JP6223739B2 - Function execution instruction system and function execution instruction method - Google Patents

Description

  The present invention relates to a function execution instruction system, a function execution instruction method, and a function execution instruction program that can instruct execution of one or more functions.

  Conventionally, a system that recognizes a user's voice and executes a function based on the recognized voice is known. For example, Patent Document 1 describes that functions such as setting a registered location as a destination and viewing a map of the registered location are executed based on voice recognized by the car navigation device.

JP 2006-23444 A

  Here, a voice agent system is considered in which a plurality of functions (tasks) can be executed and a user can give an utterance instruction (instruction by voice) multiple times. In such a conventional system, it cannot be determined whether the uttered content is an utterance of condition addition or change in the previous task or an utterance of a new requirement unrelated to the previous task.

  For example, if the user speaks to the system as “transfer to Tokyo”, the system checks the transfer from the user's nearest station to Tokyo station (a transfer guidance application is activated). After that, it is assumed that the user speaks “Yokohama” because he wants to know the transfer to Yokohama Station instead of the transfer to Tokyo Station. At this time, if the task to be newly executed is determined from the user's utterance content, the utterance “Yokohama” does not include the keyword for executing the task of checking the transfer “transfer”. For example, a task of examining a map of Yokohama that is activated only from a place name is executed instead of a task of examining a transfer.

  On the other hand, if the task for checking the transfer, which is the previous task, has been taken over, it is possible to check the transfer to Yokohama Station as desired by the user. However, if the task is to be taken over uniformly in this way, even if the user says `` curry '' because he wants to check the curry store after uttering `` transfer to Tokyo '', The task for checking the transfer is executed again. As described above, if a task is simply executed by a word that is input, an appropriate task cannot be executed by a word that is continuously input.

  The present invention has been made in view of the above problems, and an object of the present invention is to provide a function execution instruction system and a function execution instruction method that can appropriately execute a function using words that are continuously input. And

In order to achieve the above object, a function execution instruction system according to the present invention includes a function execution instruction means for instructing execution of one or more functions related to a category preset for each function, and information including words A word input means for inputting a word, a category specifying means for specifying a category of a word included in the information input by the word input means based on a word other than the word included in the information, and a category specifying means An execution function determination unit that determines a function that is instructed to be executed by the function execution instruction unit from one or more functions according to the category, and the execution function determination unit includes a category identified by the category identification unit, And based on the category related to the function that has been instructed by the function execution instructing means in the past, Input but if the execution of the indicated function is determined whether an instruction therefrom, it is determined that execution to the past by the function execution instruction means does not now instruct the execution of the indicated function, by the word input means Based on the word included in the information, the function to be instructed for execution is newly determined .

  In the function execution instruction system according to the present invention, it is determined whether to instruct execution of a function that has been instructed in the past based on the category related to the function and the category of the input word. Accordingly, for example, it is appropriately determined whether a function executed in the past should be executed subsequently or a new function should be executed. Further, in the function execution instruction system according to the present invention, the category of the input word is specified based on the word other than the word included in the information including the word, and thus the category is appropriately specified. That is, according to the function execution instructing system according to the present invention, it is possible to appropriately execute the function by the continuously input words.

  The category specifying means is based on information indicating the degree to which a word to be specified included in the information input by the word input means and a word other than the word included in the information are included in the same information. It is good also as specifying the category of a word. According to this configuration, the category of the word to be specified can be reliably specified using a word other than the word to be specified, and the present invention can be implemented reliably and appropriately.

  The function execution instruction system further includes an execution function storage means for storing a function instructed to be executed by the function execution instruction means, and the execution function determination means stores the category specified by the category specification means and the execution function storage means. Determining whether or not to instruct execution of a function that has been instructed to be executed by the function execution instructing means in the past based on a category related to the function that has been instructed to be executed by the function execution instructing means in the past. It is good.

  Whether the execution function determining means instructs the execution of the function instructed to be executed by the function execution instructing means immediately before a word related to the category is input by the word input means according to the category specified by the category specifying means. It may be determined whether or not. According to this configuration, for example, it is appropriately determined whether the function executed immediately before should be executed continuously or a new function should be executed, and the function can be appropriately executed by continuously input words. it can.

  The function that is instructed to be executed by the function execution instructing unit includes a function that is included in the information input by the word input unit and that inputs a word of a category related to the function. It is determined whether or not the words of the category specified by the category specifying means can be input to the function for which execution has been instructed by the function execution instructing means in the past, and the function execution instructing means in the past according to the determination It is also possible to determine whether or not to instruct execution of the function instructed to execute by. According to this configuration, together with the execution of the function, the input word can be used as an input to the function, and the convenience for the user can be improved. For example, in combination with the execution of the transfer guidance function described above, guidance based on the input station name can be performed. Further, it is possible to determine whether or not to execute a function executed in the past more appropriately, and it is possible to more appropriately implement an embodiment of the present invention.

  In addition, the execution function determining means determines whether the category specified by the category specifying means matches the category of the word that has been input to the function for which execution has been instructed in the past. In such a case, it may be determined that the execution of the function will be instructed. According to this configuration, the embodiment of the present invention can be more appropriately implemented.

  The function execution instruction system may further include voice recognition means for inputting voice, performing voice recognition on the input voice, and inputting the result of voice recognition to the word input means. According to this configuration, the function can be executed by the user's voice.

  By the way, the present invention can be described as the invention of the function execution instruction system as described above, as well as the invention of the function execution instruction method as follows. This is substantially the same invention only in different categories, and has the same operations and effects.

That is, the function execution instruction method according to the present invention is a function execution instruction method that is an operation method of the function execution instruction system, and instructs execution of one or more functions related to a preset category for each function. A function execution instruction step, a word input step for inputting information including a word, and a category for specifying a category of words included in the information input in the word input step based on a word other than the word included in the information An execution function determination step for determining a function to be executed in the function execution instruction step from one or more functions according to the category specified in the category specification step; , The category specified in the category specifying step, and executed in the function execution instruction step in the past Based on the categories associated with the indicated function, it determines whether to now direct the execution of the functions executed in the function execution instruction step in the past has been instructed, executed in the function execution instruction step to the last instruction When it is determined that the execution of the performed function is not instructed from now on, a function to be instructed for execution is newly determined based on the word included in the information input in the word input step .

  In the present invention, based on the category related to (associated with) the function and the category of the input word, it is determined whether or not to instruct execution of the function that has been instructed in the past. Accordingly, for example, it is appropriately determined whether a function executed in the past should be executed subsequently or a new function should be executed. In the present invention, since the category of the input word is specified based on a word other than the word included in the information including the word, the category is appropriately specified. In other words, according to the present invention, it is possible to appropriately perform the function by the continuously input words.

It is a figure which shows the structure of the function execution instruction | indication system which concerns on embodiment of this invention. It is a figure which shows the correspondence of the task and slot which are performed in a function execution instruction system. It is a table | surface which shows the information for specifying a category from a word. It is a table | surface which shows the score for every task when a user's utterance is what is "transfer to Tokyo." It is a table | surface which shows the score for every task when a user's utterance is what is still "Yokohama." It is a table | surface which shows the score for every task when a user's utterance is what is "curry." It is a table which shows the co-occurrence probability memorize | stored in a function execution instruction | indication system. It is a figure which shows the hardware constitutions of the function execution instruction | indication server and speech recognition server which comprise the function execution instruction | indication system which concerns on embodiment of this invention. It is a flowchart which shows the process (function execution instruction method) performed with the function execution instruction system which concerns on embodiment of this invention. It is a flowchart which shows the specific process of the category of a word.

  DESCRIPTION OF EMBODIMENTS Hereinafter, embodiments of a function execution instruction system and a function execution instruction method according to the present invention will be described in detail with reference to the drawings. In the description of the drawings, the same elements are denoted by the same reference numerals, and redundant description is omitted.

  FIG. 1 shows a function execution instruction system 1 according to this embodiment. The function execution instruction system 1 includes a function execution instruction server 10 and a voice recognition server 20. The function execution instruction system 1 is a system that instructs the communication terminal 30 to execute one or more preset functions. Functions to be instructed to be executed from the function execution instruction system 1 to the communication terminal 30 are, for example, transfer guidance, telephone directory search, gourmet search, image search, music search, music player, scheduler, mail, memo, route guidance, Map search is equivalent. In the present embodiment, the function to be executed is called a task. Specifically, with regard to transfer guidance, telephone directory search, gourmet search, image search, music search, and map search, for example, information on the guidance and search results is displayed and output at the communication terminal 30.

  In addition, with regard to mail creation, music player, scheduler, incoming / outgoing calls, cameras, etc., application programs for executing these tasks are started in the communication terminal 30. Further, depending on the content of the instruction, an operation after the application program is started is also performed. For example, in the mail creation task, information is entered into the mail text and destination.

  The communication terminal 30 is a device that can communicate with the function execution instruction server 10 and the voice recognition server 20 through a communication network (for example, a mobile communication network), such as a mobile phone or a PC (Personal Computer). It corresponds to. The communication terminal 30 may be a terminal equipped with a car navigation function used in a vehicle such as an automobile. In the present embodiment, the communication terminal 30 will be described as a terminal having a car navigation function. For example, the communication terminal 30 is configured to be detachable from the vehicle, and can exhibit a car navigation function when mounted and mounted on the vehicle. Note that the communication terminal 30 may be able to exhibit the car navigation function even when it is not attached to the vehicle (for example, when it is simply placed in the vehicle or carried by a user who is on the vehicle). Good.

  Since the above task is executed with the user's voice as a trigger, the communication terminal 30 has a function of inputting the user's voice. In addition, the communication terminal 30 has a function of executing a task in response to an instruction from the function execution instruction system 1, for example, an information reception function, an information processing function, a display function, and the like.

  That is, the communication terminal 30 has a function execution unit that executes one or more (or a plurality of) tasks. Specifically, the communication terminal 30 stores an application program corresponding to the task in advance, and executes the task by executing (starting) the stored application program. The task executed by the communication terminal 30 is instructed by the function execution instruction server 10 as described later. In addition to executing the application, the task may be executed by acquiring information corresponding to the task from the network.

  The function execution instruction system 1 instructs the communication terminal 30 to execute a task as follows. First, the communication terminal 30 inputs a user's voice. The input voice is transmitted from the communication terminal 30 to the voice recognition server 20. The voice recognition server 20 receives voice from the communication terminal 30 and performs voice recognition on the received voice. The voice recognition server 20 transmits a sentence or a set of words as a result of the voice recognition to the communication terminal 30. The communication terminal 30 receives the result of voice recognition and further transmits it to the function execution instruction server 10. The function execution instruction server 10 receives the result of the voice recognition, determines a task to be executed based on the result of the voice recognition, and instructs the communication terminal 30 to execute the task. For example, the function execution instruction server 10 transmits information indicating a task to be executed to the communication terminal 30. The communication terminal 30 receives an instruction from the function execution instruction server 10 and executes a task according to the instruction. Note that a set of sentences or words as a result of voice recognition may be directly transmitted from the voice recognition server 20 to the function execution instruction server 10 without being sent to the communication terminal 30 once. Furthermore, the function execution instruction server 10 and the voice recognition server 20 may be integrated.

  For example, when the user of the communication terminal 30 speaks “Transfer at Tokyo Station”, the function execution instruction server 10 executes the task of transfer guidance from the nearest station to the Tokyo station at the current position of the communication terminal 30. The communication terminal 30 is instructed to execute the transfer guidance task. When the transfer guidance task is executed as described above, the current position of the communication terminal 30 is acquired when the task is executed. As described above, additional information is acquired as necessary depending on the task. The above is the outline of the function execution instruction system 1 according to the present embodiment.

  Subsequently, functions of the function execution instruction system 1 according to the present embodiment will be described. The voice recognition server 20 is a device that is a voice recognition unit that inputs voice, performs voice recognition on the input voice, and outputs a result of the voice recognition. Specifically, as described above, the voice recognition server 20 receives voice data from the communication terminal 30. The voice recognition server 20 has a voice recognition engine, and performs voice recognition using the voice recognition engine. For speech recognition itself, any conventional speech recognition method can be used. The voice recognition server 20 acquires the voice recognition result as a set of words (a group of words including one or more words) or a sentence and transmits the result to the communication terminal 30.

  As shown in FIG. 1, the function execution instruction server 10 includes a function execution instruction unit 11, a word input unit 12, a category identification unit 13, an execution function determination unit 14, and an execution function storage unit 15. Is done.

  The function execution instructing unit 11 is a function execution instructing unit that instructs the communication terminal 30 to execute one or more tasks. Specifically, the function execution instructing unit 11 instructs the execution of the task by transmitting a command for executing the task to the communication terminal 30. The task whose execution is instructed by the function execution instructing unit 11 is determined by the execution function determining unit 14 as described later.

  A task is related to (associated with) a category set in advance for each task. The category, for example, classifies words for each attribute. For example, the categories are “station (station name)”, “commercial facility name”, “overseas travel”, “dishes name”, and “person name”. For example, in the case of a transfer guidance task, it is related to (associated with) the category of “station”. Further, if it is a gourmet search task, it is related to (associated with) the category of “dishes name”. In the case of a telephone directory search task, it is related to (associated with) the “person name” category. The task may be associated with a preset word. For example, a task of transfer guidance is associated with the word “transfer”. These associations are input and stored in the function execution instruction server 10 in advance by, for example, an administrator of the function execution instruction system 1 or the like. These associations and associations are used to determine a task to be executed as will be described later. The specific usage will be described later.

  In addition, the task instructed to execute has a function of inputting words (arguments). This is called a task slot. The slot is predetermined. For example, as shown in FIG. 2, in the case of a transfer guidance task, the task is executed with the departure station and arrival station as inputs to the slot. Alternatively, in the case of a gourmet search task, a store relating to the input dish is searched using the dish name as an input to the slot. In the case of a telephone directory search task, information such as a telephone number associated with the input person name is searched using the person name as an input to the slot. Note that zero or more slots are defined for the task. That is, some tasks do not have slots.

  As shown in FIG. 2, each slot is associated with information of “slot type”, “size”, “essential?”, And “takeover?”. The “slot type” is information specifying a category (a type of a word that can be accepted in the slot) input to the slot, and indicates (holds), for example, category ID information. That is, a word category is associated with the slot. In the example of FIG. 2, “type_station” is held as the slot type of the slot corresponding to the arrival station and the departure station. This means that “station” can be input as the word category for the slot. Represents. The slot type may be configured such that a word can be input when the word is a specific expression. For example, “to” corresponds to “to” in the case of expressions such as “to” (equivalent to to. * Type_station of the arrival station in FIG. 2) and “to from” (equivalent to from. * Type_station of the arrival station in FIG. 2). A word may be input into a slot. Further, the “slot type” does not have to specify a specific category, and a word of an arbitrary category may be input. “Size” indicates the number of words input to the slot. “Mandatory?” Indicates whether or not input of a word is essential when executing the task. If there is no word to be input to the True slot in the “required?” Field, the slot content may be returned without executing the task, and the user may be prompted to input to the slot. .

  “Takeover?” Indicates whether or not the input to the slot is taken over when the task is continuously executed, as will be described later. Each slot may be associated with information indicating whether a preset word (default word) is input when no word is input to the slot. For example, in the case of transfer guidance, for the departure station, the station name of the station closest to the current position of the user acquired by GPS (Global Positioning System) or the like may be input as a default word.

  The word input unit 12 is word input means for inputting information including words. Specifically, the word input unit 12 receives the information from the communication terminal 30 by receiving information indicating one or more words as a result of speech recognition by the speech recognition server 20. Alternatively, when the result of speech recognition by the speech recognition server 20 is a sentence (input sentence), the word input unit 12 receives the sentence and extracts the word from the sentence, or cuts the document into words. Get a word. In addition, the method of acquiring a word from sentences can use arbitrary conventional methods, such as a morpheme analysis using a morpheme dictionary. The word input to the word input unit 12 is a user's utterance unit for the communication terminal 30. That is, a word group included in one utterance is treated as one unit. The word input unit 12 inputs information indicating the acquired word to the category specifying unit 13.

  The category specifying unit 13 is a category specifying unit that specifies the category of the word input from the word input unit 12. Specifically, for example, the category specifying unit 13 stores information (dictionary) for specifying the category of the word in advance as shown in FIG. 3, and specifies the category based on the information. This information is previously input and stored in the function execution instruction server 10 by, for example, an administrator of the function execution instruction system 1 or the like. In this information, as shown in FIG. 3, a word is associated with a category ID and a category tag (category name) that are information for specifying a category. One category ID and category tag are associated with one or more words. For example, for the category whose category ID is “type_station” and whose category tag is “station”, the words “Osaka”, “Osaka Station”, “Tokyo”, “Tokyo Station”, “Yokohama”, and “Yokohama Station” correspond. It is attached.

  In the above dictionary, the same word may be associated with a plurality of categories. For example, words such as Tokyo, Shibuya, Shinagawa, Yokohama, etc. are associated with the “station” category. In addition, words such as Enomoto, Kobayashi, Shibuya, etc. are associated with the “person name” category. As in the above example, the word “Shibuya” may be associated with both the “station” category and the “person name” category.

  The category specifying unit 13 determines whether or not each word input from the word input unit 12 matches the word included in the information shown in FIG. The attached category is specified as the category of the word input from the word input unit 12. Further, the category specifying unit 13 may specify the category by, for example, a technique of named expression extraction (Named Entity Extraction) without using the information indicating the correspondence between the category and the word as shown in FIG. . Moreover, it is good also as specifying a category by chunking (chunking). Further, the category specifying unit 13 may store a regular expression in advance and specify the category depending on whether or not the word matches the stored regular expression. For example, if a word matches a regular expression, the word is identified as a category that matches the regular expression. Alternatively, the category specifying unit 13 may acquire the phone book data stored in the communication terminal 30 and specify the category by determining whether or not the word is a word included in the phone book data. For example, when a word is included in the phone book data, the word is specified as a category of words existing in the phone book data. The category identification unit 13 outputs information indicating the word input from the word input unit 12 and the category of the identified word to the execution function determination unit 14. Note that it is not always necessary to specify a category for each word, and there may be a word for which no category is specified.

  The execution function determination unit 14 is an execution function determination unit that determines a task whose execution is instructed by the function execution instruction unit 11 according to the category specified by the category specification unit 13. For example, the execution function determination unit 14 determines a task that is instructed to be executed based on the association (association) between a task and a category as described above. Specifically, the score of the category is stored for each task, the score of the word input in the user's utterance unit is specified, and the task to be instructed to execute is determined based on the specified score. The score of each category is determined according to the degree of association with the task.

  For example, it is assumed that the user's utterance is “transfer to Tokyo”. In this case, as shown in FIG. 4, the user's utterance is divided into the words “Tokyo”, “To”, “No”, and “Transfer”. And the score of the category matched with each word for every task is calculated | required, and those are totaled and the score for every task is calculated. As shown in FIG. 4, the score of each word is specified, and by summing them, the score of each task (in the example of FIG. 4, "transfer guide", "gourmet search", "map search" tasks) is totaled. The The score may be determined not only for the category but also for each word. For example, since the word “transfer” is strongly associated with the task “transfer guidance”, the task “transfer guide” is associated with the score of the word “transfer” itself. In this way, a task to be instructed to execute may be determined based on a word input by the word input unit 12 and a preset word.

  For example, the execution function determination unit 14 determines a task having the highest score as a task whose execution is instructed by the function execution instruction unit 11. The execution function determination unit 14 notifies the function execution instruction unit 11 of the determined task. When this notification is given, the function execution instructing unit 11 instructs the communication terminal 30 to execute the task. In addition, at the time of this notification, among the words input by the word input unit 12, a word that is input to the slot of the task to be executed (a word whose category matches the category of the slot) is also notified. Also good. A word to be input to the slot is also transmitted from the function execution instruction unit 11 to the communication terminal 30 together with the task execution instruction. For example, when the “transfer guide” task is executed by the utterance “transfer to Tokyo”, the word “Tokyo” in the “station” category is associated with the word in the “station” category. It is used as an input for a station slot. In this case, in the slot of the departure station, the station name of the station closest to the current position of the user is input as a default word. Thereby, the transfer from the nearest station to the Tokyo station from the current position of the user is searched, and the result is presented to the user. In addition, when the input to the slot is insufficient in executing the task, the user may be prompted to input to the slot at that time, and the word to be input to the slot may be input again.

  As described above, the execution function determining unit 14 determines a task to be instructed to be executed by the function execution instructing unit 11 from now on. The determination of this task is assigned to the task instructed to be executed by the function execution instructing unit 11 in the past. It may be based on. Specifically, it is determined whether or not a task for which execution has been instructed in the past is to be executed subsequently. For example, if execution of the task of “transfer guidance” has been instructed in the past (the task has been executed), and then a word input based on the user's utterance has been input to the function execution instruction server 10, It is determined whether to instruct the execution of the task of “transfer guidance” continuously. A configuration for this will be described below.

  The execution function storage unit 15 is an execution function storage unit that stores a task instructed to be executed by the function execution instruction unit 11. For example, the execution function storage unit 15 inputs information (for example, task name) indicating a task for which execution has been instructed from the function execution instruction unit 11, and stores information indicating the input task. At this time, the task execution instruction time may be stored together. The execution function storage unit 15 also inputs a word input to the slot at the time of task execution instruction from the function execution instruction unit 11, and stores the word together with the executed task. Also good. The information stored in the execution function storage unit 15 is appropriately referred to by the execution function determination unit 14 as follows.

  The execution function determination unit 14 is a category related to (associated with) the category specified by the category specification unit 13 and the task stored in the execution function storage unit 15 and instructed to be executed by the function execution instruction unit 11 in the past. Based on the above, it is determined whether or not to instruct execution of a task for which execution has been instructed by the function execution instructing unit 11 in the past. At this time, the task for determining whether or not to instruct execution is the task that is instructed to be executed by the function execution instructing unit 11 immediately before a word is input to the function execution instructing server 10 (one time before. It may be determined whether or not to instruct execution of a task for which execution is instructed. That is, a task for which execution has been instructed two or more times before may not be a target of a task for which execution is instructed continuously. Furthermore, only tasks that are instructed to be executed within a predetermined time (for example, within 1 hour) set in advance from the word input may be set as candidate tasks to be instructed to be executed subsequently.

  Specifically, the execution function determination unit 14 includes a category related to (associated with) a task that has been instructed by the function execution instruction unit 11 in the past and stored in the execution function storage unit 15, and a category specification unit 13. It is determined whether or not the category specified by (1) matches, and according to the determination, it is determined whether or not to instruct execution of the function that has been instructed by the function execution instructing unit 11 in the past. For example, the execution function determining unit 14 matches the category of the word specified by the category specifying unit 13 with the category associated with the slot of the task whose execution has been instructed by the function execution instructing unit 11 in the past (category The above determination is made by determining whether the word specified by the specifying unit 13 can be input to the task slot. If it is determined that the category of the word specified by the category specifying unit 13 matches the category associated with the slot of the task for which execution has been instructed by the function execution instructing unit 11 in the past, the execution instruction in the past If it is determined that the task to be executed is instructed to be executed continuously, and it is determined that the tasks do not coincide with each other, it is determined that the instruction to continue execution of the task is not performed, that is, the execution of a new task is instructed.

  At this time, the execution of the task is instructed only when the category of the word specified by the category specifying unit 13 coincides with all the categories associated with the task slots instructed to be executed by the function execution instructing unit 11 in the past. It may be possible to instruct the execution of the task when it matches the category associated with a part of the slot. Further, only when the category of the word specified by the category specifying unit 13 matches the category associated with any of the task slots instructed to be executed by the function execution instructing unit 11 in the past, the task execution is continued. (That is, all the categories of the words specified by the category specifying unit 13 match the category corresponding to any of the task slots for which execution has been instructed by the function executing unit 11 in the past, and are input. The task may be instructed to continue only when there is no noun that does not correspond to any category in the sentence and therefore does not correspond to any slot. Some categories of the designated words are slotted for tasks for which execution has been instructed by the function execution instructing unit 11 in the past. May instruct the execution of the tasks subsequent when matching the categories marked with corresponding to any one of. If the slot of the task whose execution has been instructed in the past by the function execution instructing unit 11 is not of a type that specifies a category (if it is a type in which words of an arbitrary category can be input, for example, for the task of “image search” The “search word” slot) may be instructed to always execute a task regardless of which category of word is input, so the above determination may not be performed.

  However, even in this case, it is determined whether or not the category of the input word matches the category of the word input to the slot of the task for which execution has been instructed by the function execution instructing unit 11 in the past. When it is determined that they match (when the category of the word of the immediately preceding utterance matches the category of the input word), the execution of the task may be instructed subsequently. According to this configuration, it is possible to appropriately determine whether to take over a task even if the task can input a word of an arbitrary category in the slot.

  Further, when the execution function determination unit 14 determines that a task for which execution has been instructed in the past by the function execution instruction unit 11 is to be executed (to continue the task), the category associated with the slot The function execution instructing unit 11 is instructed to execute the word that matches with the input to the slot. In addition, the execution function determination unit 14 sets, as a part of (or all) a task slot, the word input to the past slot stored by the execution function storage unit 15 as an input to the task. Also good. In this case, in each slot of the task shown in FIG. 2, even if the input of the word is performed only when the information in the “takeover?” Column is information indicating that the input of the word is inherited (True). Good. If the execution function determination unit 14 determines that the task is not continuously executed by the above determination, the execution function determination unit 14 determines an instruction to execute the task based on the score described above.

  When the execution of the “transfer guidance” task is instructed by the utterance “transfer to Tokyo” as in the example described above, “transfer guidance” is displayed in the execution function storage unit 15 as a task for which execution has been instructed in the past. The word “Tokyo” is stored as an input of the arrival station slot. Assume that the user's utterance after that is “Yokohama”. In this case, since the word “Yokohama” is the “station” category, it is determined that it matches the category associated with the slot of the arrival station of the “transfer guidance” task executed in the past. Therefore, it is determined that the “transfer guidance” task executed in the past is continuously instructed to be executed again. At that time, “Yokohama” is entered in the slot of the arrival station. As for the departure station, the station name of the nearest station from the current position of the user is inherited as in the case of “transfer to Tokyo” (the “takeover” column of the departure station is True as shown in FIG. 2). ing).

  On the other hand, a case will be described in which the score is determined by the utterance “Yokohama” without determining whether or not the execution function determination unit 14 continues to instruct execution of the task as described above. In this case, as shown in FIG. 5, the user's utterance is divided into the words “after all” and “Yokohama”. Then, as in the above-described example, the score of each word is specified for each task, and these are added together to calculate the score for each task. In this case, since the user's utterance does not include a word that is highly related to “transfer guidance”, the highest score is “map search” in which the score is high only by the place name “Yokohama”, not “transfer guidance”. It becomes. If the user utters “Yokohama” after the user utters “transfer to Tokyo”, the user wants to know the transfer to Yokohama station. When it is determined to take over the task as described above, the execution of the “transfer guidance” task is appropriately instructed. On the other hand, when a task to be executed independently of the user's utterance is determined as described above (unlike this embodiment), a task called “map search” that is different from the user's intention is executed.

  Further, it is assumed that the user's utterance after the utterance “transfer to Tokyo” is “curry”. In this case, since the word “curry” is the “cooking name” category, it is determined that the category does not match the category associated with the “transfer guidance” task slot for which execution has been instructed in the past. Accordingly, it is determined that the “transfer guidance” task that has been instructed to be executed in the past is not executed again.

  In this case, the execution function determination unit 14 determines a task instructing execution by calculating a score based on an utterance “curry”. As shown in FIG. 6, the user's utterance is composed only of the word “curry”. Then, as in the example described above, the score of the category associated with each word is specified for each task, and these are added together to calculate the score for each task. In this case, “gourmet search” has the highest score, and it is determined that the task of “gourmet search” is executed. If the user utters “curry” after the user utters “transfer to Tokyo”, it is a completely different topic, so he would like to know information about “curry” instead of “transfer guidance”. As described above, in this case, the task score is not determined to be taken over, but the task score is calculated from the user's utterance, and the execution of the task of “gourmet search” is instructed.

  As another example, if the task that was instructed in the past was “Gourmet Search” for curry, and “Tokyo” was entered in the slot indicating the target area for the gourmet search, “Yappa Osaka” When “Tokyo” is overwritten with “Osaka”, it is determined that execution of “gourmet search” regarding curry is instructed.

  Note that task transition may be forcibly performed in a specific case. For example, when information indicating that a transition is made to a new task is received from the communication terminal 30 by a user operation on the communication terminal 30, the task transition may be forcibly performed (the task is not taken over). ).

  The determination of the task instructed to be executed by the execution function determination unit 14 may be performed as follows. When determining a task to be executed from an input word group (user's utterance), the score corresponding to the word is given for each task as described above, but it is not always necessary to calculate the score. For example, a task is associated with a word (category), and it is determined whether or not the word (category) associated with the task is included in the word group input for each task. If it is determined, the execution of the task may be instructed.

  If it is determined that there is a keyword (strong keyword) that is strongly related to a specific task in the input word or sentence (when the score is extremely high), the task that has been instructed to be executed in the past Instead of taking over, a task instructing execution may be determined from a newly input word or sentence. In other words, task transition may be forced. For example, if the word “transfer” is included, a highly relevant “transfer guidance” task is executed.

  In addition, even when the execution of a new task is instructed without taking over the task, that is, when the task transitions, the word input to the slot may be taken over. In this case, all words that can be taken over are extracted from the slot contents before task transition. In the task after switching, all slot contents are once empty. An attempt is made to input (fill) the taken-out transferable word into the slot of the task after switching (those words that cannot be input are discarded). The input word is input to the slot according to the slot input rules.

  When taking over tasks as described above, if words in an utterance belong to multiple categories (if they have multiple attributes) to determine whether or not to take over tasks, take over tasks appropriately. May not be a judgment. For example, the following situations can be considered. Based on the user's utterance of “Mr. Enomoto's phone number”, the function execution instruction system 1 according to the present embodiment executes a phone book search task, and Enomoto's phone number is checked by the phone book search. The user makes a call using the telephone number. The user waits for Ms. Enomoto in Shibuya on the phone, so after the end of the talk, he / she wants to execute the transfer guidance to Shibuya, and then utters “Go to Shibuya”.

  This utterance is divided into words such as “Shibuya: noun (“ person name ”category,“ station ”category”), up to: particle, go: verb ”. The function execution instruction system 1 according to the present embodiment identifies “Shibuya” included in the utterance as a word in the “person name” category. Since the “Person Name” category is associated with the previously executed phone book search task, the function execution instruction system 1 according to the present embodiment determines that it will take over the phone book search task, and “Shibuya”. Phonebook search is executed using In this case, “Shibuya” belongs to both the “station” category and the “person name” category. In this case, since the user wants to execute the task of transferring information, it is not desirable that the task of searching for the phone book is executed again.

  In order to deal with the above case, the category specifying unit 13 specifies the category of the word input from the word input unit 12 based on a word other than the word included in the same utterance (word group) as the word. To do. That is, the category of a word is specified in consideration of the same utterance as the word, not just the word.

  Specifically, the category specifying unit 13 includes a degree to which a word to be specified included in information (utterance) input from the word input unit 12 and a word other than the word included in the information are included in the same information. The category of the word to be identified is identified based on the information indicating. More specifically, the category specifying unit 13 stores the degree that a word in the category and another word are included in the same information (utterance) for each category, and the word to be specified is determined based on the degree. A score for each category is specified for a word to be specified when another word is used as the other word, and a category of the word to be specified is specified based on the score. As the above-mentioned degree, a co-occurrence probability score can be used. The score of the co-occurrence probability indicates that the higher the value, the higher the degree that the word in the category and another word are included in the same information.

  For example, the category specifying unit 13 holds a table as shown in FIG. 7, and stores the score of the co-occurrence probability by storing information in the table. In this table, elements that co-occur with words of a specific category (included in the same utterance) are associated with the scores of co-occurrence probabilities of those words and elements. FIG. 7A shows a score of co-occurrence probabilities of elements that co-occur with words in the “station” category (included in the same utterance). FIG. 7B shows the score of the co-occurrence probability of an element that co-occurs with a word in the “person name” category (included in the same utterance). For example, for the word “go”, the co-occurrence probability score is 0.51 for words in the “station” category, and the co-occurrence probability score is 0.04 for words in the “person name” category. is there. As described above, for the word “go”, the word in the “station” category has a higher co-occurrence probability score. That is, it is shown that a word in which the word “go” is included in the same utterance is likely to be a word in the “station” category. The co-occurring element may be a word itself or a category (words in the entire category). A score may be associated with a combination of a plurality of co-occurring elements.

  The score is input and stored in advance in the function execution instruction server 10 by, for example, an administrator of the function execution instruction system 1 or the like. Further, it may be obtained by machine learning.

  As described above, the category specifying unit 13 specifies the categories of words by using the dictionary or chunking shown in FIG. 3 and, as a result, co-occurs with respect to the words specified as belonging to a plurality of categories (determined to belong). Limit categories based on probability scores. For example, in the case of the utterance “go to Shibuya” as described above, the word “Shibuya” is classified into two categories of “station” category and “person name” category in the category specification by dictionary or chunking. Identified. The category specifying unit 13 calculates a score for each category based on the co-occurrence probability score shown in FIG. 7 from words that co-occur with a category specifying target word that is a word specified for a plurality of categories. The category specifying unit 13 calculates the score for each category by adding the scores of each category of the words that co-occur with the category specifying target word for each category. The category specifying unit 13 specifies the category having the highest calculated score as the category of the word to be specified. When the co-occurring element is a verb, the use of the verb may be considered. For example, the word “go” may be handled in the same manner as “go”.

  The co-occurrence probability score shown in FIG. 7 may be used only for the category specified by the dictionary or chunking shown in FIG. 3 as described above. In the case of “Going to Shibuya” above, the word “Shibuya” is specified in the “Station” category and the “Person Name” category, so the “Station” category and the “Person Name” category Co-occurrence probability score is used. Further, since the words “until” and “go” co-occur in “Shibuya”, the score for each category is specified based on the table of FIG. For the “Station” category, a score of 0.51 for “going”, a score of 0.34 for “to”, and a score of 0.26 for a combination of “going” and “to” are totaled, and the total score is 1. .11. For the “person name” category, the “go” score of 0.04, the “to” score of 0.3, and the score “0” of the combination of “go” and “to” are summed, and the total score is 0.34. It is said. Since the total score of the “station” category is higher than the total score of the “person name” category, the category of “Shibuya” is specified (limited) as “station”.

  The category specifying unit 13 may specify a word category based only on the co-occurrence probability score without specifying the category by a dictionary or chunking. In that case, word scores based on the co-occurrence probability scores are calculated for all categories. The above is the functional configuration of the function execution instruction system 1 according to the present embodiment.

  FIG. 8 shows a hardware configuration of a server device constituting the function execution instruction server 10 and the voice recognition server 20 according to the present embodiment. As shown in FIG. 8, the server device includes a central processing unit (CPU) 101, a random access memory (RAM) 102 and a read only memory (ROM) 103, which are main storage devices, a communication module 104 for communication, and a hard disk. The computer is configured to include a computer including hardware such as the auxiliary storage device 105. The functions of the function execution instruction server 10 and the voice recognition server 20 described above are exhibited by the operation of these components by a program or the like. The above is the configuration of the function execution instruction system 1 according to the present embodiment.

  Subsequently, a function execution instruction method which is a process executed by the function execution instruction system 1 according to the present embodiment will be described with reference to the flowchart of FIG. In this process, first, a user operation on the communication terminal 30 for receiving a function execution instruction by the function execution instruction system 1 according to the present embodiment is performed, and the user's operation for causing the communication terminal 30 to execute the function is performed. Voice (utterance) is input. Subsequently, the voice is transmitted from the communication terminal 30 to the voice recognition server 20. The voice recognition server 20 receives and inputs the voice (S01, voice recognition step). Subsequently, the voice recognition server 20 performs voice recognition on the input voice (S02, voice recognition step). Information indicating a set of words or a sentence as a speech recognition result is transmitted from the speech recognition server 20 to the communication terminal 30. The communication terminal 30 receives the information and transmits it to the function execution instruction server 10.

  In the function execution instruction server 10, the word input unit 12 receives and inputs information including a word that is a voice recognition result (S03, word input step). The input information is output from the word input unit 12 to the category specifying unit 13. Subsequently, the category specifying unit 13 specifies the category of the word included in the information input from the word input unit 12 (S04, category specifying step).

  Here, with reference to the flowchart of FIG. 10, the word category specifying process (S04) will be described in detail. In this process, first, the category of the word is specified by the dictionary method shown in FIG. 3 or the chunking method as described above (S41). In this category specification, if there is no word specified in a plurality of categories, the word category specification processing (S04) ends. In this category specification, when there are words specified in a plurality of categories, one category is specified from the plurality of categories based on the co-occurrence probability score in FIG. (Limited) (S42). Information on the specified category is output from the category specifying unit 13 to the execution function determining unit 14 together with the word.

  Returning to FIG. 9, subsequently, the execution function determination unit 14 determines a candidate for a task to be executed (S 05, execution function determination step). The task candidate is determined by calculating a score for each task based on the input word and the category of the word as described with reference to FIGS. The task with the highest score is the task candidate to be executed.

  Subsequently, the execution function determining unit 14 corresponds to the category specified by the category specifying unit 13 and the slot of the task for which execution is instructed by the function execution instructing unit 11 in the past stored by the execution function storing unit 15. Based on the category, it is determined whether or not to instruct execution of a task that has been instructed to be executed by the function execution instructing unit 11 in the past (S06, execution function determining step). That is, it is determined whether to instruct execution of the task that has been instructed immediately before or to instruct execution of a new task (task transition). Note that if there is no task that has been instructed to be executed in the past, or there is a task that has been instructed to execute in the past, the task that has been instructed to execute in the past is not used for judgment because it is an execution instruction before a certain period. In some cases, the above determination need not be made. In this case, the task to be executed is determined based on the word input in S03.

  When it is determined that the execution of the task for which execution has been instructed immediately before is to be instructed, the task is determined as the task for which execution is instructed (S07, execution function determination step). On the other hand, if it is determined that the task is to be changed, the task determined as a task candidate in S05 is determined as a task that is instructed to be executed (S07, execution function determination step). Subsequently, the execution function determination unit 14 instructs the function execution instruction unit 11 to instruct execution of the determined task. At this time, the word input to the task slot is also output to the function execution instruction unit 11. Subsequently, the function execution instruction unit 11 that has received the instruction instructs the communication terminal 30 to execute the task (S08, function execution instruction step). The communication terminal 30 receives this instruction and executes a task related to the instruction.

  Subsequently, information indicating the task instructed to be executed by the function execution instruction unit 11 is stored in the execution function storage unit 15 (S09, execution function storage step). The information indicating the task stored here is referred to when the task to be instructed is determined (S06) when the task is instructed next by the user's voice. The task execution (start-up) processing by voice (the processing of S01 to S09) can be repeated. The above is the function execution instruction method which is the process executed by the function execution instruction system 1 according to the present embodiment.

  As described above, in the present embodiment, whether or not to instruct execution of a task that has been instructed in the past based on the category associated with (associated with) the task and the category of the input word is determined. To be judged. Thereby, for example, it is appropriately determined whether a task executed in the past should be executed subsequently or a new task should be executed.

  As in the example above, after the transfer to Tokyo Station is examined by the task of “Transfer Information”, if “Yokohama” is spoken after all, the task is taken over and transfer to Yokohama Station is performed. Be examined. On the other hand, when “curry” is spoken, the task of “gourmet search” is executed without taking over the task. Unlike this embodiment, if the task is assumed to be taken over uniformly, if “Curry” is spoken, “Transfer” and “Curry” do not correspond, so “Curry” is ignored and the user ’s intention In contrast, transfer to Tokyo Station will be examined again. Also, unlike this embodiment, if it is assumed that the task is not handed over uniformly, when “Yokohama” is spoken, the “map search” task is executed as described above, unlike the user's intention. End up. As described above, in the present embodiment, task takeover or task transition is appropriately performed according to the user's intention, and an appropriate task is executed.

  In the present embodiment, since the category of the input word is specified based on a word other than the word included in the information including the word, the category is appropriately specified. Specifically, as described above, the category is specified based on the co-occurrence probability score shown in FIG. After the user utters “Mr. Enomoto's phone number” and the phone book search task is executed as in the above example (for example, “Mr. Enomoto's phone number is xxx”) Suppose the user utters “I want to go to Shibuya” with the intention of executing the task of transfer guidance.

  If the category is not specified (limited) based on the co-occurrence probability score shown in FIG. 7, the telephone directory search task is taken over. However, in this embodiment, the category is identified based on the co-occurrence probability score, and the category of the word “Shibuya” is identified as the “station” category. As a result, the task of searching for the phone book is not carried over, and the task of transfer guidance is executed as a new task (for example, notification to the user such as “Here is the transfer guide to Shibuya Station” is performed). In this way, the operation as expected by the user is performed. In other words, according to the present embodiment, it is possible to appropriately perform the function using the words that are continuously input.

  Also, as in the present embodiment, a keyword that is strongly related to a specific task such as “transfer” may be set as a preset word, and a task to be executed may be determined based on the keyword. According to this configuration, for example, a task can be appropriately executed based on a word deeply related to a function.

  In addition, as in the above-described example, it may be determined whether to execute the task executed immediately before. According to this configuration, for example, it is appropriately determined whether the task executed immediately before should be executed subsequently or a new task should be executed, and the function can be appropriately executed by continuously input words. it can. However, it is not always necessary to set only the task executed immediately before as a takeover determination target. For example, a plurality of tasks executed within a predetermined time may be set as takeover targets.

  Moreover, it is good also as a structure which provides a slot in a task like this embodiment and inputs the word input by the word input part 12 into a slot. According to this configuration, together with the execution of the task, the word input by the word input unit 12 can be used as an input to the task, an appropriate task can be executed, and the convenience of the user is improved. be able to. For example, in combination with the execution of the task of “transfer guidance” described above, guidance based on the input station name can be performed.

  In addition, as in this embodiment, the determination of task takeover is made by comparing the category (slot category) related to the task (the category of the slot) with the category of the input word, and whether they match. It may be performed by judging. According to this configuration, it is possible to determine whether or not to instruct execution of a task for which execution has been instructed in the past more appropriately, and it is possible to more appropriately implement the present invention.

  Alternatively, the word input in the slot may be stored, and the stored word may be used as an input to the slot when the next task execution is instructed. According to this configuration, a word used in a task for which execution has been instructed in the past can be used as an input to a newly executed function, and user convenience can be further improved. For example, in the “transfer guidance” task described above, it is possible to provide guidance in which only the arrival station is changed.

  Further, as in the present embodiment, voice recognition may be performed by inputting voice, and the voice recognition result may be information including the input word. According to this configuration, the function can be executed by the user's voice.

  However, in the present embodiment, the function execution instruction system 1 includes the voice recognition server 20, but the voice recognition server 20 is not necessarily included. In this case, a result of speech recognition or a word group or a sentence not based on speech recognition is input to the function execution instruction system 1. Further, in the present embodiment, the function execution instruction server 10 and the voice recognition server 20 are configured separately, but they may be configured as a single unit. In this case, it is not necessary to transmit / receive the voice recognition result via the communication terminal 30. Further, a function used by the function execution instruction server 10 or the voice recognition server 20 may be provided in a terminal used by a user such as the communication terminal 30. In that case, the terminal is a function execution instruction system according to an embodiment of the present invention. Alternatively, the function execution instruction server 10 may have a function of executing a task, and the task execution result may be provided to the communication terminal 30.

  Further, in the above-described embodiment, information indicating a task for which execution is instructed by the function execution instructing unit 11 is input and stored in the execution function storage unit 15. However, the information stored in the execution function storage unit 15 may be transmitted from the communication terminal 30 instead of the function execution instruction unit 11 and received by the function execution instruction server 10. In this case, the function of the execution function storage unit 15 of the function execution instruction server 10 as described above is not necessarily required.

  In this case, the communication terminal 30 stores information indicating a task (task executed or executed in accordance with the instruction) instructed to be executed immediately from the function execution instruction server 10 (function execution instruction unit 11 thereof). The result is sent to the function execution instruction server 10 together with the result of the voice recognition. For example, the communication terminal 30 transmits the utterance information “Yokohama” or “curry” and the “transfer guidance” information indicating the task being executed to the function execution instruction server 10. In the function execution instruction server 10, the information is received, and the execution function determination unit 14 determines a function to be executed in the same manner as described above based on the received information. The execution function determination unit 14 of the function execution instruction server 10 communicates information stored in the execution function storage unit 15 other than the above information (for example, information input to the slot of the task executed immediately before). The function may be received from the terminal 30 and used to determine a function to be executed and to instruct execution.

  DESCRIPTION OF SYMBOLS 1 ... Function execution instruction system, 10 ... Function execution instruction server, 11 ... Function execution instruction part, 12 ... Word input part, 13 ... Category specification part, 14 ... Execution function determination part, 15 ... Execution function memory | storage part, 20 ... Voice Recognition server 101 ... CPU 102 ... RAM 103 ... ROM 104 ... Communication module 105 ... Auxiliary storage device 30 ... Communication terminal

Claims (8)

Function execution instruction means for instructing execution of one or more functions related to a category set in advance for each function;
A word input means for inputting information including a word;
Category specifying means for specifying a category of words included in the information input by the word input means based on words other than the words included in the information;
Execution function determining means for determining a function to be instructed to be executed by the function execution instructing means from the one or more functions according to the category specified by the category specifying means,
The execution function determining means is executed by the function execution instructing means in the past based on the category specified by the category specifying means and the category related to the function instructed to be executed by the function execution instructing means in the past. Is input by the word input unit when it is determined that the execution of the function instructed in the past by the function execution instructing unit is not instructed from now on. A function execution instruction system for newly determining a function to be instructed to execute based on a word included in the information .   The category specifying unit is based on information indicating a degree that a word to be specified included in the information input by the word input unit and a word other than the word included in the information are included in the same information. The function execution instruction system according to claim 1, wherein a category of a word to be specified is specified. An execution function storage means for storing the function instructed to be executed by the function execution instruction means;
The execution function determining means is based on the category specified by the category specifying means and the category related to the function instructed to be executed by the function execution instruction means in the past stored by the execution function storage means. The function execution instruction system according to claim 1, wherein it is determined whether or not to instruct execution of a function that has been instructed to be executed by the function execution instruction means in the past.   The execution function determining means executes the function instructed to be executed by the function execution instructing means immediately before a word related to the category is input by the word input means according to the category specified by the category specifying means. The function execution instruction system according to any one of claims 1 to 3, wherein it is determined whether or not to instruct the user. Functions that are instructed to be executed by the function execution instructing unit include a function that is included in the information input by the word input unit and that inputs a word of a category related to the function,
The execution function determining means determines whether or not the words of the category specified by the category specifying means can be input to the function for which execution has been instructed in the past by the function execution instructing means. 5. The function execution instruction system according to claim 1, wherein the function execution instruction system determines whether to instruct execution of a function that has been instructed by the function execution instruction unit in the past.   The execution function determining means determines whether the category specified by the category specifying means matches the category of the word that has been input to the function for which execution has been instructed in the past. The function execution instruction system according to claim 5, wherein if it is determined, it is determined that the execution of the function is to be instructed.   The speech recognition unit according to any one of claims 1 to 6, further comprising speech recognition means for inputting speech, performing speech recognition on the input speech, and inputting the result of speech recognition to the word input means. Function execution instruction system. A function execution instruction method that is an operation method of the function execution instruction system,
A function execution instruction step for instructing execution of one or more functions related to a preset category for each function;
A word input step for inputting information including the word;
A category specifying step for specifying a category of a word included in the information input in the word input step based on a word other than the word included in the information;
An execution function determining step for determining a function to be instructed in the function execution instruction step from the one or more functions according to the category specified in the category specifying step;
In the execution function determining step, based on the category specified in the category specifying step and the category related to the function that has been instructed in the function execution instructing step in the past, the function execution instructing step in the past Is input in the word input step when it is determined whether or not to instruct the execution of the function instructed in the function execution instruction step in the past. A function execution instructing method for newly determining a function to be instructed for execution based on a word included in the information .

Images