CN114930377A - Report pen holding auxiliary system and report pen holding auxiliary method - Google Patents

Abstract

There are: a document model decision unit that decides a document model storing a template of a copy to be presented to the user based on an answer to the selective question and a question decision model that decides a further question for the answer; a template application unit that determines a document model based on the decision. the document model of the document model and the answer or further answer, perform a descriptive question for questioning the content that is insufficient in the template of the copy stored in the document model, and apply the answer to the descriptive question to the template of the copy; and a prompt processing unit , which presents a template to which a copy of an answer to a descriptive question is applied as the copy to be prompted to the user.

Description

Report writing assistance system, report writing assistance method

technical field

The present invention relates to a report writing assistant system and report writing assistant method for assisting report writing.

Background technique

Facts and rules of thumb confirmed by practitioners in various research and development and investigations. Such tacit knowledge is extremely valuable information for the employer who employs the practitioner, and it is desired to be shared among a plurality of practitioners not limited to the practitioner. For this purpose, countermeasures such as creating and sharing reports are often adopted. However, in the writing of the report, there is a deviation in the quality of the information transmission based on the writing skills of the writer. That is, it is difficult to acquire the writing skills of conveying information to readers more or less, resulting in omissions and misunderstandings.

In addition, as another problem, there is the problem of labor required for writing. It is generally difficult to start writing from the state of blank paper, and it takes time. Therefore, in many cases, a report template is prepared in advance, and writing work is performed in a format corresponding to this. However, in this approach the template can only utilize a few patterns. This is because when there are too many templates, the user cannot judge which template should be selected from among them. Therefore, several templates are generally prepared by field. For example, a template for an experiment report may be prepared, but detailed responses such as using a different template depending on the content and result of the experiment cannot be performed. If only a specific template is selected from a plurality of templates, a method such as document retrieval can be applied, but this method requires that the user can sufficiently grasp the prepared template and the user himself must be able to determine which template to use, which is very difficult of.

prior art literature

Patent Literature

Patent Document 1: Japanese Patent Laid-Open No. 2008-262432

Patent Document 2: Japanese Patent Laid-Open No. 2004-102818

SUMMARY OF THE INVENTION

The problem to be solved by the invention

Patent Document 1 discloses a technique of generating a report by inputting a problem case into a table and applying it to a template. However, in this method, an appropriate template cannot be selected and used from a variety of templates, and it cannot be used unless the content to be reported cannot be clarified.

Patent Document 2 discloses a technique for retrieving question texts using conceptual keywords, thereby facilitating maintenance of answer data for questions. According to this method, the template can be conceptually searched, but it is not easy for the user to determine the template to be used.

The present invention has been made in consideration of such a situation, and an object of the present invention is to provide a report writing assistance system and a report that can easily reflect information memorized by a user on a document without much effort. Writing helper methods.

means of solving problems

A report writing assistance system according to an aspect of the present invention is characterized by comprising: a document model determination unit that determines, based on an answer to a selective question and a question determination model that determines further questions to the answer, a document to be stored for the user The document model of the template of the prompted copy; the template application part, which according to the determined document model and the answer or the answer to the further question, carries out the question for the question stored in the document model. a descriptive question for the content that is insufficient in the copy template, and applying an answer to the descriptive question to the copy template; and a prompt processing unit that uses the copy template to which the answer to the descriptive question is applied as a template The text prompted to the user is prompted.

Invention effect

According to one aspect of the present invention, it is possible to reflect only the information memorized by the user on the document without much effort.

Description of drawings

FIG. 1 is an example of a schematic configuration of the first embodiment.

FIG. 2 is an example of the structure of the physical installation of Embodiment 1. FIG.

FIG. 3 is an example of the utilization process of the first embodiment.

FIG. 4 is an example of an initial screen of the first embodiment.

FIG. 5 is an example of the initial question-and-answer flow of the first embodiment.

FIG. 6 is an example of user input data of the user handover according to the first embodiment.

FIG. 7 is an example of user data in the user information DB of the first embodiment.

FIG. 8 is an example of a question generation request in Embodiment 1. FIG.

9 is an example of selective question data in the selective question DB of the first embodiment.

10 is an example of a question decision model of the question and answer stage decision model of the first embodiment.

FIG. 11 is a schematic diagram of the question-and-answer phase decision in Example 1. FIG.

FIG. 12 is an example of question text data in the first embodiment.

FIG. 13 is an example of a question screen related to the selective question of the first embodiment.

FIG. 14 is an example of the flow of question and answer processing in the first embodiment.

FIG. 15 is an example of the answer data of the first embodiment.

FIG. 16 is an example of answer history data in the first embodiment.

FIG. 17 is an example of a copy creation status determination process performed by the question control unit of the first embodiment.

18 is an example of document model data in the document model DB of the first embodiment.

FIG. 19 is an example of document model data of the first embodiment.

FIG. 20 is an example of question generation processing in the first embodiment.

21 is an example of descriptive question data in the descriptive question DB of the first embodiment.

FIG. 22 is an example of a question screen related to the selective question of the first embodiment.

FIG. 23 is an example of a copy creation request in Embodiment 1. FIG.

FIG. 24 is an example of the copy generation process of the first embodiment.

FIG. 25 is an example of the copy data of the first embodiment.

FIG. 26 is an example of the text display of the first embodiment.

FIG. 27 is an example of a schematic configuration of the second embodiment.

FIG. 28 is an example of a question decision model of the second embodiment.

FIG. 29 is a schematic diagram of the question and answer decision stage in Embodiment 2. FIG.

FIG. 30 is an example of a question screen related to a choice question determined by the document model of the second embodiment.

FIG. 31 is an example of question and answer processing in the second embodiment.

Detailed ways

Example 1

FIG. 1 shows the configuration of a first embodiment of a document preparation assisting device as an example of a report writing assisting system and a report writing assisting method according to the present invention. The document creation assisting device (101) of the present embodiment is a device that accepts the operation of the user (102), and is a system including the following parts: an input/output accepting unit (110) that accepts input from the user; a user information DB ( 111), which records the utilization status of the user; the question control unit (112), which controls the overall flow of copywriting generation; the answer record DB (113), which saves the answers obtained from the user during the copywriting generation process; A document model DB (114), which manages the information related to the category of the document to be generated and the copy template of the document in correspondence; a copy generation part (115), which combines the answer record DB (113) and the document model DB ( 114) to generate a copy; a question-and-answer stage decision model (116), which holds information for determining a document model and evaluating the degree of this determination; a selective question generation section (117), which generates a Selective questions; Selective question DB (118), which holds templates of selective questions; Descriptive question generation section (119), which generates descriptive questions; and Descriptive question DB (120), which holds descriptive questions A template of a descriptive question to be generated by the question generation unit (119). In addition, in the figure, solid-line arrows indicate the flow of processing, and broken-line arrows indicate the flow of data.

FIG. 2 shows an example of the structure of the physical installation of the first embodiment. The document creation assisting device (101) can be realized using a general computer. That is, it is a device including a processor (201) with computing performance, a DRAM (202), a volatile temporary storage area capable of high-speed reading and writing, and a storage area, which is a permanent storage area using HDD, flash memory, and the like A device (203), an input device (204) such as a mouse and a keyboard for operation, a monitor (205) for showing actions to the user, and a serial port for communicating with external computers and devices etc. interface (206). The input/output reception unit (110), the question control unit (112), the copy generation unit (115), the selective question generation unit (117), and the descriptive question generation unit (119) can execute the recording and storage by the processor (201). The program in the device (203) is implemented. User information DB ( 111 ), answer record DB ( 113 ), document model DB ( 114 ), question-and-answer stage decision model ( 116 ), selective question DB ( 118 ), and descriptive question DB ( 120 ) can be processed by the processor (201) It is realized by executing a program for accumulating data in the storage device (213). Hereinafter, each part of the report writing assistance system is installed as hardware in the document preparation assistance device (101) which is a general computer, but all or part of them may be distributed and installed in one or more computers such as a cloud. The same function is achieved by communicating with each other.

FIG. 3 schematically shows the operation process of this embodiment. When the document creation assisting device (101) is activated, an initial screen (301) is displayed, and the access by the user (102) is awaited. When the user (102) inputs to the document creation support device (101) and hands over the user input data (302), the document creation support device (101) executes the initial question and answer process (303) to start a question and answer with the user. This initial process is called an initial stage.

When the initial stage is completed, the document creation assisting device (101) proceeds to a document information collection stage in which information of the document to be created is collected while questioning and answering with the user. At this stage, in the question-and-answer process (305) executed by the document creation support device (101), the question to be asked next is determined and presented to the user (102) as a question screen (304). When the user ( 102 ) inputs and sends the answer information ( 306 ) in the form of an answer, the question-and-answer process ( 305 ) is performed again to generate the next question. Through this repetition, the document to be created by the user (102) is gradually clarified.

When the draft of the document to be written by the user can be sufficiently generated, the operation of the document creation assisting device (101) proceeds to a copy reading stage in which the copy is presented to the user. When entering the copy reading stage, the result returned by the document creation assisting device (101) executes a copy prompt process (307) including the created copy, and the user is prompted to display the copy easily observably (308). Through this process, the user can passively create the copy of the document in response to the inquiries of the document creation assisting device (101), and therefore, it is possible to reduce the time and effort required to create the document from a blank state. Hereinafter, each of them will be described in detail.

FIG. 4 shows an example of an initial screen ( 301 ) generated when the document creation assisting device ( 101 ) is activated. The initial screen (301) is a screen for specifying user information. In this example, HTML data that can be created by a known Web browser (401) or the like is sent as the initial screen. This screen includes a text box ( 402 ) for acquiring the user's name and a send button ( 403 ) for prompting the start of use upon completion of the input, and the user ( 102 ) can freely use his or her name, nickname, and the like. In addition, the initial screen ( 301 ) may be in an output format capable of responding to the user ( 102 ), and may be in a format in which a speaker or a microphone conducts voice dialogue, for example.

When the user (102) operates the initial screen (301) and presses the send button (403), the user input data (302) is sent to the document creation support device (101), and the document creation support device (101) An initial question and answer process is performed (303). FIG. 5 shows an example of the flow of the initial question and answer process (303). In the initial question-and-answer process (303), first, the input/output reception unit (110) receives user input data (302), and executes the process of reflecting it in the user information DB (111) (S501).

FIG. 6 shows an example of the content of the user input data (302). The user input data ( 302 ) includes a user ID ( 601 ), which is an ID for identifying the user, and a user name ( 602 ) indicating the display name of the user. Here, the user ID ( 601 ) may be blank if it is not determined, or separately designated information (eg, employee number, etc.) that can identify the user may be used. The user name (602) is acquired from the text box of the initial screen (301).

This user input data (302) is added to the user data (503) of the user DB (111). FIG. 7 shows details of an example of the user data (503). The user data ( 503 ) includes a user ID ( 701 ) storing information contained in the user input data ( 302 ), a user name ( 702 ), data ( 703 ) indicating which stage of the question and answer stage it is currently in, and a user ID ( 703 ) indicating the use The document model ID (704) of the estimation result of the type of the document that the person (102) wants to write. The input/output reception unit (110) for the user ID (701), when the user ID (601) of the user input data (302) is blank, the user ID for the existing user data (503) Numbers that do not exist in (701) are numbered. For example, a value obtained by adding 1 to the maximum value of the user ID (701) can be used. In addition, in the initial question-and-answer process (303), the question-and-answer phase (703) is a document model determination phase, and values corresponding thereto are stored.

When the user DB (111) is updated, the question control unit (112) generates a question generation request (505) (S504). FIG. 8 shows data items included in the question generation request (505). The question generation request ( 505 ) excluding the user ID ( 601 ) included in the user data ( 503 ) of the user DB ( 111 ), the user ID ( 801 ) storing the user name ( 602 ), the user name ( 802 ) ), there is also answer history data (803), which is history data of what kind of answer the user has made so far. In the initial question-and-answer process (303), the answer history data (803) is blank, but if there is an answer obtained in another form, it may be added here.

Next, the question control unit (112) acquires the question determination model (507) from the question and answer stage determination model (116) (S506), and sends it to the selective question generation unit (117) together with the user data (503). Then, the selective question generation unit (117) refers to the question determination model (507) to determine the next question, that is, the important question (S508), and acquires the selective question data corresponding to the important question from the selective question DB (118). (510), generating question text data (511) (S509).

In order to explain this process in detail, first, the items included in the selective question data ( 510 ) of the selective question DB ( 118 ) are shown in FIG. 9 . The selective question data (510) includes a question ID (901), which is an identifier of the question data, a question text (902), which is an article presented to the user (102), a list of options, and a list of option codes that are code values thereof. The option list (903) is kept correspondingly.

Next, an example of the question determination model (507) is shown in FIG. 10 . The question decision model ( 507 ) in the first embodiment has a structure similar to that of a known decision tree, and is a combination of items of a plurality of branches ( 1001 ) ( 1002 ) and results ( 1003 ). Each branch has a branch ID (1004) as an identifier of the branch, a question ID (1005) of the question corresponding to the branch, and a destination (1006) by answer indicating the next branch according to the answer to the question. elements. One of the branches corresponds to one record of the selective question data ( 510 ), and can be associated by comparing the question ID ( 1005 ) of the branch and the question ID ( 901 ) of the selective question data ( 510 ). In addition, the item of the result has: a branch ID (907), which has the same hierarchy as the branched item; and a document model ID (908), which indicates the estimated result of the category of the document. This structure is a decision tree in which each branch can be stacked, and the last result of tracing the branch of the decision tree from branch 1 (901) is the document model ID of the document model to be used.

FIG. 11 schematically shows the principle of document model determination using this question determination model (507). The table on the upper side of the figure is the data that forms the basis of the question decision model (507). In order to create a question decision model (507), a plurality of examples of documents to be supported by the document creation support device (101) are collected in advance, and the selective question data ( 510) The table obtained by the corresponding answer is this table. Each row of the table corresponds to the collected documents. The column (1101) of the question ID corresponds to the question ID (901) of the selective question data (510), and the numbers entered in each column are the option codes in the option list (903) of the selective question data (510). value. In addition, the column (1102) of the document model ID is an identifier of the document model indicating which document model is suitable for each document.

The question determination model ( 507 ) can be generated by applying a machine learning algorithm to the table by setting the column ( 1101 ) of question IDs as explanatory variables and the column ( 1102 ) of document model IDs as target variables. The tree-like structure in the lower part of the figure is a structure in which the data of the question determination model (705) is arranged so as to be easily observed. Knowing that this structure is the same as that of a well-known decision tree, it can be constructed by learning the above table to most smoothly classify the column of document model ID (1102). The top item (1103) means the question ID of the question confirmed first, and if it is answered by the option of 1, it is in the form of the question (1104) to the left, and in the case of 2, it is The right question (1105) advances the form. Therefore, when the question ID: 1 is answered 1, and the question ID: 3 is answered 2, the document model ID: 1 (1106) is determined as the document model data to be used. When the question decision model ( 507 ) can be traced to reach the document model, it can be judged that it can be confirmed; otherwise, it can be judged that it cannot be confirmed.

In addition, as the question determination model (507), even a model other than a decision tree can be used as long as it is a model of discriminant analysis for predicting the classification of target variables from explanatory variables. In addition, by using a model that can be discriminated even when only a part of the target variable is obtained, such as a decision tree, such as the well-known Gaussian process method, the document model can be determined without answering all the questions. As a result, it is possible to reduce the number of The advantage of the number of questions asked. For example, as the question determination model (507), even when no answer to the multiple-choice question is obtained, the multiple-choice question generation unit (117) can use the Gaussian process method to estimate the answer to the multiple-choice question, and from the multiple-choice question In DB (118), the next choice question with the highest reliability among the estimated answers is selected, and the document model corresponding to the answer to the selected question is reached.

The multiple choice question generation unit (117) uses the question decision model (507) to determine the multiple choice question to be asked next. Specifically, the branch that can be traced deepest in the question determination model (507) is searched, and the question corresponding to the question ID of the branch is regarded as an important question. In the example of FIG. 11 , when answering 2 to the question with question ID: 1, the important question is question ID: 4. In addition, even when a model other than a decision tree is used, it is possible to determine and decide which question is to be answered next, thereby facilitating the identification of the document model. For example, when a model capable of evaluating the accuracy of discrimination is used, conditions can be investigated while virtually giving an answer to the next question, and the question with the highest accuracy can be set as an important question.

FIG. 12 shows elements included in question text data ( 511 ) generated from selective question data ( 510 ). Question text data ( 511 ) includes a user ID ( 1201 ) of the user, a question ID ( 1202 ) as an identifier of the question, a question type ( 1203 ) indicating the type of question, and a question text presented to the user ( 102 ) (1204), answer information (1205) indicating an item to be answered by the user (102).

In the initial question and answer process (303), the selective question generation unit (117) stores the user ID included in the user data (503) as the user ID (1201) for each element of the question text data (511). (601), the question ID (901) of the question data (510) is stored as the question ID (1202), the code value representing the alternative question is stored in the question category (1203), and the alternative question data (510) is stored as the question text (1204). ) of the question text (902), and the option list (903) is stored in the answer message (1205).

Then, the input/output reception unit (110) generates a question screen (304) based on the question text data (511) and presents it to the user (102) (S512). An example of the question screen ( 304 ) is shown in FIG. 13 . On this screen, a question text ( 1301 ), a radio button input ( 1302 ) for selectively inputting an answer, and an answer button ( 1303 ) for transmitting the answer result are displayed. In this example, the user ( 102 ) selects an answer by inputting a radio button ( 1302 ) and presses the answer button ( 1303 ), whereby the answer can be transmitted. At this time, it should be noted that the user ID ( 1201 ) and the question ID ( 1202 ) that are not displayed on the screen also need to be held. In addition, the user name (802) etc. may be displayed on this screen (in this example, the user name "Ozawa" is used).

With the above, the initial stage is completed, and after that, the document information collection stage in which the copy of the document is created by the interaction between the user ( 102 ) and the document creation support device ( 101 ) is entered. In the document information collection and creation stage, when the user ( 102 ) responds to the question screen ( 304 ), the document creation support device ( 101 ) executes the question and answer process ( 305 ) to generate the next question screen ( 304 ). repetitive form.

An example of the flow of the question and answer process (305) is shown in FIG. 14 . The question and answer process ( 305 ) is a process of accepting information from the user ( 102 ) and returning questions and the like to it, similarly to the initial question and answer process ( 303 ). First, the input/output reception unit (110) acquires the response data (1402) of the user (102) in response to the question screen (304) (S1401). FIG. 15 shows an example of the elements included in the answer data (1402). The answer data (1402) includes a user ID (1201) for generating the question text data (511) of the question screen (304), a question ID (1202), and a user ID (1501) similar to the question type (1203). ), question ID ( 1502 ), question type ( 1503 ), as well as answer content ( 1504 ), which is an answer input by the user ( 102 ) on the question screen ( 304 ). When the answer content ( 1504 ) is an answer to the multiple-choice question as shown in FIG. 13 , the option code of the selected option is entered through the radio button input ( 1302 ) etc. narrate). Next, the input/output reception unit (110) uses the user ID (1501) of the response data (1402) to search for the user ID (701) and the response data from the user data (503) in the user information DB (111). ( 1402 ) is matched with the user ID ( 1501 ) ( S1403 ). Thereby, the record of the user data (503) corresponding to the present user (102) can be acquired.

Next, the input/output accepting unit (110) sends the information of the answer data (1402) to the answer record DB (113), and adds it as answer history data (1405). The answer history data (1405) is data that stores a plurality of answer data (1402). As shown in FIG. 16, the constituent elements of the data may be the same as those of the answer data (1402). User ID (1601), question ID (1602), The question type ( 1603 ) and the answer content ( 1604 ) correspond to the user ID ( 1501 ), the question ID ( 1502 ), the question type ( 1503 ), and the answer content ( 1504 ), respectively.

When the addition of the answer history data (1104) is completed, the input/output acceptance unit (110) sends the information of the user data (503) to the question control unit (112). The question control unit (112) refers to the user ID (901) included in the user data (503), and obtains the user ID (1601) in the answer history data (1405) and the answer history data (1405) matching it. group, and executes a copy creation status determination process (S1406), which is a process for determining the current copy creation status. Then, based on the result, it is determined whether or not copy creation can be performed (S1407). When the question control unit ( 112 ) determines that it is not possible ( S1407 ; NO), it generates a question generation request ( 505 ) ( S1408 ) and executes the question generation process ( S1409 ). This question generation request ( 505 ) is the same as the case where its elements are shown in FIG. 8 , and the user ID ( 801 ) and the user name ( 802 ) are stored as answer history data by storing a set of the answer data ( 1405 ). (803) The user ID (701) and the user name (702) of the user data (503) are generated. On the other hand, if the question control unit (112) determines that copy creation is possible (S1407; YES), it generates a copy creation request (1411) (S1410), and executes the creation by sending it to the copy creation unit (115). The processing of the copy is the copy generation process (S1412). Finally, the input/output reception unit (110) receives the question text data (511) obtained as a result of the question generation process (S1409) or the text generation process (S1412), and updates the content of the user information DB (111) (S1413). , and present it to the user (102) (S1414).

FIG. 17 shows an example of a detailed flow of the copy generation status determination process ( S1406 ). In this flow, the question control unit (112) performs processing using the user data (503) and the answer history data (1405) acquired from the answer history DB (114) (S1701). The question control unit (112) first confirms the question and answer phase (703) entered in the user data (503), and determines whether or not the "text viewing phase" indicating that the copy can be created is set (S1702). The question control unit ( 112 ), when it is determined that the “copy reading stage” is set ( S1702 ; YES), determines that the copy creation can be performed. When it is determined that the "text viewing stage" is not set (S1702; NO), the question control unit (112) further determines whether the question and answer stage (703) is set to the "document model determination stage" (S1703). When the question control unit (112) determines that the "document model determination stage" is set (S1703; YES), it indicates that which template should be used has not yet been determined, and therefore determines that copy creation cannot be performed (S1703). When the question control unit (112) determines that the "document model determination stage" is not set (S1703; NO), it indicates that the present process is in the "document information collection stage". At this time, the value of the user data (503) should be stored in the document model ID (702). Therefore, the question control unit (112) uses it to obtain the corresponding document model data (1705) from the document model DB (114), thereby further making a determination (S1704).

FIG. 18 shows a data item of one record of the document model data (1705) stored in the document model DB (114). The document model data (1705) has a document model ID (1801) as an identifier of the record, a copy template (1802) representing the template of the document corresponding to the document model, and a list as a list for obtaining the template as a document. A descriptive question item list (1803) of the question ID of the required keyword is completed. In addition, the question ID included in the descriptive question item list (1803) corresponds to the identifier for identifying the question in the selective question DB (118) and the descriptive question DB (120), and also corresponds to the question text data ( 511) corresponds to the question ID (1202) contained in the answer data (1302) and the question ID (1502) contained in the answer data (1302). In addition, the document model ID (1801) corresponds to the document model ID (704) of the user data (503) and the document model ID (908) of the question decision model (507).

A specific example of ( 1705 ) of document model data used here is shown in FIG. 19 . In this example, the document model ID ( 1901 ) is a serial number of an integer value, and the part described with square brackets ({}) in the copy template ( 1902 ). An integer value is also written in square brackets, and the integer value corresponds to the question ID. The descriptive question list (1903) stores the contents of the question IDs in the square brackets that appear in the copy template (1902).

The question control unit (112) compares the document model ID (704) possessed by the user data (503) with the document model ID (1801) possessed by the document model data (1705), identifies matching records, and acquires a descriptive question list (1803). On the other hand, the answer history data ( 1405 ) also includes the question ID ( 1602 ) that has been answered. The question control unit (112) compares the two to determine whether there is a question ID of a record included in the descriptive question list (1803) and not answered in the answer history data (1405), that is, the descriptive question list Whether all the questions included in (1803) have been answered (S1705). When it is determined that all the questions included in the descriptive question list (1803) have been answered (S1705; YES), the question control unit (112) determines that copy creation is possible, and outputs the meaning (S1706). On the other hand, when the question control unit (112) determines that all questions included in the descriptive question item list (1803) have not been answered (S1705; NO), there are questions that have not been answered, and therefore, it is determined that the question cannot be answered. Copywriting is produced, and the meaning is output (S1707).

An example of the detailed processing flow of the question generation process ( S1409 ) executed in the question and answer process ( 305 ) is shown in FIG. 20 . The question generation process ( S1409 ) is a process of generating a question for specifying the type of the document to be written by the user ( 102 ), and a question for inquiring about items required to complete the template of the type.

First, the question control unit ( 112 ) compares the answer history data ( 1405 ) with the question determination model ( 507 ) included in the question and answer stage determination model ( 116 ), and obtains document model data ( 1705 ) estimated to be currently used ( S2001 ). This process is performed according to the principle of the question determination model (507) shown in FIG. 11 . That is, the result reached last by tracing the branch of the decision tree from branch 1 (901) matches the document model ID (1801) of the document model data (1705) to be used. The question control unit ( 112 ) can determine that the document model can be identified when the question determination model ( 507 ) can reach the document model ID, and can determine that the document model cannot be identified otherwise.

The question control unit (112) determines whether or not the document model can be identified, and if the document model cannot be identified (S2002; NO), the alternative question determination unit (118) determines the alternative question to be asked next. On the other hand, when it is determined that the document model can be specified (S2002; YES), the question control unit (112) causes the descriptive question determination unit (120) to generate a question text for acquiring information insufficient to create a text.

When the process is transferred to the selective question generation unit (117), the selective question generation unit (117) compares the answer history data (1405) with the question determination model (507), and decides the next question, that is, the important question ( S2003). This process can be performed similarly to the process ( S508 ) of identifying important questions in the initial question and answer process ( 303 ). Next, the multiple choice question generation unit (117) generates question text data (511). Similar to the process ( S509 ) in the initial question and answer process ( 303 ), selective question data ( 510 ) of important questions are acquired from the selective question DB ( 118 ), and question text data ( 511 ) are generated ( S2004 ).

On the other hand, the case where it is determined that the document model can be specified and the process is shifted to the descriptive question generation unit ( 119 ) will be described. First, the descriptive question generation unit (119) compares the answer history data (1405) with the document model data (1705), and determines the question to be performed next. Specifically, the descriptive question generation unit (119) compares the list of question IDs in the descriptive question item list (1803) of the document model data (1705) with the question ID (1602) of the answer history data (1405), The question IDs that are not included in the answer history data (1405) but are included in the descriptive question item list (1803) are investigated. The descriptive question generation unit (119) arranges such question IDs in a predetermined order, for example, in ascending order of question IDs, and sets the smallest question ID as the next important question (S2005).

Next, the descriptive question generation unit (119) acquires descriptive question data (2007) corresponding to the question ID of the important question from the descriptive question DB (120) (S2006). Fig. 21 shows the data included in the descriptive question data (2007). The descriptive question data (2007) contains the question ID (2101) as the identifier of the question data, the question text (2102) as the text presented to the user (102), and the answer item name ( 2103) correspond to hold. The descriptive question generation unit (119) acquires descriptive question data (2007) corresponding to the question ID of the important question, and generates question text data (511). The descriptive question generation unit (119) stores the user ID (701) included in the user data (503) as the user ID (1201) of the question text data (511), and stores the descriptive question data as the question ID (1202). (2007) question ID (2101), question type (1203) stores a code value representing a descriptive question, as question text (1204) stores a question text (2102) of descriptive question data (2007), and in answer information (1205) ) to store the answer item name (2103). Thereby, question text data can be generated (511).

At the end of the question generation process ( S505 ), the question control unit ( 112 ) returns the question text data ( 511 ) and the document model ID to the input/output accepting unit ( 110 ) together with the document model ID if the document model is identified. , thereby ending (S2008). Then, in ( S1413 ) shown in FIG. 14 , if the descriptive question generation unit ( 119 ) operates, the input/output acceptance unit ( 110 ) executes the question-and-answer phase ( 703 ) of the user data ( 503 ). Updating to the document information collection phase, the document model ID (704) is updated to its resulting document model ID. In addition, the input/output accepting unit (110) sets the question-and-answer phase (703) as a copy-reading phase when the copy generation unit (115) operates.

After that, the input/output reception unit (110) presents the text to the user (102). When the question type ( 1203 ) of the question text data ( 511 ) is a selective question, a question screen ( 304 ) as shown in FIG. 13 is displayed. When the question type ( 1203 ) is a descriptive question, a question screen ( 304 ) as shown in FIG. 22 is displayed. On this screen, a question text ( 2201 ), a text box ( 2202 ) corresponding to the answer item name ( 2103 ) and in which an answer can be input, and an answer button ( 2203 ) for transmitting the answer result are displayed. In this example, the user ( 102 ) inputs an answer into the text box ( 2202 ) and presses the answer button ( 2203 ), whereby the answer can be sent. At this time, it should be noted that the user ID ( 601 ) and the question ID ( 602 ) that are not displayed on the screen also need to be held. In addition, the user name ( 602 ) and the like may be displayed on this screen.

As described above, when the question-and-answer process ( 305 ) in the document information collection stage is repeatedly executed, the result that the copy-writing status determination process ( S1406 ) in FIG. 14 can generate a copy is always obtained. At this time, the copy creation request (1411) is sent to the copy generation unit (115), and the copy generation process is executed (S1412). The state in which this copy generation process ( S1412 ) has been performed is the copy reading stage. FIG. 23 shows the information included in the copy creation request (1411). The copy creation request (1411) has a user ID (2301) based on the user ID (701) of the user data (503), and a user name (702) based on the user name (702) of the user data (503). (2302) Data of the document ID (2303) determined in the question and answer process (305).

An example of the flow of the copy generation process (S1412) is shown in FIG. 24 . In the copy generation process (S1412), the copy generation unit (115) generates copy based on the answer history data (1405) and the document model data (1705). Therefore, based on the user ID (2301), the copy generation unit (115) selects and acquires data matching the user ID (1601) from the answer history data (1405) of the answer record DB (113), In (114), the document model data (1705) whose document model ID (1801) matches the document ID (2303) of the copy creation request (1411) is acquired (S2401). Next, the copy generation unit (115) uses the document template (1802) of the acquired document model data (1705) to insert the answer content (1604) of the answer history data (1405) into the portion corresponding to the question ID in the square brackets of the template. (S2402). Finally, the copy generation unit (115) generates copy data (2404) by adding information such as the user name to a necessary range (S2403). FIG. 25 shows the information contained in the generated copy data (2404). The copy data (2404) has a user ID (2501) based on the user ID (2301) of the copy creation request (1411), and a user name (2502) based on the user name (2302) as well. The copy text (2503) generated in the processing of the template (S2403).

The copy data ( 2404 ) is passed to the input/output accepting unit ( 110 ) after the copy generation process ( S1412 ) is completed, and is used for generation of the copy display ( 308 ). An example of the information presentation based on the copy display (308) is shown in FIG. 26 . On this screen ( 2601 ), the generated copy ( 2602 ) and a completion button ( 2603 ) for ending the display are displayed. The displayed copy is obtained by filling the necessary items in the template selected based on the user's answer, and is considered to be very close to the report text that the user should write. When the completion button ( 2603 ) is pressed, the input/output accepting unit ( 110 ) generates a text file in which the screen display is completed and the copy is entered. In addition, the copy and the like can also be copied and pasted by a known method.

According to the above-mentioned embodiment, when documenting a report such as an experimental investigation, it is possible to write in a short time without hesitating what kind of document should be written. The copy does not have to be complete, but it is possible to make a report sufficient for full use to the extent that corrections are applied. Therefore, the business practitioner can frequently write reports without wasting time, thereby suppressing the loss of empirical insights.

For example, there is a document model determination unit (for example, the selective question generation unit 117, the question control unit 112, S2001 in FIG. 20) that determines a model based on the answer to the selective question and the question that determines further questions for the answer (for example, the question decision model (507)), which determines the document model (for example, the document model data (1705)) in which the template (for example, the copy template (1902)) of the copy presented to the user is stored; the template application unit (for example, The descriptive question generation unit 119, the question control unit 112, (S2002, S2005, S2006, S2008 in FIG. 20, S1705 in FIG. 17)) based on the determined document model and the above-mentioned answers or answers to the above-mentioned further questions , performing a descriptive question for asking the content that is insufficient in the template of the above-mentioned copy stored in the above-mentioned document model, and applying the answer to the descriptive question to the template of the above-mentioned copy; and a prompt processing unit (for example, a question control section 112, input/output accepting section 110, S1412, S1414 in FIG. 14), which presents a template of a copy to which the answer to the above-mentioned descriptive question is applied as a copy to be presented to the user, so that it is easy to Only the information memorized by the user is reflected in the document. As a result, reports can be produced frequently, and insights that are only memorized by the practitioner and easily lost can be retained as a document and reused.

Further, the document model determination unit repeats a further question to the answer and an answer from the user to the question until the document model is determined. Therefore, a document model suitable for the user can be determined with high accuracy.

Further, the template application unit repeats the descriptive question and the user's answer to the descriptive question until there is no insufficient content in the template of the copy, so that a copy with a high degree of completion can be created.

In addition, the document model determination unit uses the question determination model (for example, the decision tree shown in FIG. 10 ) composed of the branch that selects the selective question based on the answer and the answer to the further question, and combines the The document model corresponding to the answer to the multiple-choice question with the deepest branch is determined as the document model in which the template of the above-mentioned copy is stored. Therefore, the document model to be determined can be efficiently determined.

Further, the document model determination unit estimates an answer to the multiple-choice question using a Gaussian process method, and determines a document model in which the template of the copy is stored using the multiple-choice question to be performed next with the highest reliability among the estimated answers. Therefore, even when there is an unanswered question, the next question for determining the document model can be determined, and the document model can be determined.

In this way, in the present system, the inquiry is performed in the following two stages: (1) Selective inquiry for determining an opinion model is carried out, and closed questions are thrown along the decision tree, thereby determining what kind of opinion to output and (2) descriptive questions as detailed questions for documentation, questions (mainly open-ended questions) that ask the detailed parts of the insights to be output along the determined insight model, and are learned through prior learning. By determining the decision tree, the insight pattern can be accurately and efficiently specified with a small number of questions, and the load on the user can be reduced.

Example 2

FIG. 27 shows the configuration of the second embodiment of the document preparation assisting device, which is another example of the report writing assisting system of the present invention. The document creation assisting device (101) of the second embodiment is different from the configuration of the first embodiment in that it has a function of updating the question-and-answer stage determination model (116), and has a stage determination model updating unit (2701). In addition, the question determination model ( 507 ) of the question and answer stage determination model ( 116 ) can be updated by the stage determination model update unit ( 2701 ).

An example of the data structure of the question determination model (507) is shown in FIG. 28 . This configuration is different from Example 1 in the configuration of the result, and in the figure, the content of the result 1 (903) is different. In FIG. 28, the result is not associated with one document model ID, but is associated with a plurality of document model IDs, and a plurality of document ID tables (2901) containing question IDs are stored. A count table (2902) that counts the document model IDs. In the count table ( 2902 ), a group is stored in which the question ID, the document model ID, and the count indicating the number of times the document model has been selected are associated with each other. Each of these elements corresponds to a question.

FIG. 29 shows a schematic diagram of the question-and-answer stage determination in the second embodiment. In the decision tree for question-and-answer phase determination in the second embodiment, the document ID table (2901) at the end of the tree is a count table (2902) in which question IDs corresponding to document IDs and the count of the number of selections are associated as follows: corresponding form. This count ( 2803 ) is also used in the learning of a known decision tree, but the document creation assisting device ( 101 ) can follow the decision tree for the data ( 2903 ) that is the basis of the question decision model ( 507 ) and arrive at the tree. Finally, the data (2903) is generated by summing up the number of documents for each document model ID. This question ID is associated with a question for confirming the validity of each document model. If there is no data corresponding to the answer history data ( 1405 ) for any question, the question ID with the largest count ( 2803 ) is selected as the important question, and the input/output accepting unit ( 110 ) displays the question based on the important question screen. An example of such a question screen ( 304 ) is shown in FIG. 30 . "Regular report of on-site inspection" in the question text (3001) in the figure is a sentence corresponding to the document model ID, and a yes/no (3002) answer is required.

The answer to the question ID ( 2801 ) is recorded in the answer history data ( 1405 ) as in the first embodiment. When the document creation assisting device (101) reaches the count table (2902) through the same procedure as in the first embodiment in the process determined in the question and answer stage after that, it is checked whether the answer history data (1405) does not match the present count. After the data corresponding to the table (2902), if there is an answer like "Yes", it is the result of determining the article model with the document model ID. If there is only "No", the document creation assisting device (101) sets the question ID with the largest count among the unanswered question IDs as the next important question. With this implementation, the document model can be determined even if the question-and-answer phase decision model (116) cannot correctly determine the document model.

Premised on the described implementation, the question-and-answer phase decision model (116) can be updated. FIG. 31 shows the processing flow of the question and answer process (305) of the second embodiment. The difference from the first embodiment is that after the question control unit (112) generates the question, the question-and-answer phase determination model can be updated by the phase determination model update unit (2701) (S3101). As this update process, the stage determination model update unit (2701) increases the count of the question ID for which "Yes" is selected for the count table (2902), which is the end of the tree reached by tracing the decision tree. Ask the questions in the appropriate order. In addition, the step determination model update unit (2701) adds the user's answer to the basic data table (2903) to reconstruct the decision tree, thereby improving the accuracy of determining the question and answer phase.

However, at the time of this update, it is not limited to have all the answers, and it should be noted that it is necessary to use a well-known algorithm that functions even when data is missing. By adopting the implementation of the second embodiment, the accuracy of the determination process of the question and answer phase can be improved, and the template can be determined more appropriately.

In this way, according to the system of the second embodiment, in the document model determination unit, there are a plurality of document model candidates (for example, a count table (2902) determined in response to an answer to the optional question or an answer to the further question )), a question for confirming the validity of the document model is presented to the user (for example, the question text (3001) shown in FIG. 30), and it is determined that the above-mentioned text is stored based on the answer to the question presented to the user The template's document model. Therefore, even when there are a plurality of document model candidates, the document model to be presented can be reliably specified.

In addition, since there is a question determination model updating unit (for example, the stage determination model updating unit (2701)) that associates the document model determined by the document model determining unit with the question determination model, the determined document can be The model remains in the updated state.

Further, the question determination model updating unit may select the document model from the plurality of document model candidates determined in response to the answer to the optional question or to the further question based on the number of times (for example, stored in a count table) (Count value in (2902)), the document model corresponding to the question determination model described above is determined, so the document model to be presented can be determined based on the selected number of times, that is, the frequency.

Symbol Description

101 Document production aids

110 Input and output reception section

111 User information DB

112 Question Control Department

113 Answer record DB

114 Document Model DB

115 Copywriting Department

116 Question and answer phase decision model

117 Selective Question Generation Department

118 Selected Question DB

119 Descriptive Question Generation Section

120 Descriptive question DB

507 Question Decision Model for Template Decision

701 indicates the user ID of the user's identifier.

Claims (9)

1. A report writing assistant system, characterized in that it has: a document model determination unit that determines a document model in which a template of a copy to be presented to the user is stored, based on the answer to the selective question and the question determination model for determining further questions for the answer; A template application unit that, based on the determined document model and the answer or an answer to the further question, performs description of the content that is insufficient in the template for questioning the copy stored in the document model a descriptive question to which the answer to the descriptive question is applied to a template for said copy; and A presentation processing unit that presents a template of a copy to which the answer to the descriptive question is applied as a copy to be presented to the user. 2. The report writing assistance system according to claim 1, characterized in that: The document model determination unit repeats a further question to the answer and an answer from the user to the question until the document model is determined. 3. The report writing assistance system according to claim 1, characterized in that: The template application unit repeats the descriptive question and the user's answer to the descriptive question until there is no insufficient content in the template of the copy. 4. The report writing assistance system according to claim 1, wherein: The document model determination unit uses the question determination model composed of the branch that selects the alternative question based on the answer and the answer to the further question, and uses the question decision model for the alternative question with the deepest branch. The document model corresponding to the answer is determined as the document model that stores the template of the copy. 5. The report writing assistance system according to claim 1, wherein: The document model determination unit estimates an answer to the optional question using a Gaussian process method, and determines a document in which the template of the copy is stored using the next optional question with the highest reliability among the estimated answers. Model. 6. The report writing assistance system according to claim 1, wherein: The document model determination unit issues a question for confirming the validity of the document model when there are a plurality of document model candidates determined for the answer to the optional question or the answer to the further question The user is prompted, and based on the answer to the question prompted to the user, a document model in which the template of the copy is stored is determined. 7. The report writing assistance system according to claim 1, wherein: The report writing assistance system includes a question determination model update unit that associates the document model determined by the document model determination unit with the question determination model. 8. The report writing assistance system according to claim 7, wherein: The question decision model updating unit decides the number of times a document model is selected from among a plurality of document model candidates determined in response to the answer to the selective question or to the further question The document model corresponding to the model. 9. An auxiliary method for writing a report, characterized in that: The document model determination unit determines the document model in which the template of the copy presented to the user is stored, based on the answer to the selective question and the question determination model for determining further questions in response to the answer, The template application unit performs a descriptive question for questioning the content that is insufficient in the template of the copy stored in the document model, based on the determined document model and the answer or the answer to the further question , applying the answer to the descriptive question to the template of the copy, The presentation processing unit presents a template of the copy to which the answer to the descriptive question is applied as the copy to be presented to the user.

Images