WO2021153403A1

WO2021153403A1 - Text information editing device and text information editing method

Info

Publication number: WO2021153403A1
Application number: PCT/JP2021/001975
Authority: WO
Inventors: 太亮尾崎; 祐太是枝; 皓文森下
Original assignee: 株式会社日立製作所
Priority date: 2020-01-27
Filing date: 2021-01-21
Publication date: 2021-08-05
Also published as: JP2021117759A

Abstract

A text information editing device is connected to a display device, holds dialog text comprising one or more utterance text and editing processing information that indicates at least one of an important expression and an important expression extraction algorithm, refers to the editing processing information and extracts an important expression from the one or more utterance text pieces, displays the dialog text and the extracted important expression on the display device, accepts editing input to at least one of the utterance text included in the dialog text and an annotation to which the utterance text is added, and executes editing that corresponds to the editing input.

Description

Text information editing device and text information editing method

This application claims the priority of Japanese Patent Application No. 2020-010977 filed on January 27, 2020, and incorporates it into this application by referring to its contents.

The present invention relates to a text information editing device and a text information editing method.

It is important in various tasks to refer to and utilize useful information that exists in dialogue texts such as transcriptions of dialogues made by voice such as meetings and discussions and text chats. However, while general sentences and documents are often written with sufficient elaboration assuming the reader in advance, it is important that dialogue is established between the speakers in those dialogue texts. Will be done. As a result, in the dialogue text, useful information is not always described in a clear style and expression, and one useful information is constructed in consideration of dialogues of a plurality of people and past utterances. There are many.

As a background technology in this technical field, there is International Publication No. 2014/20298 (Patent Document 1). In this publication, "As a sentence classification device, a classification target section extraction unit that extracts and processes sections that contribute to sentence classification from input sentences that are textualized conversation contents to be classified based on clue sentences, and a classification target section. A sentence classification unit for determining which classification the input sentence belongs to using the text of the section contributing to the sentence classification extracted by the extraction means is provided. ”(See summary).

International Publication No. 2014/20298

However, in the invention described in Patent Document 1, it is necessary that the classification method and the viewpoint of classification for classifying sentences are defined in advance. Further, the invention described in Patent Document 1 is applicable only to a semi-standard dialogue text such as a response in a call center, and it is difficult to extract useful information from a general dialogue text.

For example, the utterance "I came here on foot because the weather was fine today" comes from the perspective of "the means of transportation of the speaker" and the perspective of "the weather at the place where the utterance was made". Contains information on two perspectives. However, whether one or both of the means of transportation and the weather are useful depends on the perspective of the person using the dialogue text, including this utterance text. Therefore, in the general context, useful information extraction by classification cannot meet all information requests.

Therefore, one aspect of the present invention aims to easily obtain useful information based on an arbitrary viewpoint from the input dialogue text.

In order to solve the above problems, one aspect of the present invention adopts the following configuration. The text information editing device has a processor and a storage device, and is connected to a display device. The storage device includes a dialogue text consisting of one or more spoken texts, and at least one of an important expression and an extraction algorithm of the important expression. The processor retains the editing processing information indicating the above, extracts important expressions from the one or more spoken texts with reference to the editing processing information, and obtains the dialogue text and the extracted important expressions. It is displayed on the display device, accepts edit input for at least one of the utterance text included in the dialogue text and the comment to which the utterance text is attached, and executes the edit corresponding to the edit input.

According to one aspect of the present invention, useful information based on an arbitrary viewpoint can be easily obtained from the input dialogue text.

Issues, configurations and effects other than those described above will be clarified by the explanation of the following embodiments.

It is a block diagram which shows the structural example of the text information editing apparatus in Example 1. FIG. This is an example of dialogue text data output by the dialogue input reception unit in the first embodiment. This is an example of data output by the important expression detection unit in the first embodiment to the interactive editing unit. It is a flowchart which shows an example of the text information editing process in Example 1. FIG. This is an example of a display screen for interactive editing in the first embodiment. This is an example of a display screen for interactive editing in the second embodiment. This is an example of the display screen for interactive editing in the third embodiment. It is a block diagram which shows the structural example of the text information editing apparatus in Example 4. FIG. This is an example of dialogue text data output by the topic boundary identification unit in the fourth embodiment. It is a flowchart which shows an example of the text information editing process in Example 4. This is an example of a display screen for interactive editing in the fourth embodiment. It is an example of the display screen of the interactive editing in Example 5. This is an example of the display screen for interactive editing in the sixth embodiment. It is an example of the display screen of the interactive editing in Example 7.

This embodiment describes a text information editing device. In this embodiment, an example in which the text information editing device handles Japanese text will be described, but the language used for the text is not limited. Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the present embodiment, in principle, the same components are designated by the same reference numerals, and the repeated description will be omitted. It should be noted that the present embodiment is merely an example for realizing the present invention and does not limit the technical scope of the present invention.

FIG. 1 is a block diagram showing a configuration example of a text information editing device. The text information editing device 100 is composed of, for example, a computer having a CPU (Central Processing Unit) 101, a memory 102, an auxiliary storage device 103, and a communication device 104.

The CPU 101 includes a processor and executes a program stored in the memory 102. The memory 102 includes a ROM (Read Only Memory) which is a non-volatile storage element and a RAM (Random Access Memory) which is a volatile storage element. The ROM stores an invariant program (for example, BIOS (Basic Input / Output System)) and the like. The RAM is a high-speed and volatile storage element such as a DRAM (Dynamic Random Access Memory), and temporarily stores a program executed by the CPU 101 and data used when the program is executed.

The auxiliary storage device 103 is, for example, a large-capacity and non-volatile storage device such as a magnetic storage device (HDD (Hard Disk Drive)) or a flash memory (SSD (Solid State Drive)), and is a program and a program executed by the CPU 101. Stores the data used when executing. That is, the program is read from the auxiliary storage device 103, loaded into the memory 102, and executed by the CPU 101.

The text information editing device 100 may have an input interface 105 and an output interface 108. The input interface 105 is an interface to which an input device such as a keyboard 106 or a mouse 107 is connected and receives input from an operator. The output interface 108 is an interface to which a display device 109 such as a printer or a display is connected and outputs a program execution result in a format that can be visually recognized by an operator. Further, for example, a microphone for inputting a voice dialogue may be connected to the input interface 105 as an input device.

The communication device 104 is a network interface device that controls communication with other devices according to a predetermined protocol. Further, the communication device 104 may include a serial interface such as USB.

The program executed by the CPU 101 is provided to the text information editing device 100 via a removable medium (CD-ROM, flash memory, etc.) or a network, and is stored in the non-volatile auxiliary storage device 103, which is a non-temporary storage medium. .. Therefore, the text information editing device 100 may have an interface for reading data from removable media.

The text information editing device 100 is a computer system composed of physically one computer or a plurality of computers logically or physically configured, and operates in separate threads on the same computer. It may operate on a virtual computer built on a plurality of physical computer resources. The same applies to the terminal 200.

In the present embodiment, the input device connected to the text information editing device 100 inputs to the text information editing device 100, but another computer such as a tablet terminal owned by each participant in the dialogue, for example. Input may be made from. Similarly, in the present embodiment, the calculation result by the text information editing device 100 is output to the display device 109 connected to the text information editing device 100, but for example, a tablet terminal owned by each participant in the dialogue. The calculation result may be output to another computer.

The CPU 101 includes, for example, an interactive input receiving unit 111, an important expression detecting unit 112, an editing input receiving unit 113, an interactive editing unit 114, an output information generating unit 115, and an output unit 116.

The dialogue input reception unit 111 receives the input of the dialogue text via the input device connected to the input interface 105, and divides the dialogue text into utterance units. Texts transcribed from dialogues and conferences, text chats, texts in web bulletin boards and forums, and texts in e-mail and SMS are all examples of dialogue texts. The dialogue text in the present embodiment is composed of utterance texts by a plurality of humans, but a text composed of one or more utterance texts by one or more humans may be used as the dialogue text.

The important expression detection unit 112 detects important expressions such as important utterances, phrases, and a plurality of utterances from the utterance text. The edit input reception unit 113 accepts the input of the annotation attached in the utterance text. The interactive editing unit 114 adds the annotation input to the editing input receiving unit 113 to the utterance text. The output information generation unit 115 generates information for outputting the annotated utterance text to the display device 109.

The output unit 116 outputs the utterance received by the dialogue input reception unit 111, the important expression extracted by the important expression detection unit 112, the editing result report by the interactive editing unit 114, and the like to the display device 109. The output unit 116 provides the user with a user output result in a form that can be used by the user, such as a GUI (Graphical User Interface), a text file, an image file, or an audio file.

For example, the CPU 101 functions as the dialogue input reception unit 111 by operating according to the dialogue input program loaded in the memory 102, and operates according to the important expression detection program loaded in the memory 102, so that the important expression detection unit 112 Functions as. The relationship between the program and the functional unit is the same for the other functional units included in the CPU 101.

Note that some or all of the functions of the functional unit included in the CPU 101 may be realized by hardware such as ASIC (Application Specific Integrated Circuit) or FPGA (Field-Programmable Gate Array).

The auxiliary storage device 103 holds, for example, the utterance annotation DB (DataBase) 121 and the editing processing information DB 122. Note that some or all of the information stored in the auxiliary storage device 103 may be described in the program stored in the memory 102.

The utterance annotation DB 121 holds the dialogue text input to the dialogue input receiving unit 111 and the annotation given by the interactive editing unit 114. The editing processing information DB 122 holds information such as an algorithm for detecting an important expression from the uttered text and a correspondence between a handy word that can be an important expression and the content of an annotation to be given.

FIG. 2 is an example of dialogue text data output by the dialogue input reception unit 111. The dialogue input reception unit 111 receives the input of the dialogue text via the input interface 105, divides the input dialogue text into utterance texts, assigns an utterance ID that can uniquely identify the utterance, and sets the utterance text and the utterance ID. Correspondence is transmitted to the output unit 116. The output unit 116 displays the dialogue text data in which the utterance ID and the utterance text are associated with each other on the display device 109.

FIG. 3 is an example of data output by the important expression detection unit 112 to the interactive editing unit 114. The important expression detection unit 112 adds an important expression detected from each utterance text as an annotation. The number of important expressions included in one utterance does not have to be one or less, and may be plural, or important expressions may be extracted across a plurality of utterances. The utterance annotation DB 121 retains the correspondence between the utterance ID, the utterance text, and the annotation shown in FIG. Further, the utterance annotation DB 121 may include a dialogue ID indicating which dialogue text includes each utterance text.

FIG. 4 is a flowchart showing an example of the text information editing process. The text information editing process is executed, for example, every time the dialogue input receiving unit 111 accepts the input of the dialogue text. For example, when the dialogue text is input by voice, the dialogue input reception unit 111 may input, for example, the start timing and the end timing of the dialogue, or the dialogue is performed after a predetermined time has elapsed from the previous utterance. It is determined that it is the end timing of the dialogue, and when the first utterance after the end timing of the dialogue starts, it is determined to be the start timing of the dialogue. Further, for example, when the dialogue input by the text file is input, the dialogue input receiving unit 111 determines that the text in the text file is one dialogue, for example.

First, the dialogue input reception unit 111 generates an utterance text by dividing the input dialogue text into utterance units, assigns a utterance ID to each of the generated utterance texts, and utters the utterance text and the utterance ID. It is stored in the DB 121 (S401).

Specifically, for example, the dialogue input reception unit 111 divides the dialogue text at a point where the utterance time interval is more than a predetermined time when the dialogue text by voice is input. Further, for example, the dialogue input reception unit 111 may receive a predetermined character code (for example, a punctuation mark, an exclamation mark, a space, or a space) in the text when the dialogue text is input by the text file. Divide the dialogue text before and after (enter, etc.). In addition, the dialogue text divided for each utterance may be input to the dialogue input reception unit 111 in advance, and in this case, the process of step S401 may not be performed.

The dialogue input reception unit 111 may transmit the utterance ID, the utterance text, and the output unit 116, and the output unit 116 may output the received utterance ID and the utterance text, for example, as shown in FIG.

The important expression detection unit 112 detects the important expression by referring to each utterance text (S402). Specifically, for example, the important expression detection unit 112 extracts the clue word stored in the editing processing information DB 122 from the utterance text as an important expression. For example, in the utterance text in which the utterance ID of FIG. 3 is "1", the expressions "example" and "for example" are important expressions defined in advance in the editing processing information DB 122 (or described in the program). When a representation is given as an annotation, the annotation is an "example". Further, for example, the important expression detection unit 112 may detect the important expression by applying a machine learning discriminator, TF-IDF, a statistical model, or the like to the dialogue text and / or the utterance text.

In step S402, the important expression detection unit 112 may add the detected important expression to the annotation of the utterance text and store it in the utterance annotation DB 121. In this case, the user deletes the important expression from the annotation. Remove the important expression from the annotation when instructed to do so.

The important expression detection unit 112 transmits the utterance ID, the utterance text, and the detected important expression to the output unit 116, and the output unit 116 sends the received utterance ID, the utterance text, and the important expression to the user. It is displayed on the display device 109 in an editable manner (S403). When the process of step S403 is completed, the user can edit the annotation, sort the utterances, search the utterances, and so on. On the display screen where editing input is possible, for example, an utterance in which an utterance ID is assigned to a text box on the GUI is displayed so that the user can easily recognize the text to be edited.

The interactive editing unit 114 determines whether or not the editing input receiving unit 113 has accepted the editing input from the user (S404). In step S404, the edit input receiving unit 113 may accept the user's input via the image and video, the sensor data, and the GUI such as click and tap in the dedicated application, in addition to the text and voice. The specific editing contents by the user will be described later.

When the interactive editing unit 114 determines that the editing input receiving unit 113 has received the editing input from the user (S404: YES), the interactive editing unit 114 performs editing according to the user's editing input, and stores the edited utterance and annotation in the utterance annotation DB 121. (S405). Specifically, for example, the interactive editorial unit 114 can add, delete, and replace lines, and add and delete annotations to the data shown in FIG. For example, for the utterance text or the like that is not related to the edit input, the information output in step S403 may continue to be output as it is.

Subsequently, the output information generation unit 115 generates information for displaying the edited utterance and annotation, and the output unit 116 displays the information on the display device 109 (S406).

When the interactive editing unit 114 determines that the editing input receiving unit 113 does not accept the editing input from the user (S404: YES), the output information generating unit 115 provides information for displaying unedited utterances and comments. The output unit 116 generates and displays the information on the display device 109 (S406).

Further, the edit input receiving unit 113 may receive a plurality of times and a plurality of types of edit inputs in one text editing process. In that case, the interactive editing unit 114 performs sequential editing, and the output unit 116 thereof. The editing result may be output sequentially. As a result, the user can continue editing after confirming the editing results one by one.

FIG. 5 is an example of a display screen for interactive editing. The display screen 500 shows a text box 501 and a text box 502 for displaying the spoken text, a text box 503 for showing the important expression detected by the important expression detection unit 112, and a text box 504 for showing the result of editing input by the user.

In the example of FIG. 5, since the important expression detection unit 112 determines that the expression "seeing the input and processing it interactively" in the text in the text box 502 is an important expression, the expression is referred to as an important expression in step S403. A text box 503 indicates whether the text should be annotated. Then, in step S404, the user makes an edit input indicating that the expression is an important expression in the text box 504.

Note that the user can also input an annotation as an important expression for the utterance text in which the important expression is not detected by the important expression detection unit 112 in the edit input. In this case, the annotation as an important expression input by the user is displayed in the text box 504.

Further, the display screen may include the emotion icon 505. For example, the important expression detection unit 112 may perform the sentiment analysis of the dialogue from the text of the dialogue text according to a predetermined algorithm (and when the dialogue text is given by voice, from the frequency of the speaker's voice or the like. (Sentiment analysis may be performed), in this case, the result of the sentiment analysis is presented to the user by the sentiment icon 505.

The emotion icon 505 makes it easier for the user to determine whether or not the expression contained in the presented dialogue text is an important expression. The emotion indicated by the emotion icon 505 may be determined for the entire dialogue text, or may be determined for each one or more utterances.

Further, for example, when the result of the sentiment analysis for a certain utterance is a predetermined emotion (for example, "joy" that promotes discussion), the important expression detection unit 112 includes important information in the immediately preceding utterance. It can be regarded as an important expression and the utterance immediately before the said can be annotated.

Further, the user may specify an emotion in the edit input and input the emotion. In this case, the emotion icon 505 indicates the input emotion, and the important expression detection unit 112 inputs the important expression based on the input emotion. It may be detected.

For example, when the important expression displayed in the text box 503 represents homework matters in the meeting, the user can make an utterance (edit input) as displayed in the text box 504 to express the important expression. Can be annotated as homework items, and homework items can be extracted and referred to via the output information generation unit 115 and the output unit 116 after the meeting is completed.

Further, for example, the user can input that there is an unnecessary utterance in the edit input (for example, the unnecessary utterance can be deleted from the utterance commentary DB 121 by performing an edit input for inputting the utterance ID of the unnecessary utterance). By adding an explanation to the unclear utterance (for example, by performing an edit input to input the utterance ID and the explanation of the unclear utterance, the additional explanation is recorded in the utterance corresponding to the unclear utterance in the utterance comment DB. ) Can also be done.

Whether or not the important expression detected by the important expression detection unit 112 corresponds to the homework matter and whether or not the dialogue contains unnecessary utterances may depend on the user's point of view. By configuring the text information editing device 100 as described above, the text information editing device 100 can proofread, summarize, and extract information from the dialogue text from the user's point of view.

For the examples after the second embodiment, the differences from the examples prior to the embodiment will be explained, and the description of the same configuration as that of the previous embodiment will be omitted in principle. FIG. 6 is an example of the interactive editing display screen output in step S406. The display screen 500 includes a text box 501 and a text box 502 for displaying the spoken text, and a button 601 for rearranging the order of the spoken text.

The user can change the order of utterances by selecting button 601 in the edit input. Specifically, for example, when the button 601 is selected, the output unit 116 has a display screen for inputting the order of each utterance text, or a position (for example,) in which the utterance text selected by the user is specified by the user. , Top or bottom) to display the display screen. According to this embodiment, for example, important utterances can be moved higher and less important utterances can be moved lower, and the dialogue text can be proofread, summarized, and information from the user's point of view. Extraction can be performed.

FIG. 7 is an example of the interactive editing display screen output in step S406. The display screen 500 includes a text box 501 and a text box 502 for displaying the spoken text, and a card 701 for displaying important expressions.

The important expression detected by the important expression detection unit 112 is displayed on the card 701. By selecting card 701 in the edit input, the user can save important utterances as annotations from any viewpoint such as homework, and proofread, summarize, and extract information from the dialogue text according to the user's viewpoint. It can be performed. The card 701 may display an important expression specified by the user in the edit input.

FIG. 8 is a block diagram showing a configuration example of the text information editing device 100 of this embodiment. The text information editing device 100 of the present embodiment is different from the text information editing device 100 of the first embodiment in that the CPU 101 includes the topic boundary identification unit 117.

The topic boundary identification unit 117 can identify one or more utterance texts (hereinafter, also referred to as utterance groups) belonging to the same topic by dividing the utterances included in the dialogue text by the topic boundary. For example, in a dialogue between a plurality of people, an utterance containing a clue word indicating an aizuchi or consent (for example, a word such as "yes" or "yes") is made on the same topic as the immediately preceding utterance. Therefore, it is considered that the utterance showing the aizuchi or consent and the utterance immediately before it belong to the same group of topics.

In addition, utterances that include clues and conjunctions that indicate a causal relationship are also conducted on the same topics as the immediately preceding utterance. Therefore, it is considered that the utterance showing the aizuchi or consent and the utterance immediately before it belong to the same group of topics.

Further, for example, the topic boundary identification unit 117 uses morphological analysis / syntactic analysis to identify an utterance that states a question and an utterance that states an answer corresponding to the question, and these utterances are the same. Determined to belong to the utterance group. Further, for example, the topic boundary identification unit 117 calculates the similarity between utterances by using parsing, and the utterances whose similarity is equal to or higher than a predetermined value have overlapping contents, so that the same utterance group is used. It may be determined that the utterances that have a similarity less than the predetermined value belong to a different utterance group (that is, a topic boundary is generated).

Further, for example, the topic boundary identification unit 117 may calculate the amount of information of the utterance and determine that the utterance with a small amount of information belongs to the same utterance group as the previous or subsequent utterance. For the amount of utterance information, for example, a predetermined weight may be applied to each speaker. Further, for example, the topic boundary identification unit 117 may specify the utterance group (topic boundary) by applying a classifier by machine learning or the like to the dialogue text and / or the utterance text.

Further, for example, the topic boundary identification unit 117 may specify the utterance group (topic boundary) by applying a predetermined weight to a part or all of the above-mentioned utterance group (topic boundary).

Further, the topic boundary identification unit 117 can identify the topic of each utterance group by extracting the words and phrases that characterize the utterance group by using, for example, morphological analysis and syntactic analysis.

Information necessary to generate topic boundaries, such as clues and algorithms that identify the boundaries of utterances as described above, clues and algorithms that identify that multiple utterances belong to the same utterance group, and topics. Information necessary for specifying a topic such as an algorithm for specifying the above may be stored in advance in the editing process information DB 122, or may be described as a program.

In the above, the boundary between utterance groups is the topic boundary, but the utterance group indicated by this topic boundary is exclusive, that is, the utterances do not have to overlap between the utterance groups, and one utterance becomes a plurality of utterance groups. May belong.

FIG. 9 is an example of dialogue text data output by the topic boundary identification unit 117. In the topic boundary identification unit 117, the dialogue input reception unit 111 receives the utterance ID and the utterance text, identifies the topic boundary from the utterance text, and transmits the topic boundary to the output unit 116 via the output information generation unit 115. In addition to the data of FIG. 2, the dialogue text data of FIG. 9 includes a topic boundary that can identify the utterance at which the topic starts and the utterance at which the topic ends.

FIG. 10 is a flowchart showing an example of the text information editing process. Following step S401, the topic boundary identification unit 117 identifies the topic boundary in the dialogue text by the method described above (S1001). At this time, the topic boundary identification unit 117 transmits the topic boundary to the output information generation unit 115, the output information generation unit 115 generates output information indicating the topic boundary, and the output unit 116 displays the output information 109. It may be displayed in. When the process of step S1001 is completed, the process proceeds to step S402.

FIG. 11 is an example of the display screen output in step S406. The display screen 500 includes the utterance group box 1101 and the utterance group box 1104, and the dialogue text is divided and displayed for each utterance group. The utterance group box 1101 includes a text box 501 and a text box 502 that display the text of the utterance included in the utterance group, a topic box 1102 that indicates the topic of the utterance group, and a tag that characterizes the topic with respect to the topic. Includes button 1103 for presentation. In the topic box 1102, not the topic but the first utterance of the utterance group, the important expression in the utterance group detected by the important expression detection unit 112, and the like may be displayed.

In the area next to the button 1103, a tag that characterizes the topic specified by the topic boundary identification unit 117 is presented, and the user can input information indicating whether or not the tag is appropriate in the edit input. It should be noted that only one tag may be attached to one utterance group, a plurality of tags may be attached, or no tag may be attached.

Further, a pop-up display or the like may be performed using the button 1103 so that the user can freely input a tag indicating the topic according to the pop-up display in the edit input, for example, and the user can input a tag that characterizes the topic. You can freely save it as an annotation. Tags can describe, for example, topical topics and special notes such as homework in free words and sentences.

In this way, when the tag is attached, the text information editing device 100 can investigate the utterance group in which a similar topic is spoken, for example, by using the feature amount of the utterance in the utterance group. As this feature amount, the frequency of words appearing in the utterance group and the feature amount applicable to any sentence such as TF-IDF can be used.

When similar utterance groups are extracted, the user can efficiently edit the dialogue text for similar utterance groups, and provide necessary information based on the topics and homework annotated as tags. It can be easily extracted and summarized. Further, the user can delete the tag attached to the similar utterance group by editing input.

With the text information editing device 100 configured in this way, it is possible to proofread, summarize, and extract information from the dialogue text from the user's point of view.

FIG. 12 is an example of the interactive editing display screen output in step S406. The display screen 500 includes the utterance group box 1101 and displays the dialogue text divided for each utterance group. Further, the display screen 500 includes a search result display box 1204. The utterance group box 1101 includes a text box 501 and a text box 502 that display the text of the utterance included in the topic, a topic box 1102 that indicates the topic of the utterance group, a search box 1202, and a search button 1203.

The user can extract and rank utterance groups and utterances related to the utterance group by inputting a query such as a keyword in the search box 1202 and selecting the search button 1203 in the edit input.

The search result display box 1204 includes snippet 1205 and snippet 1206. Snippets 1205 and snippets 1206 can display important expressions extracted as search results as headings. As a result, the user can efficiently search for utterances and topic boundary ranges related to the utterance group.

Further, when the user selects snippet 1205 and snippet 1206 in the edit input, the utterance or topic boundary range related to the utterance group can be linked to the utterance group, and the interactive editorial unit 114 can connect the result to the utterance group, for example. It can be stored in the utterance annotation DB 121 in the form of an annotation or the like.
With the text information editing device 100 configured in this way, it is possible to proofread, summarize, and extract information from the dialogue text from the user's point of view.

FIG. 13 is an example of the interactive editing display screen output in step S406. The display screen 500 includes the utterance group box 1101 and the utterance group box 1104, and the dialogue text is divided and displayed for each utterance group. The utterance group box 1101 includes a text box 501 and a text box 502 for displaying the utterance text included in the topic, and a similar utterance display box 1301.

The topic boundary identification unit 117 extracts, for example, utterances that are desirable to belong to the utterance group from the utterances other than the series of utterances included in the utterance group box 1101 as similar utterances, and displays them in the similar utterance display box 1301. Specifically, for example, the topic boundary identification unit 117 extracts utterances other than the series of utterances as similar utterances if the degree of similarity with the series of utterances included in the utterance group box 1101 is equal to or higher than a predetermined value.

When the user selects the similar utterance display box in the edit input, the interactive editorial unit 114 adds the similar utterance to the utterance group and changes the topic of the similar utterance to the topic corresponding to the utterance group. Add and save in the utterance annotation DB 121.

FIG. 14 is an example of the interactive editing display screen output in step S406. The display screen 500 includes a topic box 1102 and a topic box 1105, and a text box 501 and a text box 502.

The utterance of the text box 501 displayed directly under the topic box 1102 is an utterance belonging to the utterance group indicating the topic of the topic box 1102, and the utterance of the text box 502 displayed directly under the topic box 1105 is a topic. Box 1105 An utterance belonging to an utterance group indicating a topic.

Since the topic boundary identification unit 117 determines that the topic in the topic box 1102 and the utterance in the text box 501 are related, they are connected by a solid line edge. Similarly, since the topic boundary identification unit 117 determines that the topic of the topic box 1107 and the utterance of the text box 502 are related, they are connected by a solid line edge.

Further, the display screen 500 includes an edge addition button 1401 and an edge deletion button 1402. The user can add a dotted edge by selecting two topic boxes or text boxes and selecting the edge addition button 1401 in the edit input, and the interactive editorial unit 114 can add the two topics or utterances. The utterances connected by the edge may be registered in the utterance commentary DB 121, or the utterances connected by the edge may be included in the same utterance group.

The user can delete the edge by selecting the edge and selecting the edge delete button 1402 in the edit input, and the interactive editorial unit 114 may delete the information about the edge from the utterance annotation DB 121. However, the utterance with the edge removed may be excluded from the utterance group.

As a result, the text information editing device 100 can connect utterances and utterance groups that are considered to be related from the user's point of view at an edge, and can accumulate this as a new annotation or change the utterance group.

Further, when the user specifies related utterances and utterance groups in the edit input, the interactive editorial unit 114 edits so as to continuously display, for example, related utterances according to the user's designation. , Related utterances and utterance groups can be extracted as shown in Example 4 and Example 5.
With the text information editing device 100 configured in this way, it is possible to proofread, summarize, and extract information from the dialogue text from the user's point of view.

The present invention is not limited to the above-described embodiment, and includes various modifications. For example, the above-described embodiment has been described in detail in order to explain the present invention in an easy-to-understand manner, and is not necessarily limited to the one including all the described configurations. It is also possible to replace a part of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add the configuration of another embodiment to the configuration of one embodiment. Further, it is possible to add / delete / replace a part of the configuration of each embodiment with another configuration.

Further, each of the above configurations, functions, processing units, processing means, etc. may be realized by hardware by designing a part or all of them by, for example, an integrated circuit. Further, each of the above configurations, functions, and the like may be realized by software by the processor interpreting and executing a program that realizes each function. Information such as programs, tables, and files that realize each function can be placed in a memory, a hard disk, a recording device such as an SSD (Solid State Drive), or a recording medium such as an IC card, an SD card, or a DVD.

In addition, the control lines and information lines indicate those that are considered necessary for explanation, and not all control lines and information lines are necessarily indicated on the product. In practice, it can be considered that almost all configurations are interconnected.

Claims

It is a text information editing device
It has a processor and a storage device,
Connected to the display device,
The storage device is
Dialogue text consisting of one or more utterance texts and
The editing processing information indicating at least one of the important expression and the extraction algorithm of the important expression is retained.
The processor
An important expression is extracted from the one or more utterance texts with reference to the editing processing information.
The dialogue text and the extracted important expressions are displayed on the display device.
Accepts edit input for at least one of the utterance text included in the dialogue text and the annotation to which the utterance text is attached.
A text information editing device that executes editing corresponding to the editing input.
The text information editing device according to claim 1.
When the processor receives an instruction to associate the important expression displayed on the display device with the utterance text including the important expression as the annotation in the editing input, the utterance text and the important expression are described. A text information editing device that is associated as an annotation and stored in the storage device.
The text information editing device according to claim 1.
The processor
As the edit input, the specification of the text included in the dialogue text is accepted.
A text information editing device that associates a text that has received the designation and an utterance text that includes the text as the annotation and stores the text in the storage device.
The text information editing device according to claim 1.
The processor is a text information editing device that receives a rearrangement instruction of one or more utterance texts included in the dialogue text as the editing input.
The text information editing device according to claim 1.
The editing processing information indicates an algorithm for classifying the dialogue text into utterance groups which are utterance texts related to the same topic.
The processor
With reference to the editing process information, the dialogue text is classified into one or more utterance groups.
The important expression is stored in the storage device in association with the utterance group including the important expression, and displayed on the display device.
A text information editing device using the information indicating the classification result of the one or more utterance groups as the annotation.
The text information editing device according to claim 5.
The editing processing information indicates an algorithm for identifying the topic of the utterance group.
The processor
With reference to the editing process information, the topic of the one or more utterance groups is identified, and the topic is identified.
The one or more utterance groups are displayed on the display device in association with the specified topic.
A text information editing device that stores the specified topic as the annotation in the storage device.
The text information editing device according to claim 5.
The editing process information holds a clue word for classifying the dialogue text into the one or more utterance groups.
The processor is a text information editing device that determines that the utterance text including the clue word and the utterance text before or after the utterance text belong to another utterance group.
The text information editing device according to claim 7.
The clue word is a text information editing device including at least one of an expression indicating an aizuchi, a conjunction, and an expression indicating a causal relationship.
The text information editing device according to claim 5.
The processor
As the edit input, the input of the tag for the first utterance group is accepted.
A text information editing device that stores the tag in the storage device in association with the first utterance group.
The text information editing device according to claim 5.
The processor
As the edit input, the input of the search query for the second utterance group is accepted.
A text information editing device that displays a search result for the second utterance group by the search query on the display device.
The text information editing device according to claim 5.
The processor
Identify the utterance texts that are not included in the third utterance group and whose similarity with the utterance texts included in the third utterance group is equal to or greater than a predetermined value.
The identified utterance text is displayed on the display device in association with the third utterance group.
When an instruction to include the specified utterance text in the third utterance group is received as the edit input, the annotation indicating that the specified utterance text is included in the third utterance group is sent to the storage device. A text information editing device that stores.
The text information editing device according to claim 5.
When the processor receives an instruction to connect utterance texts included in different utterance groups at an edge as the edit input, the processor associates the utterance texts connected at the edge, stores the utterance text in the storage device, and stores the utterance text in the display device. A text information editing device to display.
The text information editing device is a method of editing text information.
The text information editing device is connected to the display device and is connected to the display device.
The text information editing device is
Dialogue text consisting of one or more utterance texts and
The editing processing information indicating at least one of the important expression and the extraction algorithm of the important expression is retained.
The method is
The text information editing device extracts important expressions from the one or more utterance texts with reference to the editing processing information.
The text information editing device displays the dialogue text and the extracted important expression on the display device.
The text information editing device accepts edit input for at least one of the utterance text included in the dialogue text and the annotation to which the utterance text is attached.
A text information editing method in which the text information editing device executes editing corresponding to the editing input.