section 4, 6. IP Address"} {} {
{Available VPN connection IP address} {Where can I find it? }
{See "Chapter 2, Section 4, 6. IP Address"} {} {See}.

(3) The output result of "confirmation" is over.

(4) The output result of "Case (necessary)" is shown below.

{When using the group identifier / company identifier used in the Fretz Group or Fretz Office as the company identifier in this service} {} {} {Before using the identifier in this service, Fretz · Group or Fretz offices need to be abolished}
{When using the group identifier / company identifier used in the Fretz Group or Fretz Office as the company identifier in this service} {} {} {Before using the identifier in this service, Fretz · Need to abolish group or Fretz office}?
{If you want to use the group identifier / company identifier used by FLET'S Group or FLET'S Office as the company identifier in this service}, it is required.

{1 When changing the administrator in the basic service} {} {} {Center line connection service must be transferred}
{1 When changing the administrator in the basic service} {} {} {The center line connection service must be transferred}?
{1 When changing the administrator in the basic service} is required.

(4) The output result of "Case (necessary)" is over.

(5) The output result of "Case (possible)" is shown below.

{When using a global IP address as an IP address} {} {} {It is also possible for a third party to use the global IP address assigned by a NIC such as JPNIC}
{When using a global IP address as an IP address} {} {} {It is also possible for a third party to use the global IP address assigned by a NIC such as JPNIC}?
{} {} {It is also possible for a third party to use the global IP address assigned by NIC such as JPNIC}.

{If you change the IP address that is set to discard unnecessary packets} {} {} {You can change the setting to discard unnecessary packets}
{If you change the IP address that is set to discard unnecessary packets} {} {} {You can change the setting to discard unnecessary packets}?
{} {} {You can change the settings for discarding unnecessary packets at your request}.

(5) The output result of "Case (possible)" is above.

(6) The output result of "Negation" is shown below.

{} {Minimum usage period and its} {} {There is no cancellation fee}
{} {Minimum usage period and its} {} {There is a cancellation fee}?
{There is no cancellation fee}.

{Note 4. If the person in charge of construction does not arrive by the designated time,} {charge will not be charged} {} {will not be done}
{Note 4. If the person in charge of construction does not arrive by the designated time} {Bill a fee} {} {I will do it}?
{Do not do}.

(6) The output result of "Negation" is over.

(7) The output result of "negative word" is shown below.

{Note 1. Of the basic construction costs, the amount equivalent to the procedure cost = 1,000 yen} {extra} {} {not applicable}
{Note 1. Of the basic construction costs, the amount equivalent to the procedure cost = 1,000 yen} {extra} {} {target}?
{Not applicable}.

(7) The output result of "negative word" is as above.

(8) The output result of "impossible" is shown below.

{When using VPN} {Use the same user ID as the already connected user ID} {Use twice} {You cannot log in}
Is it possible to {when using a VPN} {use the same user ID as the already connected user ID} {use twice} {login}?
{You cannot log in}.

{} {Multiple IP addresses can be used with one user ID} {} {Cannot be used}
{} {Can I use multiple IP addresses with one user ID} {} {can I use}?
{Cannot be used}.

(8) The output result of "impossible" is over.

(9) The output result of "unnecessary" is shown below.

{} {There is no need to prepare a new authentication function} {No need} when using this service.
{} {Is it necessary to prepare a new authentication function} {to use this service}?
{No need}.

{No fee for succession or renaming of the right to use this service} {} {} {Not required}
{A fee for succession or renaming of the right to use this service} {} {} {Is it necessary}?
{there is no need}.

(9) The output result of "unnecessary" is above.

(10) The output result of "Other" is shown below.

{Lithium-ion batteries are} {precious recyclable} {} {resources}
{Lithium-ion batteries} {what} {} {resources}?
{Recyclable precious} {} {Resource}

{The secondary SSID is} {on the wireless LAN setting label affixed to the inside of this product} {} {is described}
Is the {secondary SSID} {where} {} {listed}?
{On the wireless LAN setting label affixed to the inside of this product} {} {It is described}

(10) The output result of "Other" is as described above.

Although the embodiments of the present invention have been described in detail with reference to the drawings, the specific configuration is not limited to this embodiment, and includes designs and the like within a range that does not deviate from the gist of the present invention.

100 ... document creation support device, 110 ... communication unit, 120 ... control unit, 121 ... target sentence acquisition unit, 122 ... division unit, 123 ... sentence end determination unit, 124 ... selection unit, 125 ... question generation unit, 126 ... answer generation Department, 130 ... Support information storage department, 131 ... Sentence end database, 132 ... Subject database, 133 ... Sentence database, 134 ... Adaptive database

One aspect of the present invention is divided into an acquisition unit that acquires a sentence from a manual, a division unit that divides the sentence acquired by the acquisition unit into each element of the subject , the sentence, and the end of the sentence, and the division unit. A question generation unit that generates a question sentence using the elements, an answer generation unit that generates an answer corresponding to the question sentence generated by the question generation unit using the elements divided by the division unit, and the above . The division unit includes a determination unit for determining the type of sentence end divided by the division unit, and a selection unit for selecting a question template and an answer template according to the type of sentence end determined by the determination unit . If there is a predetermined character string in one sentence, divides the sentence from immediately after the corresponding character string to the end of the sentence as the end of the sentence, and the question generation unit is the question selected by the selection unit. The answer generation unit is a document creation support device that generates a question sentence using a template and generates an answer sentence using the answer template selected by the selection unit .

One aspect of the present invention is the document creation support device, and the predetermined character string is a character string including a character string indicating a time .

One aspect of the present invention is the document creation support device, wherein the predetermined character string is "for", "to", "when", or "sometimes". ..

One aspect of the present invention is the document creation support device, wherein the predetermined character string includes "if," or "if," in one sentence, and "is possible." Or, if "necessary" is included, "possible" or "necessary" .

One aspect of the present invention is the above-mentioned document creation support device, and the types of sentence endings include "because", "denial", "confirmation", "negative word", "case (necessary)", and "necessary". , "Impossible", "Date and time", and "Case (possible)", and the types of sentence endings are "Because", "Negative", "Confirm", "Negative word", and "Case (necessary)". , "Necessary", "impossible", "date and time", and "case (possible)". The question words contained in the question are automatically selected by BERT .

One aspect of the present invention is an acquisition stage in which a document creation support device acquires a sentence from a manual, and a sentence acquired by the document creation support device in the acquisition stage is used as a subject , in a sentence, and at the end of a sentence. The division stage to be divided, the question generation stage in which the document creation support device generates a question sentence using the elements divided by the division step, and the document creation support device to the question sentence generated by the question generation step. A response generation stage in which the corresponding answer is generated using the elements divided by the division stage, a determination stage in which the document creation support device determines the type of the sentence end divided by the division stage, and a document creation support device. However, the division stage includes a question template and a selection stage for selecting an answer template according to the type of sentence end determined by the determination stage, and the division stage corresponds to a predetermined character string in one sentence. If there is something, it is divided from immediately after the corresponding character string to the end of the sentence as the end of the sentence, and the question generation stage generates a question sentence using the question template selected by the selection stage, and the answer generation stage. Is a document creation support method for generating an answer sentence using the answer template selected by the selection step .

One aspect of the present invention is a program that causes a computer to function as a document creation support device, wherein the computer has an acquisition unit for acquiring a sentence from a manual and a sentence acquired by the acquisition unit as a subject , in a sentence, and so on. And the division unit that divides into each element at the end of the sentence, the question generation unit that generates a question sentence using the elements divided by the division unit, and the answer corresponding to the question sentence generated by the question generation unit. Question template according to the answer generation unit generated by using the elements divided by the division unit, the judgment unit that determines the type of the sentence end divided by the division unit, and the type of the sentence end determined by the determination unit. And, as a selection part to select the answer template, and the division part, if there is a thing corresponding to the predetermined character string in one sentence, from immediately after the corresponding character string to the end of the sentence. Is divided as the end of the sentence, the question generation unit generates a question sentence using the question template selected by the selection unit, and the answer generation unit uses the answer template selected by the selection unit to generate the answer sentence. Is a program that generates .

Claims (7)

The acquisition department that acquires a sentence from the manual,
A division unit that divides a sentence acquired by the acquisition unit into each element of the subject, the sentence, and the end of the sentence according to a predetermined rule.
A question generation unit that generates a question sentence using the elements divided by the division unit, and a question generation unit.
An answer generation unit that generates an answer corresponding to a question sentence generated by the question generation unit using the elements divided by the division unit, and an answer generation unit.
Document creation support device equipped with. The predetermined rule divides a sentence by using the character string having the highest score among the character strings included in the sentence by using the correspondence information in which the character string and the score corresponding to the character string are associated with each other. The document creation support device according to claim 1, which is a rule. A determination unit that determines the type of sentence end divided by the division unit, and
A selection unit for selecting a question template and an answer template according to the type of sentence end determined by the determination unit, and a selection unit.
The document creation support device according to claim 1 or 2, comprising the above. The document creation support device according to claim 3, wherein the question generation unit generates a question sentence using the question template selected by the selection unit. The document creation support device according to claim 3, wherein the answer generation unit generates an answer sentence using the answer template selected by the selection unit. The acquisition stage where the document creation support device acquires a sentence from the manual,
A division step in which the document creation support device divides the sentence acquired in the acquisition step into each element of the subject, the sentence, and the end of the sentence according to a predetermined rule.
A question generation stage in which the document creation support device generates a question sentence using the elements divided by the division stage, and
An answer generation step in which the document creation support device generates an answer corresponding to the question sentence generated by the question generation step using the elements divided by the division step.
Document creation support method equipped with. A program that makes a computer function as a document creation support device.
The computer
The acquisition department that acquires a sentence from the manual,
A division unit that divides a sentence acquired by the acquisition unit into each element of the subject, the sentence, and the end of the sentence according to a predetermined rule.
A question generation unit that generates a question sentence using the elements divided by the division unit, and a question generation unit.
A program that functions as an answer generation unit that generates an answer corresponding to a question sentence generated by the question generation unit by using the elements divided by the division unit.

Images