JP6740187B2 - Dialogue device and program - Google Patents

Description

The present invention relates to a dialogue device and a program.

Many of the current dialogue systems respond uniformly to any user regardless of their characteristics. However, if the user responds uniformly regardless of the preference or knowledge of the user, the satisfaction and understanding of the response content cannot be improved. Therefore, it is necessary to automatically estimate the characteristics of the user and present the response content according to the estimation result.

One of the characteristics of users is the amount of knowledge about ICT (Information and Communication Technology) such as PCs, smartphones, and tablets. At present, a voice or text dialogue system is often used for inquiries from users regarding ICT devices and services. Therefore, the amount of knowledge about ICT is estimated from the use of the dialogue system, and a dialogue corresponding to the amount of knowledge is performed. It is important to do.

Conventionally, there has been proposed a technique for converting a term into an expression that is easy for the user to understand, depending on the level of literacy about the user's ICT, knowledge of the operation method of the device, and knowledge of the term (Patent Documents 1 and 2). ). In addition, a technique for estimating the amount of knowledge of a user's ICT has been proposed (Patent Documents 3 and 4).

JP, 2013-250624, A JP, 2014-2517, A JP, 2013-117842, A JP, 2014-115719, A

In the technique described in Patent Document 1, an operator manually inputs the literacy level of the user based on the age of the user, the years of Internet use, etc. at a call center, etc., and then converts the ICT terminology for easy understanding according to the level. And outputs it to support the operator's response.

Further, in the technique described in Patent Document 2, the ICT skill level of the user is calculated in advance based on the result of the questionnaire in advance, and then the degree of feeling given by each ICT term to the user of that level is estimated. ..

As described above, it is possible to study an appropriate expression according to the level of the user's ICT-related knowledge amount and the like using the techniques of Patent Documents 1 and 2. However, the method of calculating the amount of knowledge about the ICT of the user is based on the operator's input or the answer to the questionnaire, and cannot be automatically estimated.

On the other hand, in the technology described in Patent Document 3, the knowledge amount of the user is "large knowledge amount" in two steps or three steps using the information of the number of questions of the user or operator and the timing thereof, It is estimated such as "," the amount of knowledge is small. However, from this estimation result, it is not possible to specifically know what terms a user with a large amount of knowledge knows and what terms a user with a small amount of knowledge does not know. Therefore, even if an attempt is made to respond to the user's knowledge amount using the estimation result, for example, it is not possible to know how to explain to a user having a small knowledge amount.

In addition, in the technique described in Patent Document 4, a vocabulary network of ICT terms is created, and a user corresponding to a node corresponding to the ICT term and a directed link that knows this term if another user knows the term. Knowledge is estimated from the input information as to whether or not a certain term is known. With this technique, the known and unknown terms of the user can be known. However, if the input information regarding the understanding of the term from the user is insufficient, the accuracy is reduced.

The present invention has been made in view of the above points, and an object of the present invention is to improve the accuracy of estimating the knowledge of the user of the other party of the dialogue.

Therefore, in order to solve the above-mentioned problem, the dialog device outputs a second question that asks the user whether or not he/she understands a term related to the first question, and a receiving section that receives an input of the first question from the user. An output unit, and an estimation unit that estimates the level of knowledge of the user in the field related to the first question based on the content of the answer to the second question and the state of the input operation of the character string indicating the answer. And a generation unit that generates the content of the answer to the first question according to the level estimated by the estimation unit.

It is possible to improve the accuracy of estimating the knowledge of the user with whom the user is interacting.

It is a figure which shows the hardware structural example of the interactive apparatus 10 in embodiment of this invention. It is a figure which shows the function structural example of the interactive apparatus 10 in embodiment of this invention. 6 is a flowchart for explaining an example of a processing procedure executed by the dialogue device 10. It is a figure which shows the structural example of the scenario table of the dialogue scenario storage part 161. It is a figure which shows the structural example of the character input speed storage part 162. 9 is a flowchart for explaining an example of a processing procedure of question sentence generation processing. It is a figure for demonstrating the example of determination of the presence or absence of term understanding. It is a figure which shows the structural example of the term understanding storage part 163. FIG. 6 is a diagram showing a configuration example of a term explanation storage unit 164. It is a figure which shows the structural example of the task table of TETHERING of the conversation scenario storage part 161. It is a flow chart for explaining an example of a processing procedure of answer sentence generation processing. It is a figure which shows the 1st example of the calculation method of a term comprehension degree. It is a figure which shows the 2nd example of the calculation method of term comprehension degree. It is a figure which shows the 1st example of the estimation method of a knowledge amount. It is a figure which shows the 2nd example of the estimation method of a knowledge amount. It is a figure which shows the structural example of the answer sentence table of TETHERING of the conversation scenario storage part 161.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the present embodiment, an example in which the dialogue device 10 estimates the amount of knowledge about the user's ICT (Information and Communication Technology) from the input information about the non-linguistic information and the understanding of the term, and presents an answer according to the amount of knowledge explain.

FIG. 1 is a diagram showing a hardware configuration example of a dialogue device 10 according to an embodiment of the present invention. The dialog device 10 of FIG. 1 includes a drive device 100, an auxiliary storage device 102, a memory device 103, a CPU 104, an interface device 105, a display device 106, and an input device 107, which are connected to each other by a bus B.

The program that realizes the processing in the dialogue device 10 is provided by the recording medium 101 such as an SD card or a CD-ROM. When the recording medium 101 storing the program is set in the drive device 100, the program is installed in the auxiliary storage device 102 from the recording medium 101 via the drive device 100. However, the program does not necessarily have to be installed from the recording medium 101, and may be downloaded from another computer via a network. The auxiliary storage device 102 stores the installed program and also stores necessary files and data.

The memory device 103 reads the program from the auxiliary storage device 102 and stores the program when an instruction to activate the program is given. The CPU 104 realizes the function related to the dialog device 10 according to the program stored in the memory device 103. The interface device 105 is used as an interface for connecting to a network. The display device 106 displays a GUI (Graphical User Interface) or the like according to a program. The input device 107 includes a keyboard and a mouse, a touch panel, or the like, and is used to input various operation instructions.

The dialogue device 10 may be a terminal used by the user to make an inquiry, or may be a device different from the terminal. However, in the present embodiment, the dialogue device 10 makes an inquiry by the user. It will be described as a terminal used to perform. In this case, for example, a mobile phone, a smartphone, a tablet terminal, a PC (Personal Computer), or the like is given as an example of such a terminal. However, the dialogue device 10 may be realized by one or more computers connected to a terminal operated by the user via a network.

FIG. 2 is a diagram showing a functional configuration example of the dialogue device 10 in the embodiment of the present invention. The dialog device 10 determines the amount of knowledge about the field (ICT field in this embodiment) related to the initial question from the user (hereinafter referred to as “user initial question”) during the dialog between the user and the interactive device 10. It is a device that estimates from non-linguistic information and information input by a user regarding understanding of terms and presents a description according to the amount of knowledge. Note that the interaction between the user and the dialogue device 10 is basically a question sentence output from the dialogue device 10 in order to estimate the amount of knowledge of the user regarding a field related to the user opening question, and the user's response to the question sentence. It is performed by inputting a character string indicating an answer (hereinafter, referred to as “user answer”) one or more times. When sufficient dialogue is performed to estimate the amount of knowledge of the user, the dialogue device 10 generates an answer to the user's opening question (hereinafter referred to as “system answer”) according to the amount of knowledge of the user. Output the system answer.

In FIG. 2, the dialogue device 10 includes a question sentence generation unit 11, a text input unit 12, a dialogue control unit 13, a dialogue sentence output unit 14, an answer sentence generation unit 15, and the like. The question sentence generation unit 11 includes an option determination unit 111 and a term understanding determination unit 112. In addition, the answer sentence generation unit 15 includes a knowledge amount estimation unit 151. The knowledge amount estimation unit 151 includes a character input speed calculation unit 1511 and a term understanding level calculation unit 1512. Each of these units is realized by a process that causes the CPU 104 to execute one or more programs installed in the dialog device 10. The dialogue device 10 also uses a dialogue scenario storage unit 161, a character input speed storage unit 162, a term understanding storage unit 163, a term explanation storage unit 164, and the like. Each of these storage units can be realized by using, for example, the auxiliary storage device 102 or a storage device that can be connected to the dialog device 10 via a network.

Hereinafter, the processing procedure executed by the dialog device 10 will be described. FIG. 3 is a flowchart for explaining an example of a processing procedure executed by the dialogue device 10. The flowchart of FIG. 3 shows the entire processing procedure executed by the dialog device 10 when the dialog device 10 estimates the user's knowledge amount and presents a system answer corresponding to the user's initial question in accordance with the knowledge amount. Show.

In step S1, the dialogue device 10 activates the dialogue function according to a predetermined operation by the user. The interactive function has a functional configuration shown in FIG. 2, which is realized by, for example, a process such as a text interactive system operating on the browser of the interactive apparatus 10 or an independent interactive application program executed by the CPU 104. Is.

Subsequently, the dialogue control unit 13 refers to the scenario table stored in the dialogue scenario storage unit 161, obtains the question sentence to be output first (S2), and sends the obtained question sentence to the dialogue sentence output unit 14. Output. The question indicated by the question sentence that the dialogue device 10 first outputs is hereinafter referred to as a "system initial question".

FIG. 4 is a diagram showing a configuration example of the scenario table of the dialogue scenario storage unit 161. The scenario table is a table in which a scenario of a dialogue between the user and the dialogue device 10 is defined, such as “status”, “task”, “choice”, “question statement”, “keyword”, and “jump destination status”. Has columns (items). In FIG. 4, the unit of the line delimited by “status” is hereinafter referred to as “status record”.

"Status" is a concept that distinguishes the status of a conversation. The “question sentence” is a question sentence that should be output to the user in the corresponding status. That is, there is a one-to-one correspondence between the status and the question text. The “keyword” is a keyword for distinguishing the status of the next transition destination (jump destination status, etc.) according to the content of the character string input by the user with respect to the question sentence. The value of “keyword TRUE” in “keyword” indicates a case where the specified keyword is not included in the character string input by the user for the question sentence. In addition, the fact that the keyword is not included in the question text, that is, that the keyword is not registered, indicates that the output of the question text transits to the next status without waiting for the input of the character string from the user. For example, in FIG. 4, the “keyword” column for the “status” is “AGAIN( )” and “EXPLAIN( )” is blank. Therefore, when the “question text” whose “status” is “AGAIN( )” or “EXPLAIN( )” is output, the question text corresponding to the next status is displayed without waiting for the user to input a character string. Is output.

The “jump destination status” indicates the status of the transition destination according to the keyword included in the character string input by the user. When the “jump destination status” includes a plurality of statuses, it means that each status is selected in order and the question sentence corresponding to each status is output. Further, "next" in the "jump destination status" means that the status is the transition base of the status including the "next" and that the status is the next status after the statuses listed in the "jump destination status". To do. That is, when the "jump destination status" includes a plurality of statuses, the value of the "jump destination status" of the status other than the end is "next".

“Task” indicates whether or not the corresponding status is a status for executing the task. Specifically, the status in which the value of “task” is “task TRUE” indicates that the task is executed, and the status in which the “task” is blank indicates that the task is not executed. A task is a process that the dialogue device 10 finally executes in response to a dialogue with a user. In the present embodiment, an output of an answer sentence indicating a system answer to a user opening question corresponds to a task.

The “option” indicates whether or not the content of the “question sentence” in the corresponding status is a question sentence that gives a choice of user answers for term understanding.

In step S2, the dialogue control unit 13 acquires a question sentence corresponding to the current status as a text from the scenario table as shown in FIG. 4, for example. In this embodiment, the first status is "START". Therefore, the dialogue control section 13, to get the text of the "question" corresponding to the "START" status "Hello! Tell me your business!".

Subsequently, when the dialogue control unit 13 outputs the acquired text of the question sentence to the dialogue sentence output unit 14, the dialogue sentence output unit 14 outputs the input text of the question sentence to the display device 106 such as a display. And output (S3). For example, the question text is displayed as text on the display device 106. Subsequently, the dialogue control unit 13 determines whether or not the keyword corresponding to the current status is registered in the scenario table (FIG. 4) (S4). When the keyword is not registered (No in S4), the process proceeds to step S8. That is, in the scenario table (FIG. 4 ), when the “keyword” item is blank, the dialogue control unit 13 does not wait for the user to input a character string.

When the keyword is registered in the current status (Yes in S4), the text input unit 12 receives the input of the character string for the question sentence from the user (S5). For example, the user inputs a character string using the input device 107 such as a keyboard. At this time, the text input unit 12 accepts the character string input by the user and measures the time required from the start of the input of the character string by the user to the completion thereof. The timing of completion of the input of the character string by the user can be detected by, for example, a predetermined operation after the input of the character string (for example, pressing a “send” button). It should be noted that when step S5 is first executed, the dialogue device 10 outputs a question sentence indicating a system initial question. The character string input by the user in this state is usually a character string indicating the user's initial question. For example, a character string indicating the user's opening question such as “how to tether?” is input. On the other hand, the character string input when step S5 is executed for the second time and thereafter is a character string indicating the user answer to the question sentence output by the dialogue device 10. The user response is used by the interactive device 10 to estimate the amount of knowledge of the user regarding the field related to the user initial question.

Then, the text input unit 12 outputs the input character string and the required time to the dialogue control unit 13 (S6). Then, the dialogue control unit 13 saves the character string and the required time input from the text input unit 12 in the character input speed storage unit 162 (S7).

FIG. 5 is a diagram showing a configuration example of the character input speed storage unit 162. As shown in FIG. 5, the character input speed storage unit 162 stores the character string and the time required to input the character string each time the user inputs a character string for the question sentence.

Subsequently, the dialogue control unit 13 updates the status with reference to the scenario table (FIG. 4) (S8). That is, the “jump destination status” specified based on the scenario table (FIG. 4) and the keyword included in the character string input by the user is selected as the new status. For example, when the status before update is “START” and the character string input by the user is “How to tether?”, “TETHERING” is set as the new status. In addition, when a plurality of statuses are registered in the "jump destination status" and the status is the processing target first, the first (highest) status of the plurality of statuses is newly updated. Selected as the correct status. The second and subsequent statuses of the plurality of statuses are selected based on the "next" of the "jump destination status" in the new status.

Subsequently, the dialogue control unit 13 determines whether or not the task status has been transitioned (jumped) and the acquisition of the parameter related to the task has been completed (S9). The status as a task means a status in which the value of the task is “task TRUE” in the scenario table (FIG. 4).

If the task status has not been transitioned to, or if the acquisition of the parameter related to the task is not completed (No in S9), the question statement generation unit 11 executes the question statement generation process according to a request from the dialogue control unit 13. Yes (S10). The question sentence generation process (S10) is always executed when step S9 is first executed. In the question sentence generation process, a question sentence is mainly generated for the dialog device 10 to provide an appropriate system answer to the user initial question input by the user in step S5. Subsequent to the question sentence generation processing, step S3 and subsequent steps are executed. That is, step S3 and subsequent steps are executed for the question text generated in the question text generation processing.

On the other hand, when the status of the task has already been transited and the acquisition of the parameter related to the task is completed (Yes in S9), the response sentence generation unit 15 generates the response sentence in response to the request from the dialogue control unit 13. The process is executed (S11). In the answer sentence generation process, an answer sentence indicating a system answer to the user initial question input by the user in step S5 is generated. Subsequently, when the dialogue control unit 13 outputs the answer sentence generated by the answer sentence generation unit 15 to the dialogue sentence output unit 14, the dialogue sentence output unit 14 outputs the input answer sentence (S12). Note that FIG. 3 shows an example in which the process ends as it is, but the process may be returned to step S2 and a dialogue may be performed regarding another inquiry.

Next, details of step S10 will be described. FIG. 6 is a flowchart for explaining an example of the processing procedure of the question sentence generation processing.

First, the option decision unit 111 of the question sentence generation unit 11 refers to the scenario table (FIG. 4) (S10-1), and the question sentence finally output by the dialogue device 10 indicates the choices of user answers for term understanding. It is determined whether or not the question text is given (S10-2). Specifically, it is determined whether or not the value of “option” corresponding to the immediately preceding status (status before update) is “option TRUE”.

For example, if the procedure of the first to FIG. 6 is executed, the question statement that interactive device 10 is output to the last is a "Hello! Tell me your business!". Such a question sentence is not a question sentence that gives a choice of user answers to term understanding. That is, the value of "option" in the status of "START" is not "option TRUE". Therefore, in this case (No in S10-2), the process proceeds to step S10-8.

On the other hand, if the dialogue between the user and the dialogue device 10 progresses and the status corresponding to the question sentence last output by the dialogue device 10, that is, the immediately previous status is “THERING”, the dialogue device 10 displays “tethering”. By the way, in addition to the question sentence "Do you know tethering itself?", the user was presented with the option "Yes/No" in text. If the previous status is “THEHERING_OS”, the user is presented with an option “OS-A/OS-B/don't know” in addition to the question sentence “Which OS of the terminal are you using?”. I was doing it. Note that the options presented here are options that allow the user to determine the understanding of terms depending on which one the user selects.

When the question sentence finally output by the dialog device 10 is a question sentence that gives a user answer option for term understanding (Yes in S10-2), the term understanding determination unit 112 performs step S5 for the question sentence. The option input in (1) is used as an input, the presence or absence of the user's term understanding is determined based on the option, and the determination result is stored in the term understanding storage unit 163 (S10-3).

FIG. 7 is a diagram for explaining an example of determining whether or not a term is understood. As a method of determining whether or not the term is understood, as shown in FIG. 7, for example, “OS-A”, “OS-B”, “I don't know Is presented and the user answer is “OS-A” or “OS-B”, it is interpreted that both terms are understood, and the determination result is (OS-A, 1), (OS-B). , 1). Here, the notation of the determination result is (term, 0 or 1), 0 means not understood, and 1 means understood.

In addition, “yes” and “no” are presented as choices for the user answer to the question text that directly asks if the user knows the term itself. If the user answer is “yes”, the term is understood to be understood. However, if “No”, it is interpreted that the term is not understood, and the determination result is generated in the same manner as above. Note that the method of determining whether or not the term is understood is not limited to this.

The term understanding determination unit 112 stores the determination result of whether or not the term is understood in the term understanding storage unit 163. FIG. 8 is a diagram showing a configuration example of the term understanding storage unit 163. In the term understanding storage unit 163, terms related to ICT and their difficulty levels are stored in advance. The degree of difficulty of the term may be set in advance based on, for example, a questionnaire survey. In the present embodiment, 5 is the highest difficulty level, and 1 is the lowest difficulty level. However, the difficulty level is not limited to five levels, and may be any level or continuous value.

In step S10-3, the determination result of whether or not the term understanding is determined by the term understanding determination unit 112 is stored in the term understanding storage unit 163 as a value of the presence or absence of understanding of the term that is the determination target.

Then, the question sentence generation unit 11 determines whether or not the user understands the term for which the presence or absence of term understanding is determined, that is, the value stored in the term understanding storage unit 163 is 1 for the term. It is determined whether or not (S10-4). When the user understands the term (Yes in S10-4), that is, when 1 is stored, the process proceeds to step S10-8.

On the other hand, when the user does not understand the term (No in S10-4), that is, when 0 is stored, the question sentence generation unit 11 refers to the dialogue scenario storage unit 161, and responds to the current status. The question sentence to ask is acquired (S10-5). That is, the question sentence corresponding to the character string input by the user is acquired.

For example, in FIG. 4, the question message last output by the dialogue device 10 is “Tethering method” in the “THERING” status. By the way, do you know tethering itself? Yes/No. It is assumed that the user inputs the character string "No". In this case, the dialogue control unit 13 selects “EXPLAIN( )” which is the first status of the “jump destination status” corresponding to “NO”, and thus the status becomes “EXPLAIN( )”. Note that the "EXPLAIN()" status is passed as an argument a term that is determined to be a term that the user does not understand. It should be noted that, of the terms constituting the question sentence lastly output by the dialogue device 10, the term registered in the term understanding storage unit 163 is specified as the term that the user does not understand.

In this case, as shown in FIG. 4, the question sentence """#WORD"" corresponding to the "EXPLAIN()" status is "#EXP"! Is acquired. Here, “#WORD” and “#EXP” indicate specific embedding locations of the character string. “#WORD” indicates a term that is determined as a term that the user does not understand, which is passed as an argument to the “EXPLAIN( )” status. “#EXP” indicates an explanation sentence (term explanation sentence) corresponding to the term.

Subsequently, the question sentence generation unit 11 determines whether or not the question sentence acquired in step S10-5 has an embedded portion (S10-6). When there is an embedding place such as “#WORD” or “#EXP” (Yes in S10-6), the question sentence generation unit 11 refers to the term explanation storage unit 164 and fills the embedding place of the question sentence ( S10-7).

FIG. 9 is a diagram showing a configuration example of the term explanation storage unit 164. As shown in FIG. 9, the term explanation storage unit 164 stores the degree of difficulty, the term explanation, and the like for each term.

For example, if the "question text" is """#WORD"", it means "#EXP"! , And if the term that is determined not to be understood is “tethering”, “tethering” turns a smartphone into a mobile router, and uses a smartphone's 3G line or LTE to transfer a computer to the Internet. It is a function that can be connected! Is generated. In the present embodiment, an example in which the question sentence generation unit 11 fills in “#WORD”, “#EXP”, and the like in the templated “question sentence” to generate a question sentence that explains the term However, the method of generating the question sentence indicating the explanation of the term is not limited to this. For example, the question sentence generation unit 11 acquires a term explanation sentence corresponding to a term determined not to be understood from the term explanation storage unit 164, and automatically generates a natural sentence including the term and the term explanation sentence. It may be generated.

Subsequently, the question sentence generation unit 11 outputs the question sentence to the dialogue control unit 13 (S10-9). In this case, the question sentence is output in step S3 of FIG.

On the other hand, in step S10-8, the question sentence generation unit 11 acquires the question sentence corresponding to the current status in the same manner as in step S10-5, and then in step S10-9, the question sentence generation unit 11 issues the question sentence to the dialogue control unit 13. Output question text.

In FIG. 4, when the last status is “THERING” and the character string input by the user with respect to the question text is “No”, the “jump destination status” is “EXPLAIN( )”. Next to, "See the task table of TETHERING" is included. The corresponding keyword is not registered in the "EXPLAIN()" status, and the "jump destination status" of the "EXPLAIN()" status is "next". Therefore, in this case, "Tethering" is a function that allows you to connect your PC to the Internet using your smartphone's 3G line or LTE by turning it into a mobile router! Is output in step S3, steps S5 to S7 are not executed and the process proceeds to step S8. In step S8, the dialogue control unit 13 refers to "next" and refers to "the task table of Tethering" which is the second status of the "jump destination status" corresponding to "no" of the "tethering" status. select.

In this case, in step S10-5, the question sentence generation unit 11 refers to the task table of TETHERING stored in the dialogue scenario storage unit 161, based on the fact that the current status is “see task table of TETHERING”. To do.

FIG. 10 is a diagram showing a configuration example of a task table of TTHERING in the dialogue scenario storage unit 161. In the task table, the status in which the value of the “task” is “task TRUE” is selected, and the “jump destination status” based on the character string input by the user with respect to the question text in the status is “TETHERING”. Is a table that is referred to when it is determined to execute the “refer to the task table of TETHERING”.

In FIG. 10, the task table stores “task name” for each task, 1 or more “parameters”, “usage status” for each parameter, “acquired value” for each parameter, “acquisition destination of answer sentence”, etc. It is configured to be possible. The "task name" corresponds to the "status" value of the scenario record whose "task" value is "task TRUE" in the scenario table in FIG.

Of the items in the task table, the “acquired value” is initially empty. This is because the “acquired value” is stored according to the interaction with the user. Specifically, in step S10-5, when the current status is “refer to the task table of TETHERING”, the question statement generation unit 11 selects the “parameter” of the record whose “task name” is “TETHERING”. , The question sentence corresponding to the “use status” of the “parameter” in which the “acquired value” is not stored is acquired from the scenario table (FIG. 4). For example, in FIG. 10, the “parameters” of “OS” and “CONNECT” are defined for the task “TETHERING”, but initially, no “acquired value” is stored. The sentence generation unit 11 first obtains from the scenario table (FIG. 4) the question sentence “Which terminal are you using?” corresponding to “TETHERING_OS”, which is the “usage status” corresponding to “OS”. Then, the dialogue sentence output unit 14 outputs the question sentence in step S3 of FIG.

Therefore, from the user's point of view, when the user inputs the character string “No” in response to the question “Is this tethering method! By the way, do you know tethering itself? "Is a function that allows you to connect your PC to the Internet using your smartphone's 3G line or LTE by turning your smartphone into a mobile router! The question text "Which terminal are you using?... "is output. That is, two question sentences are continuously output.

In addition, when the user inputs the character string of "Yes" to the question sentence "Is tethering method known, do you know tethering itself? Yes/No" in the "THETHERING" status. As shown in FIG. 4, the “jump destination status” does not include “EXPLAIN( )” and only “refer to the task table of TETHERING”, and thus “refer to the task table of TETHERING” is executed. In this case, the question statement generation unit 11 sequentially searches for “usage status” of “parameter” for which “acquired value” is not stored in the task table of FIG. Further, the question sentence corresponding to the “use status” of the “parameter” in which the “acquired value” is not stored is acquired from the scenario table (FIG. 4). For example, when the “acquired value” corresponding to the parameter “OS” is not stored, the question sentence generation unit 11 corresponds to “TETHERING_OS” that is the “use status” of the parameter “OS” as described above. The question sentence to be asked is acquired from the scenario table (FIG. 4), and the question sentence “Which terminal are you using?

The method of acquiring the question sentence when the current status is “refer to the task table of TETHERING” is the same for step S10-8.

The character string input in step S5 for the question text output based on the task table is stored as the “acquired value” corresponding to the “parameter” that is the source of the question text in the task table of the question text. To be done.

Furthermore, in the scenario table (FIG. 4), the “jump destination status” of the “TETHERING_OS” status is “see the TETHERING task table”. Therefore, in this case, the task table of FIG. 10 is referred to again. When the task table is referred to for the second time, in step S10-5 or S10-8, the question sentence generation unit 11 obtains the question sentence corresponding to “TETHERING_CONNECT” that is the “usage status” of the parameter “CONNECT”. Then, in step S10-9, the question sentence is output to the dialogue control unit 13. Note that, in FIG. 10, for convenience, the scenario record of the “TETHERING_CONNECT” status is omitted.

In this case, the character string for the question sentence is input by the user in step S5 of FIG. 3, and the character string is stored in the task table (FIG. 10) as an acquired value for “CONNECT” in FIG. As a result, all the parameters corresponding to the "TETHERING" task have been acquired.

In this state, the determination in step S9 of FIG. 3 is Yes. In the task table (FIG. 4), the value of “task” has already been changed to “TETHERING” status which is “task TRUE”, and acquisition of all the parameters corresponding to the task related to “TETHERING” status has been completed. Is. In this case, steps S11 and S12 are executed.

Note that there may be a plurality of task tables stored in the dialogue scenario storage unit 161 other than THETHERING, such as those relating to Wi-Fi (registered trademark) and other ICT terms. In this case, a status corresponding to each task table (for example, “See task table of Wi-Fi (registered trademark)” or the like) may be set in the “jump destination status” of the scenario table (FIG. 4 ).

Next, details of step S11 will be described. FIG. 11 is a flowchart for explaining an example of the processing procedure of the response sentence generation processing.

In step S<b>11-1, the response sentence generation unit 15 requests the knowledge amount estimation unit 151 to estimate the user's knowledge amount.

Then, the character input speed calculation unit 1511 of the knowledge amount estimation unit 151 calculates the character input speed by referring to the character input speed storage unit 162 (FIG. 5) (S11-2). For example, the speed (the number of input characters per second) is calculated for each character string by counting the number of characters for each character string input by the user and dividing the number of characters by the time required for the character string. Further, the character input speed calculation unit 1511 calculates the average value of the speeds of all the character strings input so far as the character input speed.

For example, assuming that the stored contents of the character input speed storage unit 162 are as shown in FIG. 5, the character input speed is calculated as follows.
No. 1 speed=12 characters/10 seconds=1.2 characters/second No. 2 speed=3 characters/3 seconds=1 character/second No. 3 speed=5 characters/3 seconds=1.667 characters/second Character input speed=(1.2+1+1.667)/3=1.289
Subsequently, the term comprehension degree calculation unit 1512 of the knowledge amount estimation unit 151 refers to the term comprehension storage unit 163 (FIG. 8) to calculate the term comprehension degree (S11-3).

FIG. 12 is a diagram showing a first example of the method of calculating the term comprehension level. In FIG. 12, the result of determination as to whether or not the term is understood (no understanding = 0, understanding = 1) is on the vertical axis, and the difficulty is on the horizontal axis on the coordinates, and the storage content of the term understanding storage unit 163 (FIG. 8). Is plotted, and an example in which the maximum difficulty below the inflection point is used as the term comprehension level is shown by approximating the logistic curve. In the example of FIG. 12, the calculation result of the term comprehension degree is 2.

The method of calculating the term comprehension level is not limited to this. FIG. 13 is a diagram showing a second example of the method of calculating the term comprehension level. In FIG. 13, an example is shown in which, as a result of plotting the storage content of the term understanding storage unit 163 (FIG. 8) on the same coordinates as in FIG. 12, the maximum difficulty level where 0 does not exist is the term understanding level. Also in this case, the calculation result of the term comprehension degree is 2.

Subsequently, the knowledge amount estimation unit 151 estimates the knowledge amount using the character input speed calculated by the character input speed calculation unit 1511 and the term comprehension level calculated by the term comprehension level calculation unit 1512 (S11-4). .. The knowledge amount is an index indicating the level of the user's knowledge of a field (ICT field) related to the user's initial question or a term of the field.

FIG. 14 is a diagram showing a first example of the knowledge amount estimation method. In FIG. 14, a correspondence table in which the knowledge amount corresponding to the character input speed and the term comprehension degree is registered is prepared, and the calculated character input speed and term comprehension degree are applied to the correspondence table to estimate the knowledge amount. An example is shown (doing). The correspondence table may be stored in the auxiliary storage device 102 or the like in advance, for example.

In addition, FIG. 15 is a diagram illustrating a second example of the knowledge amount estimation method. In FIG. 15, there is an example in which a calculation formula in which the character input speed and the term comprehension degree are variables is prepared, and the knowledge amount is calculated based on the calculation result obtained by applying the calculated character input speed and term comprehension degree to the calculation formula. It is shown.

14 and 15, both the character input speed and the term comprehension level are discrete values, but the character input speed and the term comprehension level may be continuous values. The knowledge amount is also a discrete value in the present embodiment, but may be a continuous value. That is, the knowledge amount may be any value as long as the knowledge amount can be estimated based on the character input speed and the term comprehension level.

Subsequently, the answer sentence generation unit 15 acquires the answer sentence from the acquisition destination of the answer sentence corresponding to the task to be executed (S11-5). The acquisition source can be specified by referring to the value of “acquisition source of answer sentence” in the task table (FIG. 10) corresponding to the task. In the example of FIG. 10, the “acquiring source of the answer sentence” for the task “TETHERING” is “TETHERING_R(OS, CONNECT)”. Here, “TETHERING_R” means the answer sentence table stored in the dialogue scenario storage unit 161 in association with the task of “TETHERING”. “(OS, CONNECT)” is information for identifying one record from a plurality of records stored in the answer statement table, and each value of “OS” and “CONNECT” is a parameter. Is replaced with a concrete acquired value corresponding to.

FIG. 16 is a diagram showing an example of the structure of the TERRORING answer sentence table in the dialogue scenario storage unit 161.

As shown in FIG. 16, the “answer sentence” is stored in the answer sentence table in association with the “condition”. The “condition” is a condition for which the corresponding “answer sentence” is to be acquired, and corresponds to the value of “acquisition destination of answer sentence” in the task table. The “answer sentence” is a character string that constitutes the answer sentence.

For example, when the “obtained value” for the parameter “OS” of the task “TETHERING” is “OS-B” and the “obtained value” for the parameter “CONNECT” is “Wi-Fi (registered trademark)”, In 16, the answer sentence stored in the first record is acquired.

Subsequently, the response sentence generation unit 15 determines whether the knowledge amount estimated in step S11-4 is the highest level at the defined stage (S11-6). For example, in the present embodiment, the difficulty level is defined in 5 levels of 1 to 5, and 5 is the highest level. If the knowledge amount is the highest level, the process proceeds to step S11-9, and if not, the process proceeds to step S11-7.

In step S11-7, the response sentence generation unit 15 refers to the term understanding storage unit 163 (FIG. 8) and refers to a term having a degree of difficulty higher than the knowledge amount level (hereinafter, referred to as “conversion target term”) from the response sentence. .) is extracted. For example, when the answer sentence of the first record in the answer sentence table of FIG. 16 is acquired, the term “Wi-Fi (registered trademark)” with a difficulty level of 2 and the “4” with a difficulty level of 4 are included in the answer sentence. The term “SSID” and the term “tethering” with a difficulty level of 4 are included. In this case, when the user's knowledge amount is 3, "Wi-Fi (registered trademark)" whose difficulty level is equal to or lower than the user's knowledge amount level is not extracted as a conversion target term, and the knowledge amount level is “SSID” and “tethering”, which have a higher degree of difficulty, are extracted as conversion target terms.

Subsequently, the answer sentence generation unit 15 refers to the term explanation storage unit 164 (FIG. 9) to acquire the term explanation sentence of the conversion target term, and includes the term explanation sentence in the answer sentence to answer the user's opening question. An answer sentence indicating a system answer is generated (S11-8). For example, "... 3. Press tethering..." is replaced with "... 3. Tethering (By converting your smartphone to a mobile router, you can use a 3G line or LTE to connect a computer, etc. A term explanation may be included immediately after the conversion target term, such as "Pressing a function capable of connecting to the Internet)". Note that the answer sentence generation unit 15 may generate a natural sentence including a term explanation sentence immediately after the conversion target term without using the answer sentence table.

In step S11-9, the response sentence generation unit 15 outputs the response sentence to the dialogue control unit 13. The dialogue control unit 13 outputs the answer sentence to the dialogue sentence output unit 14. As a result, the dialogue sentence output unit 14 outputs the answer sentence in step S12 of FIG.

In the above, instead of the character input speed, the number of typographical errors (for example, the number of deletions of a character) or the character input time (absolute time instead of speed) is information (index) indicating the status of the response input operation. ).

Further, in the above description, an example in which the user's input question and the question sentence are input by the user in the form of a character string, and the output of the question sentence and the answer sentence are displayed in the character string on the display device 106 has been described. It may be input or output. For example, the user opening question may be entered by the user's utterance. In this case, the text input unit 12 inputs the text obtained as a result of recognition of the user's utterance by a voice recognition unit (not shown) instead of the character string input by the user, and completes after the user starts the utterance. Measure the time required to do so. The start and completion times of the user's utterance are, for example, the timing at which the user's utterance starts when the last output process by the system is completed, and a predetermined operation on the input device 107 (for example, pressing of a “send” button) It is possible to detect when time is the timing of completion of utterance. In this way, instead of the character string input by the user, the text obtained as a result of recognition of the user's utterance by a voice recognition unit (not shown) may be used as the character string. Further, the output of the question sentence or the answer sentence by the dialogue device 10 may be outputted in a form in which the agent of the dialogue device 10 speaks as a voice.

As described above, according to the present embodiment, with respect to the user with whom the user is interacting, the input information indicating whether or not the term is understood, and the non-linguistic information of the user (the input operation status of the user answer such as character input speed). Based on, the level of knowledge (knowledge amount) of the user in the field related to the question from the user (user initial question) can be estimated. Therefore, the accuracy of estimating the knowledge of the user of the conversation partner can be improved, and a more appropriate conversation (generation of answer sentence) can be realized. For example, it is possible to estimate even what the user's known and unknown terms are.

Note that this embodiment may be applied to the estimation of knowledge in fields other than ICT.

In addition, in the present embodiment, the text input unit 12 is an example of a reception unit. The dialogue sentence output unit 14 is an example of an output unit. The knowledge amount estimation unit 151 is an example of an estimation unit. The answer sentence generation unit 15 is an example of a generation unit.

Although the embodiments of the present invention have been described in detail above, the present invention is not limited to such specific embodiments, and various modifications are possible within the scope of the gist of the present invention described in the claims. It can be modified and changed.

10 Dialogue Device 11 Question Sentence Generation Unit 12 Text Input Unit 13 Dialogue Control Unit 14 Dialogue Sentence Output Unit 15 Response Sentence Generation Unit 100 Drive Device 101 Recording Medium 102 Auxiliary Storage Device 103 Memory Device 104 CPU
105 Interface device 106 Display device 107 Input device 111 Choice determination unit 112 Term understanding determination unit 151 Knowledge amount estimation unit 161 Dialog scenario storage unit 162 Character input speed storage unit 163 Term understanding storage unit 164 Term explanation storage unit 1511 Character input speed calculation unit 1512 Term comprehension calculation section B bus

Claims (4)

A reception unit that receives input of the first question from the user,
An output unit for outputting a second question asking the user whether he/she understands the term related to the first question;
An estimation unit that estimates the level of knowledge of the user in the field related to the first question based on the content of the answer to the second question and the situation of the input operation of the character string indicating the answer;
A generation unit that generates the content of the answer to the first question according to the level estimated by the estimation unit;
An interactive device comprising: The generation unit is configured so that the user understands, among the terms included in the answer to the first question, based on the difficulty level stored in the storage unit for each term and the level estimated by the estimation section. Identifying a term that is presumed not to exist, and including an explanatory note stored in the storage unit for the identified term in the answer to the first question,
The dialogue apparatus according to claim 1, characterized in that: The estimation unit estimates the level based on the difficulty level stored in the storage unit for the term understood by the user, which is specified based on the content of the answer to the second question. ,
The dialogue apparatus according to claim 1 or 2, characterized in that. A reception procedure for receiving the input of the first question from the user,
An output procedure for outputting a second question asking the user whether he/she understands the term related to the first question,
An estimation procedure for estimating the level of knowledge of the user in the field related to the first question based on the content of the answer to the second question and the situation of the input operation of the character string indicating the answer;
A generation procedure for generating the content of the answer to the first question according to the level estimated in the estimation procedure;
A program characterized by causing a computer to execute.

Images