JPWO2018229937A1 - Intention estimation apparatus and intention estimation method - Google Patents

Abstract

  A morphological analysis unit (103) that analyzes morphemes included in the character string based on the acquired character string, and the number of intentions for the character string is estimated, and only one of the character strings is calculated according to the estimated number of intentions The intention number estimation unit (106) that determines whether it is a single intention string that does not contain an intention or a multiple intention string that includes multiple intentions, and the intention number estimation unit (106) If it is determined to be a character string, an intention for the single intention character string using a single intention estimation model in which the degree of association with the morpheme is associated with each intention based on the morpheme analyzed by the morpheme analysis unit (103) If the single intention estimation unit (108) for estimating the single intention and the intention number estimation unit (106) determine that the character string is a double intention character string, the morpheme analysis unit (103) analyzes the morpheme based on the analysis. , Morphemes with multiple intentions A compound intention estimating unit (110) that estimates a plurality of intentions for the compound intention string by using a compound intention estimation information model associated with a degree of association, and a plurality of intentions estimated by the compound intention estimating unit (110) And a combined estimation result unit (111).

Description

  The present invention relates to an intention estimation apparatus and an intention estimation method for recognizing an input character string and estimating a user's intention.

  Conventionally, there is known an intention estimation apparatus which performs speech recognition of a voice uttered by a user, converts the speech into a character string, and deduces from the character string an intention of the user as to what kind of operation to perform. Since there is also a case where a plurality of intentions are included in one utterance (hereinafter also referred to as a double intention utterance), the intention estimation device is required to be able to estimate an intention for a double intention utterance.

  For example, in the method using supervised learning disclosed in Non-Patent Document 1, a character string is expressed in a form called Bag of words, and the Bag of words is used as a feature amount to support vector machine or log linear model ( A classifier (intense understanding model) called maximum entropy model is trained, and intention is estimated based on probability values calculated using learning results. According to the method, for example, one string such as “search for ramen restaurant and Chinese food.” Includes the intention “search for ramen restaurant” and the intention “search for Chinese food”. Even in the case of having the following structure, the intention of the speaker or the like can be estimated.

Takamura Takaya, "Introduction to Machine Learning for Language Processing", 5th Edition, Corona Co., Ltd., August 5, 2010, p. 99-146

When the intention estimation method disclosed in Non-Patent Document 1 is applied to the case where one utterance may include a plurality of intentions, a separate model is learned for each intention and each model is executed at the time of execution. It will integrate the judgment results based on the model.
However, in the method of integrating the determination results based on a plurality of models at the time of execution with respect to one utterance as described above, even if the utterance includes only one intention (hereinafter, also referred to as a single intention utterance) In order to perform intention estimation based on each of the models, there is a problem that a plurality of intentions may be estimated and output, and the estimation accuracy of the intention may be lowered as a whole.

  The present invention has been made to solve the problems as described above, and it is possible to accurately estimate intention even when the acquired character string can be both a single intention character string and a double intention character string. It aims at providing an intention estimation device.

  The intention estimation apparatus according to the present invention estimates a morpheme analysis unit that analyzes morphemes included in the character string based on the acquired character string, estimates the number of intentions for the character string, and determines the character according to the estimated intention number. Intended number estimation unit that determines whether a string is a single intention string that contains only one intention or a multiple intention string that includes multiple intentions; If it is determined to be a character string, using the single intention estimation model in which the degree of association with the morpheme is associated with each intention based on the morpheme analyzed by the morpheme analysis unit, the intention for the single intention character string is single intention If the single intention estimation unit to estimate as and the intention number estimation unit determine that the character string is a double intention character string, the degree of association with the morpheme for each of the plurality of intentions is based on the morpheme analyzed by the morpheme analysis unit. Combined intention estimation model associated With, those having a composite intention estimation unit for estimating a plurality of intentions for the double intent string, the estimated result integration unit that integrates a plurality of intent composite intention estimation unit has estimated as a composite intended.

  According to the present invention, it is possible to improve the accuracy of estimating the user's intention.

FIG. 1 is a diagram showing an example of a configuration of an intention estimation device according to a first embodiment. FIG. 7 is a diagram showing an example of an intention number estimation model according to the first embodiment. 5 is a diagram showing an example of a single intention estimation model in Embodiment 1. FIG. 5 is a diagram showing an example of a composite intention estimation model in Embodiment 1. FIG. FIG. 5A and FIG. 5B are diagrams showing an example of the hardware configuration of the intention estimation apparatus according to the first embodiment. FIG. 2 is a diagram showing an example of a configuration of an intention number estimation model generation device according to Embodiment 1. FIG. 6 is a diagram showing an example of learning data stored in a learning data storage unit in the first embodiment. In Embodiment 1, it is a flowchart for demonstrating the process which the intention number estimation model production | generation apparatus produces | generates the intention number estimation model. In Embodiment 1, it is a figure which shows the example of interaction performed between a user and a navigation apparatus. 5 is a flowchart for explaining the operation of the intention estimation device according to the first embodiment. FIG. 11 is a flowchart for describing an operation of an intention number estimation unit in step ST1004 of FIG. 10 in the first embodiment. In Embodiment 1, it is a figure which shows an example of the score of the dependency information with respect to each intention number which the intention number estimation part acquires. In Embodiment 1, it is a figure which shows the calculation formula which an intention number estimation part uses in order to calculate a final score. In Embodiment 1, it is a figure which shows an example of the final score of each intention number which an intention number estimation part calculates. In Embodiment 1, it is a figure which shows an example of the final score of each intention number which an intention number estimation part calculates. In the first embodiment, the intention number estimation unit is an example of the determination result of the user's intention, which is the estimation result of the combined intention estimation unit. In this Embodiment 1, it is a figure which shows an example of the integration result of the intent integrated by the presumed result integration part. FIG. 7 is a diagram showing an example of configuration of an intention estimation apparatus according to a second embodiment. In Embodiment 2, it is a figure which shows the example of a dialog performed between a user and a navigation apparatus. FIG. 10 is a flowchart for explaining the operation of the intention estimation device in the second embodiment. In Embodiment 2, it is an example of the determination result of a user's intention which the compound intention estimation part determined. In this Embodiment 2, it is a figure which shows an example of the integration result of the intent integrated by the presumed result integration part. In Embodiment 2, it is a figure which shows an example of the content of the final intention estimation result produced | generated by the estimation result selection part.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
Embodiment 1

The intention estimation device 1 according to the first embodiment is, for example, mounted on a navigation device for providing route guidance and the like to a user such as a driver of a vehicle, and estimates the user's intention from the utterance content uttered by the user. Control is performed to cause the navigation device to execute an operation according to the estimated user's intention. The intention estimation device 1 may be connected to the navigation device via a network or the like.
The example installed in the navigation device is only an example, and the intention estimation device 1 according to the first embodiment is not limited to the user of the navigation device, but receives the information input from the user by speech etc. The present invention can be applied to an intention estimation apparatus that estimates the intention of the user of the apparatus in any apparatus that performs an operation according to the information.

FIG. 1 is a diagram showing an example of the configuration of an intention estimation device 1 according to a first embodiment.
As shown in FIG. 1, the intention estimation apparatus 1 includes a speech reception unit 101, a speech recognition unit 102, a morphological analysis unit 103, a dependency analysis unit 104, an intention number estimation model storage unit 105, and an intention number estimation. Unit 106, single intention estimation model storage unit 107, single intention estimation unit 108, combined intention estimation model storage unit 109, combined intention estimation unit 110, estimation result integration unit 111, command execution unit 112, response A generation unit 113 and a notification control unit 114 are provided.
In the first embodiment, as shown in FIG. 1, the intention number estimation model storage unit 105, the single intention estimation model storage unit 107, and the combined intention estimation model storage unit 109 are provided in the intention estimation device 1. The intention number estimation model storage unit 105, the single intention estimation model storage unit 107, and the combined intention estimation model storage unit 109 are not limited to this, and the intention estimation device 1 outside the intention estimation device 1 is It may be provided at a place where reference is possible.

  The voice receiving unit 101 receives a voice including the user's speech. The voice receiving unit 101 outputs the received voice information to the voice recognition unit 102.

  The voice recognition unit 102 performs voice recognition on voice data corresponding to the voice received by the voice receiving unit 101 and converts the voice data into a character string. The speech recognition unit 102 outputs the character string to the morphological analysis unit 103.

The morphological analysis unit 103 performs morphological analysis on the character string output from the speech recognition unit 102.
Here, the morphological analysis is an existing natural language processing technology that divides a character string into morphemes that are the smallest units having meaning as a language, and adds parts of speech using a dictionary. For example, when morphological analysis is performed on a character string “Go to Tokyo Tower”, the character string is divided into morphemes such as “Tokyo Tower / proper noun, to / case particle, go / verb”.
The morphological analysis unit 103 outputs the morphological analysis result to the dependency analysis unit 104 and the intention number estimation unit 106.

The dependency analysis unit 104 analyzes the relationship between the morphemes with respect to the character string after the morphological analysis by the morphological analysis unit 103, and generates dependency information. Here, the relation between morphemes is a relation of dependency of morphemes included in a character string. The dependency relationship means, for example, a relationship between morphemes such as “operation target”, “parallel relationship” and the like. As the dependency analysis method, the dependency analysis unit 104 may use an existing analysis method such as Shift-reduce or a spanning tree, for example.
The dependency analysis unit 104 outputs the analysis result of the relationship between morphemes to the intention number estimation unit 106 as dependency information.

  The intention number estimation model storage unit 105 stores an intention number estimation model. The intention number estimation model is a model for performing intention number estimation using the dependency information as a feature quantity.

FIG. 2 is a diagram showing an example of an intention number estimation model according to the first embodiment.
In the intention number estimation model illustrated in FIG. 2, the degree of association between the number of intentions and the dependency information is described as a score.
In the first embodiment, the dependency information is expressed in a form in which the relationship between morphemes and the number of occurrences thereof are connected by “_”.
For example, as shown in FIG. 2, when the set of morphemes in the “parallel relation” relationship appears once in one character string, the dependency information is “parallel relation_1”.
Among the pieces of dependency information shown in FIG. 2, “1 operation target” indicates that there is only one set of morphemes having a “operation target” relationship in one character string, and therefore the number of intentions is “ It is often 1 ". Therefore, as shown in FIG. 2, the score for the number of intentions “1” is higher than the score for the numbers of intentions “2” and “3” for “1 operation target”. On the other hand, the scores for the number of intentions “2 cases” and “3 cases” are high, as the number of intentions is likely to be 2 or more for both “1 parallel relation” and “2 operation targets”. It is higher than the score for the number of intentions “1 case”. As described above, in the intention number estimation model, a higher score is set as the degree of association is higher, according to the number of intentions and the degree of association of the dependency information.
In addition, in order to demonstrate easily, in FIG. 2, only three types of "one case", "two cases", and "three cases" are shown in FIG.
In the first embodiment, the intention number of users is estimated by a statistical method using the intention number estimation model as illustrated in FIG.

The intention number estimation unit 106 estimates the number of intentions included in the character string using the intention number estimation model stored in the intention number estimation model storage unit 105 based on the dependency information output from the dependency analysis unit 104. Do. The specific method of the intention number estimation by the intention number estimation unit 106 will be described later.
The intention number estimation unit 106 determines whether the character string based on the voice accepted by the voice acceptance unit 101 is a single intention utterance or a multiple intention utterance according to the estimated intention number, and In response, the morpheme analysis result of the character string output from the morpheme analysis unit 103 is output to the single intention estimation unit 108 or the combined intention estimation unit 110. Specifically, when the intention number estimation unit 106 determines that the character string based on the speech accepted by the speech acceptance unit 101 is a single intention string by a single intention utterance, the character output by the morphological analysis unit 103 The morphological analysis result of the column is output to the single intention estimation unit 108. Also, when it is determined that the character string based on the voice received by the voice receiving unit 101 is a double intention utterance, the morpheme analysis result of the character string output by the morphological analysis unit 103 is output to the combined intention estimation unit 110 .

  In the first embodiment, although the intention number is estimated by a statistical method using the intention number estimation model, the present invention is not limited to this. Instead of the statistical method, correspondence between dependency information and the number of intentions may be prepared in advance as a rule, and the number of intentions may be estimated. For example, “If there is only one“ parallel relation ”of the facility name and facility type in the character string, the number of intentions included in the character string is“ 2 ”. It is possible to estimate the number of intentions by a rule such as "."

  Further, as a method of intention estimation in this embodiment 1 to be described later, for example, the maximum entropy method can be used. The single intention estimation unit and the composite intention estimation unit use statistical methods to estimate the intention, and the intention corresponding to the input morpheme is likely to be plausible from a set of morphemes and intentions collected in large amounts in advance To estimate.

The single intention estimation model storage unit 107 stores an intention estimation model for performing intention estimation using a morpheme as a feature amount. The intention can be expressed as "<main intention>[<slotname> = <slot value>, ...]". Here, the main intention indicates the classification or function of the intention. In the example of the navigation device, the main intention is the upper layer generated in response to the input performed by the user operating the input device (not shown) for example, such as destination setting or listening to music. Corresponds to the command of.
The slot name and slot value indicate the information necessary to carry out the main intention. For example, the intention included in the string "search for nearby restaurants" is that the main intention is "periphery search", the slot name is "facility type", and the slot value is "restaurant". Therefore, the intention included in the character string “search nearby restaurants” can be expressed as “nearby search [facility type = restaurant]”.

FIG. 3 is a diagram showing an example of the single intention estimation model in the first embodiment.
As shown in FIG. 3, the single intention estimation model is “destination setting [facility = ○○]” (○ is a specific facility name and the same applies hereinafter) or “peripheral search [facility type = restaurant]” Represents the score of each morpheme for the intention such as. In the single intention estimation model of the first embodiment, the score of each morpheme for the intention is the degree of association between the intention and each morpheme, and the higher the degree of association between the intention and each morpheme, the higher the score of each morpheme is It is set. The single intention estimation model, as shown in FIG. 3, is a model created by learning the degree of association between an intention and a morpheme, in which the degree of association with the morpheme is associated with each intention.
For example, as shown in FIG. 3, the user is likely to set a destination for the morpheme "go" or "destination", so that the intention "destination setting [facility = ○○]" is used. The scores of the morpheme "go" or "destination" are higher than the scores of other morphemes. On the other hand, with regard to the morpheme "delicious" or "meal", since the user is likely to search for nearby restaurants, the morpheme "delicious" or "delicious search" in the "periphery search [facility type = restaurant]" The score of "meal" is higher than the scores of other morphemes.

  The single intention estimation unit 108 estimates the user's intention using the single intention estimation model stored in the single intention estimation model storage unit 107 based on the morpheme analysis result of the character string output by the morphological analysis unit 103. Specifically, using the single intention estimation model, the single intention estimation unit 108 uses the single intention estimation model to determine the intention that the score in which the morpheme analyzed by the morphological analysis unit 103 is associated with the intention is the largest. Estimate. The single intention estimation unit 108 outputs the estimation result to the command execution unit 112 as a single intention estimation result.

  The combined intention estimation model storage unit 109 stores, for each intention, a combined intention estimation model created by learning a separate model. The compound intention estimation model is a model created by learning by a statistical method, with learning data of the estimation target intention as a positive example and learning data of other intentions as negative examples for each intention, It is a model for judging about binary of whether each intention belongs to presumed object intention.

FIG. 4 is a diagram showing an example of a combined intention estimation model in the first embodiment.
The combined intention estimation model includes a plurality of judgment ready drawing estimation models generated for each intention.
In FIG. 4, in order to facilitate the description, the number of intentions is “destination setting [facility = ○○]” (see FIG. 4A), “peripheral search [facility type = restaurant]” (see FIG. 4B), And the example is shown as three of "via addition place [facility = ○○]" (refer to FIG. 4C). In the combined intention estimation model according to the first embodiment, the score of each morpheme for the intention is the degree of association between the intention and each morpheme, and the higher the degree of association between the intention and each morpheme, the higher the score for each morpheme is It is set. As shown in FIG. 4, the compound intention estimation model is a model in which a plurality of intentions are separately prepared by learning the degree of association between the intention and the morpheme, and the degree of association with the morpheme is associated with each intention.

The composite intention estimation unit 110 uses the composite intention estimation model stored in the composite intention estimation model storage unit 109 to determine the morpheme analysis result of the character string output by the morphological analysis unit 103 for each of the determination preparation diagram estimation models. Based on the determination, it is determined whether the character string based on the voice received by the voice receiving unit 101 is the corresponding intention. Specifically, the composite intention estimation unit 110 determines, for each determination preparation drawing estimation model, whether the score in which the morpheme analyzed by the morphological analysis unit 103 and the intention are associated with each other is equal to or more than a preset threshold. It determines and determines whether the character string is the corresponding intention.
The combined intention estimation unit 110 outputs the determination result for each of the determination preparation drawing estimation models included in the combined intention estimation model to the estimation result integration unit 111 as an estimation result.

The estimation result integration unit 111 integrates the estimation results for each of the preparation preparation drawing estimation models included in the composite intention estimation model, which are output from the composite intention estimation unit 110.
The estimation result integration unit 111 outputs the integration result of the estimated intentions to the command execution unit 112 as a composite intention estimation result.

The command execution unit 112 transmits a corresponding command to the command processing unit of the navigation device based on the single intention estimation result output from the single intention estimation unit 108 or the composite intention estimation result output from the estimation result integration unit 111. Make it run. For example, when the single intention estimation unit 108 estimates the intention of “periphery search [facility type = restaurant]” and outputs it as a single intention estimation result in response to the user's utterance “searching for a good store”, the command The execution unit 112 causes the command processing unit of the navigation device to execute a command to search for a nearby restaurant.
The command execution unit 112 outputs execution operation information indicating the content of the command executed by the command processing unit to the response generation unit 113.

The response generation unit 113 generates response data corresponding to the command executed by the command processing unit 112 based on the execution operation information output from the command execution unit 112. The response data may be generated in the form of text data or may be generated in the form of audio data.
When the response generation unit 113 generates the response data in the form of voice data, the response generation unit 113 outputs a synthetic sound such as, for example, “Searched nearby restaurants. Please select from a list”. It suffices to generate audio data for the
The response generation unit 113 outputs the generated response data to the notification control unit 114.

  The notification control unit 114 causes the response data output from the response generation unit 113 to be output, for example, from an output device such as a speaker included in the navigation device, and notifies the user. That is, the notification control unit 114 controls the output device to notify the user that the command processing unit has executed the command. In addition, about the aspect of a notification, the notification by a display, the notification by an audio | voice, the notification by a vibration etc. may be anything as long as a user can recognize notification.

Next, the hardware configuration of the intention estimation device 1 according to the first embodiment will be described.
FIGS. 5A and 5B are diagrams showing an example of a hardware configuration of the intention estimation device 1 according to the first embodiment of the present invention.
In the first embodiment of the present invention, speech recognition unit 102, morphological analysis unit 103, dependency analysis unit 104, intention number estimation unit 106, single intention estimation unit 108, compound intention estimation unit 110, estimation Each function of the result integration unit 111, the command execution unit 112, the response generation unit 113, and the notification control unit 114 is realized by the processing circuit 501. That is, the intention estimation apparatus 1 is a processing circuit for performing a process of estimating a user's intention or a process of executing and notifying a machine command according to the estimated intention based on the received information on the user's utterance. It has 501.
The processing circuit 501 may be dedicated hardware as shown in FIG. 5A or a CPU (Central Processing Unit) 506 that executes a program stored in the memory 505 as shown in FIG. 5B.

  When the processing circuit 501 is dedicated hardware, the processing circuit 501 may be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an application specific integrated circuit (ASIC), an FPGA (field-programmable) Gate Array) or a combination thereof is applicable.

  When the processing circuit 501 is the CPU 506, the speech recognition unit 102, the morphological analysis unit 103, the dependency analysis unit 104, the intention number estimation unit 106, the single intention estimation unit 108, the combined intention estimation unit 110, and the estimation result Each function of the integration unit 111, the command execution unit 112, the response generation unit 113, and the notification control unit 114 is realized by software, firmware, or a combination of software and firmware. That is, the speech recognition unit 102, the morphological analysis unit 103, the dependency analysis unit 104, the intention number estimation unit 106, the single intention estimation unit 108, the compound intention estimation unit 110, the estimation result integration unit 111, and the command The execution unit 112, the response generation unit 113, and the notification control unit 114 execute processes such as a hard disk drive (HDD) 502, a CPU 506 that executes a program stored in the memory 505, or a system LSI (Large-Scale Integration) It is realized by a circuit. The program stored in the HDD 502 or the memory 505 is a combination of the speech recognition unit 102, the morphological analysis unit 103, the dependency analysis unit 104, the intention number estimation unit 106, and the single intention estimation unit 108. It can also be said that the computer is made to execute the procedures and methods of the estimation unit 110, the estimation result integration unit 111, the command execution unit 112, the response generation unit 113, and the notification control unit 114. Here, the memory 505 is, for example, a non-volatile memory such as a random access memory (RAM), a read only memory (ROM), a flash memory, an erasable programmable read only memory (EPROM), or an electrically erasable programmable read only memory (EEPROM). Semiconductor or volatile semiconductor memory, magnetic disk, flexible disk, optical disk, compact disk, mini disk, DVD (Digital Versatile Disc), etc. correspond.

The speech recognition unit 102, the morphological analysis unit 103, the dependency analysis unit 104, the intention number estimation unit 106, the single intention estimation unit 108, the compound intention estimation unit 110, the estimation result integration unit 111, and the command The functions of the execution unit 112, the response generation unit 113, and the notification control unit 114 may be partially realized by dedicated hardware and partially realized by software or firmware. For example, the function of the speech recognition unit 102 is realized by the processing circuit 501 as dedicated hardware, and the morphological analysis unit 103, the dependency analysis unit 104, the intention number estimation unit 106, and the single intention estimation unit 108 The processing circuit of the combined intention estimation unit 110, the estimation result integration unit 111, the command execution unit 112, the response generation unit 113, and the notification control unit 114 reads and executes the program stored in the memory 505. It is possible to realize that function.
The intention number estimation model storage unit 105, the single intention estimation model storage unit 107, and the combined intention estimation model storage unit 109 use, for example, the HDD 502. Note that this is merely an example, and the intention number estimation model storage unit 105, the single intention estimation model storage unit 107, and the combined intention estimation model storage unit 109 are constituted by a DVD, a memory 505, or the like. It is also good.
Further, the intention estimation device 1 includes an input interface device 503 and an output interface device 504 that communicate with an external device such as a navigation device.
The voice receiving unit 101 is configured of an input interface device 503.

Next, the operation of the intention estimation device 1 according to the first embodiment will be described.
First, an operation related to generation processing of an intention number estimation model, which is a premise of an operation of estimating a user's intention in the intention estimation device 1, will be described.
Here, it is assumed that the intention number estimation model generation process is performed by the intention number estimation model generation device 2 different from the intention estimation device 1.

FIG. 6 is a diagram showing a configuration example of the intention number estimation model generation device 2 according to the first embodiment.
As shown in FIG. 6, the intention number estimation model generation device 2 includes a learning data storage unit 115, a morphological analysis unit 103, a dependency analysis unit 104, and an intention number estimation model generation unit 116.
The configurations and operations of the morphological analysis unit 103 and the dependency analysis unit 104 are the same as the configurations and operations of the morphological analysis unit 103 and the dependency analysis unit 104 described with reference to FIG. Duplicate descriptions will be omitted.

  The learning data storage unit 115 stores the correspondence between the character string and the number of intentions as learning data. Here, although it is assumed that the intention number estimation model generation device 2 includes the learning data storage unit 115, the present invention is not limited to this, and the learning data storage unit 115 is outside the intention number estimation model generation device 2. The intention number estimation model generation device 2 may be provided at a referenceable location.

Here, FIG. 7 is a diagram showing an example of learning data stored in the learning data storage unit 115 in the first embodiment.
As shown in FIG. 7, the learning data is data in which the number of intentions corresponding to an utterance sentence example (hereinafter referred to as an utterance sentence example) which is an example sentence of a character string voice output by speech or the like. For example, the intention number “1 case” is assigned to the utterance sentence example 701 “I want to go to ○”.
The learning data is created in advance by a model creator or the like. The creator of the model creates learning data in which the intention number is given in advance for each utterance sentence example for a plurality of utterance sentence examples, and stores it in the learning data storage unit 115.

  The intention number estimation model generation unit 116 calculates the number of intentions corresponding to the utterance sentence example based on the learning data stored in the learning data storage unit 115 and the analysis result of the relationship between the morphemes by the dependency analysis unit 104. Are learned by a statistical method, and an intention number estimation model (see FIG. 2) indicating the correspondence between dependency information and the number of intentions is generated. The intention number estimation model generation unit 116 stores the generated intention number estimation model in the intention number estimation model storage unit 105.

FIG. 8 is a flowchart for explaining processing in which the intention number estimation model generation device 2 generates an intention number estimation model in the first embodiment.
First, the morphological analysis unit 103 performs morphological analysis on each sentence example of learning data stored in the learning data storage unit 115 (step ST801). For example, in the case of the utterance sentence example 701 of FIG. 7, the morphological analysis unit 103 performs morphological analysis on "I want to go to ○ ○", and "名詞 / noun, to / case particle, go / verb, tai / auxiliary verb Result of morpheme analysis. The morphological analysis unit 103 outputs the morphological analysis result to the dependency analysis unit 104.

  The dependency analysis unit 104 performs dependency analysis using the morpheme analyzed by the morphological analysis unit 103 based on the morpheme analysis result output from the morphological analysis unit 103 (step ST802). For example, in the case of the utterance sentence example 701, the dependency analysis unit 104 performs dependency analysis on the morphemes “○”, “to”, “to go” and “to”. The dependency analysis unit 104 obtains the analysis result of the relationship between the morphemes “operation target” from the morpheme, adds the number of intentions to the analysis result, and uses “operation target _1” as the dependency information. It is output to the number estimation model generation unit 116.

The intention number estimation model generation unit 116 generates an intention number estimation model using learning data stored in the learning data storage unit 115 based on the dependency information output by the dependency analysis unit 104 (step ST 803). For example, in the case of the utterance sentence example 701 “I want to go to ○”, the dependency information is “1 operation target” and the number of intentions included in the learning data is “1 intention number” as shown in FIG. is there. Therefore, when the intention number estimation model generation unit 116 uses the utterance sentence example 701, the score of “1 intention number” is higher than the score of other intention numbers for the dependency information “1 operation target”. Learn to be high. The intention number estimation model generation unit 116 performs the same processing as the above-described steps ST801 to ST803 on all utterance sentence examples included in the learning data, and finally, the intention number estimation model as shown in FIG. Generate
Then, the intention number estimation model generation unit 116 stores the generated intention number estimation model in the intention number estimation model storage unit 105. The intention number estimation model storage unit 105 is provided, for example, at a place where the intention number estimation model generation device 2 can access via the network.

  Here, although the intention number estimation model generation unit 116 uses all dependency information output from the dependency analysis unit 104 as a feature amount for the intention number estimation, the intention number estimation model generation unit 116 The configuration is not limited to this. The intention number estimation model generation unit 116 determines the number of intentions by determining a clear rule such as “use only parallel relation” or “use only operation target”, or uses a statistical method to estimate the number of intentions It is also possible to use only dependency information that is highly effective.

  Furthermore, although here the intention number estimation model generation device 2 different from the intention estimation device 1 generates the intention number estimation model and stores it in the intention number estimation model storage unit 105, the present invention is not limited thereto. The intention estimation apparatus 1 may generate an intention number estimation model and store the intention number estimation model storage unit 105 as the intention number estimation model. In this case, the intention estimation apparatus 1 further includes a learning data storage unit 115 and an intention number estimation model generation unit 116 in addition to the configuration described with reference to FIG. 1. Note that the learning data storage unit 115 may be provided outside the intention estimation apparatus 1 at a place where the intention estimation apparatus 1 can refer.

  Subsequently, on the assumption that the intention number estimation model is generated as described above and stored in the intention number estimation model storage unit 105, the intention estimation device 1 according to the first embodiment uses the intention number estimation model. The operation related to the intention estimation process in FIG.

Here, FIG. 9 is a diagram showing an example of interaction performed between the user and the navigation device in the first embodiment.
FIG. 10 is a flowchart for explaining the operation of the intention estimation device 1 according to the first embodiment.

First, as shown in FIG. 9, the navigation device outputs, for example, a voice "Please talk when it sounds" from a speaker provided in the navigation device (S1). Specifically, the voice control unit (not shown) of the intention estimation device 1 causes the navigation device to output a voice saying "Please talk when it rings".
When the navigation device outputs a voice saying "Please talk when it sounds", the user utters "I want to go to ○" with respect to the voice (U1). In addition, in FIG. 9, the audio | voice which a navigation apparatus receives and outputs an instruction | indication from the intention estimation apparatus 1 represents as "S", and the speech from a user is represented as "U".

When the user utters "I want to go to ○ (U1)" (U1), the voice receiving unit 101 receives the voice according to the voice. The speech recognition unit 102 performs speech recognition processing on the speech received by the speech reception unit 101 (step ST1001), and converts the speech into a character string. The speech recognition unit 102 outputs the converted character string to the morphological analysis unit 103.
The morphological analysis unit 103 performs morphological analysis processing on the character string output from the speech recognition unit 102 (step ST1002). For example, the morpheme analysis unit 103 obtains morphemes of “○”, “to”, “go”, and “want”, and the information of the morpheme as the morpheme analysis result as the dependency analysis unit 104 and the intention number estimation unit Output to 106.

  The dependency analysis unit 104 performs dependency analysis processing on the morphological analysis result output from the morphological analysis unit 103 (step ST1003). For example, since the dependency analysis unit 104 is the target of the operation of “Go” as the morpheme “○”, the character string output from the speech recognition unit 102 has the relationship between the morphemes of “operation target”. Analyze that there is. In addition, since the “operation target” is one, the morphological analysis unit 103 analyzes “one operation target”. Then, the morphological analysis unit 103 outputs the analysis result of “1 operation target” as dependency information to the intention number estimation unit 106.

  The intention number estimation unit 106 determines the intention number estimation model stored in the intention number estimation model storage unit 105 with the dependency information “1 operation target” output from the dependency analysis unit 104 in step ST1003 as a feature amount. The number of intentions is estimated using (step ST1004). The estimation operation of the intention number by the intention number estimation unit 106 will be described in detail using FIG.

FIG. 11 is a flowchart for explaining the operation of the intention number estimation unit 106 in step ST1004 of FIG.
First, the intention number estimation unit 106 collates the dependency information output from the dependency analysis unit 104 with the intention number estimation model, and acquires a score of each dependency information for each intention number (step ST1101).

Here, FIG. 12 is a diagram illustrating an example of a score of dependency information with respect to each intention number acquired by the intention number estimation unit 106 according to the first embodiment.
As shown in FIG. 12, when the dependency information to be the feature amount is “operation target_1”, for example, the intention number estimation unit 106 determines that the feature amount “operation target_1” for the number of intentions “1”. As a score, get 0.2. The intention number estimation unit 106 similarly obtains the score of the feature amount “1 operation target” for other intention numbers.

Next, the intention number estimation unit 106 calculates, based on the score of each intention number acquired in step ST 1101, a final score of each intention number for the estimation object that is one character string for which the intention number is to be estimated Step ST1102). In the first embodiment, the final score obtained by the intention number estimation unit 106 is a product calculated by multiplying all the scores of the dependency information with respect to the intention number for each intention number. That is, the final score is a product calculated by multiplying the number of intentions by the score of each feature amount used for estimation of the number of intentions.
FIG. 13 is a diagram showing a calculation formula used by the intention number estimation unit 106 to calculate the final score in the first embodiment.
In FIG. 13, S is a final score of a number of intentions (hereinafter referred to as the number of intention intentions) for which the final score is to be calculated among a plurality of intention numbers for the estimation target. Moreover, in FIG. 13, Si is a score of the i-th feature amount with respect to the number of intended intentions.

FIG. 14 is a diagram illustrating an example of the final score of each intention number calculated by the intention number estimation unit 106 in the first embodiment.
The intention number estimation unit 106 calculates the final score shown in FIG. 14 using the calculation formula shown in FIG. In this example, since the dependency information to be the feature amount is one of “the operation target_1”, the score corresponding to the final score and the feature amount “the operation target_1” is the same.
As shown in FIG. 14, the score of the feature amount “1 motion target” is 0.2 and the final score S is 0.2 as well with respect to the number of intentions “1”. Similarly, the intention number estimation unit 106 calculates the final score for each of the other intention numbers.

It returns to the flowchart of FIG.
The intention number estimation unit 106 estimates the number of intentions based on the final score of each intention number calculated in step ST1102 (step ST1103). Specifically, the intention number estimation unit 106 estimates the number of intentions having the highest final score among the calculated intention numbers of the estimation target as the number of intentions of the estimation target.
Here, the intention number estimation unit 106 estimates the number of intentions “one case” as the number of intentions.

It returns to the flowchart of FIG.
As a result of estimating the number of intentions in step ST1004, it is determined whether the number of intentions is greater than 1 (step ST1005).
In step ST1005, when the estimated number of intentions is larger than 1 (in the case of "YES" of step ST1005), it progresses to step ST1010-step ST1014. The details of the process after step ST1010 when the estimated number of intentions becomes larger than 1 in step ST1005 will be described later with a specific example.

In step ST1005, when the estimated number of intentions is 1 or less ("NO" of step ST1005), it progresses to step ST1006.
For example, in the example of U1 in FIG. 9, as a result of the intention number estimation unit 106 estimating the number of intentions, the number of intentions is “1”, so the process proceeds to step ST1006.
In step ST1006, the intention number estimation unit 106 outputs, to the single intention estimation unit 108, a character string that is the morpheme analysis result that the morpheme analysis unit 103 performs the morpheme analysis in step ST1002. Then, using the single intention estimation model (see FIG. 3) stored in the single intention estimation model storage unit 107, the single intention estimation unit 108 uses a single intention utterance text as a morphological analysis result. , The user's intention is estimated (step ST1006). For example, when the character string is "I want to go to ○.", "Destination setting [facility = ○ 0"] is estimated as the user's intention. Specifically, the single intention estimation unit 108 uses the single intention estimation model to estimate the intention with which the score of the morphological analysis result of the character string by the morphological analysis unit 103 is the largest is the intention of the user.
The single intention estimation unit 108 outputs the intention estimation result to the command execution unit 112 as a single intention estimation result.

The command execution unit 112 causes the command processing unit of the navigation device to execute a command corresponding to the single intention estimation result output from the single intention estimation unit 108 in step ST1006 (step ST1007). For example, the command execution unit 112 causes the command processing unit of the navigation device to execute an operation of setting the facility ○ as a destination.
The command execution unit 112 also outputs execution operation information indicating the content of the command executed in step ST1007 to the response generation unit 113.

  The response generation unit 113 generates response data corresponding to the command executed by the command processing unit 112 based on the execution operation information output from the command execution unit 112 in step ST1007 (step ST1008). The response generation unit 113 outputs the generated response data to the notification control unit 114.

  The notification control unit 114 causes, for example, a speaker provided in the navigation device to output a voice based on the response data output from the response generation unit 113 in step ST1008 (step ST1009). As a result, as shown in “S2” of FIG. 9, a voice such as “O is set as the destination” is output, and the user can be notified of the executed command.

Next, as shown by “U2” in FIG. 9, assuming that the user utters “Drop down Δ, select expressway”, the operation of the intention estimation device 1 in this case is shown in FIG. Explain along.
When the user utters as indicated by "U2", the voice receiving unit 101 receives the voice by the voice, and the voice recognition unit 102 performs voice recognition processing on the voice by the received voice (step ST1001), and the character string Convert to The speech recognition unit 102 outputs the converted character string to the morphological analysis unit 103 and the intention number estimation unit 106.
The morphological analysis unit 103 performs morphological analysis processing on the character string output from the speech recognition unit 102 (step ST1002). For example, the morphological analysis unit 103 obtains morphemes of “ΔΔ”, “Momo”, “Give”, “Te”, “Highway”, “O”, “Select” and “Te”, and The information is output to the dependency analysis unit 104 as a morphological analysis result.

  Next, the dependency analysis unit 104 performs dependency analysis processing on the morphological analysis result output from the morphological analysis unit 103 (step ST1003). Here, “△” is the target of the “stop” operation, “highway” is the target of the “select” operation, and the “stop” and the “select” are in a parallel relationship. The dependency analysis unit 104 outputs the analysis results of “operation target _2” and “parallel relationship _1” as dependency information to the intention number estimation unit 106.

The intention number estimation unit 106 uses the intention number estimation model stored in the intention number estimation model storage unit 105 as the feature amount of the acquired dependency information “2 operation objects” and “parallel relation_1” as the feature amount. The number is estimated (step ST1004).
Although the specific operation of step ST1004 is as described in detail with reference to FIG. 11 as described above, first, the intention number estimation unit 106 determines the number of intentions as in the case of “U1”. The dependency information output from the reception analysis unit 104 is compared with the intention number estimation model, and the score of each dependency information for each intention number is acquired (see step ST1101 in FIG. 11).
Subsequently, the intention number estimation unit 106 calculates a final score with respect to the number of intentions to be estimated from the calculation formula shown in FIG. 13 (see step ST1102 in FIG. 11).

FIG. 15 is a diagram showing an example of the final score of each intention number calculated by the intention number estimation unit 106 in the first embodiment.
The intention number estimation unit 106 calculates the final score shown in FIG. 15 for the speech "U2" by the user, using the calculation formula shown in FIG. Here, with respect to the number of intentions “1 case”, the score of the feature amount “2 motion targets” is 0.01, and the score of “parallel relationship_1 case” is 0.01. As a result, the intention number estimation unit 106 calculates the final score S of the number of intentions “one case” for the utterance “U2” as 1e−4 (= 0. 0001). Similarly, the intention number estimation unit 106 calculates the final score for each of the other intention numbers for the utterance “U2”.

  The intention number estimation unit 106 estimates the number of intentions based on the final score of each calculated intention number (see step ST1103 in FIG. 11). Specifically, the intention number estimation unit 106 estimates the number of intentions “2” having the highest final score among the calculated intention numbers of the estimation target as the number of intentions of the estimation target.

It returns to the flowchart of FIG.
As a result of estimating the number of intentions in step ST1004, it is determined whether the number of intentions is greater than 1 (step ST1005).
In step ST1005, when the estimated number of intentions is larger than 1 (in the case of "YES" of step ST1005), it progresses to step ST1010.
Here, since the estimated number of intentions is “2 cases” larger than 1 (in the case of “YES” in step ST1005), the process proceeds to step ST1010.

  In step ST1010, the intention number estimation unit 106 outputs, to the combined intention estimation unit 110, a character string that is the morpheme analysis result that the morpheme analysis unit 103 performs the morpheme analysis in step ST1002. Then, compound intention estimation unit 110 uses the compound intention estimation model (see FIG. 4) stored in compound intention estimation model storage unit 109 to generate a character string as a morpheme result, that is, a compound intention utterance text, The intention of the user is estimated (step ST1010).

Here, FIG. 16 is an example of the determination result of the user's intention, which is the estimation result of the compound intention estimation unit 110 in the first embodiment.
In FIG. 16, as the combined intention estimation model stored in the combined intention estimation model storage unit 109, the determination preparation figure estimation model of the intention “add via route [facility = ΔΔ]”, the intention “in order to facilitate the description. It is assumed that there are three models of the judgment preparation map estimation model of route change [highway priority] and the judgment preparation map estimation model of intention “destination setting [facility = ΔΔ]”. That is, the composite intention estimation unit 110 determines whether the character string that is the result of the morphological analysis by the morphological analysis unit 103 corresponds to the three intentions. The intention number estimation unit 106 determines that the intention estimation score exceeds 0.5 when the intention estimation score for the intention to be determined using the three determination ready drawing estimation models exceeds 0.5. Shall be determined to be the corresponding intention.
The intention estimation score is a probability value calculated based on the sum of the scores of the morphemes. Therefore, when the intention estimation scores are summed up in each determination preparation drawing estimation model, it becomes “1”.

In FIG. 16, FIG. 16A shows the determination result of the preparation preparation chart estimation model with the intention “add via point [facility = ΔΔ]”. The combined intention estimation unit 110 obtains 0.75 as the intention estimation score of the intention “add via route [facility = ΔΔ]”. In this case, since the intention estimation score exceeds 0.5, the combined intention estimation unit 110 determines that the intention “via addition [facility = ΔΔ]” is the corresponding intention of the character string “U2”.
In FIG. 16, FIG. 16B shows the determination result of the preparation preparation chart estimation model of the intention "route change [highway priority]". Since the intention estimation score of the compound intention estimation unit 110 is 0.7 and exceeds 0.5 (see FIG. 16B), the intention “route change [highway priority]” is also the corresponding intention of the character string “U2”. Determine that there is.
In FIG. 16, FIG. 16C shows the determination result of the preparation preparation chart estimation model with the intention “destination setting [facility = ΔΔ]”. Since the intention estimation score of the intention “destination setting [facility = ΔΔ]” is 0.5 or less, the combined intention estimation unit 110 does not “intention destination setting [facility = ΔΔ]” but “others”. It is determined that "intention" is the corresponding intention of the character string "U2".

  The combined intention estimation unit 110 has the corresponding intentions obtained by the three intention estimation models shown in FIGS. 16A to 16C, “add via route [facility = ΔΔ]”, “route change [highway priority]], and , “Other intention” is output to the estimation result integration unit 111 as an intention estimation result.

  The estimation result integration unit 111 adds the corresponding intention other than the “other intention” to the integration result among the plurality of corresponding intentions output as the intention estimation result from the compound intention estimation unit 110 in step ST 1010. Are integrated (step ST1011).

As shown in FIG. 16A, the determination result of the preparation preparation diagram estimation model of the intention “append route addition [facility = ΔΔ]” is the intention “append route addition [facility = ΔΔ]”, so the estimation result integration unit 111 adds the intention "add via point [facility = ΔΔ]" to the integration result. The estimation result integration unit 111 adds the intention “route change [highway priority]” to the integration result.
On the other hand, as shown in FIG. 16C, since the determination result of the determination preparation diagram estimation model of the intention “destination setting [facility = ΔΔ]” is “other intention”, the estimation result integration unit 111 Neither destination setting [facility = Δ △] ”nor“ other intention ”is added to the integration result.

FIG. 17 is a diagram showing an example of the integration result of the intention integrated by the estimation result integration unit 111 in the first embodiment.
The estimation result integration unit 111 outputs the integration result of the estimated intentions to the command execution unit 112 as a composite intention estimation result.

The command execution unit 112 causes the command processing unit of the navigation device to execute a command corresponding to the composite intention estimation result output from the composite intention estimation unit 110 in step ST1011 (step ST1012). For example, the command execution unit 112 causes the command processing unit of the navigation device to execute an operation of adding the facility ΔΔ to the via point. The command execution unit 112 also causes the command processing unit of the navigation device to execute an operation of changing the route to expressway priority.
Also, the command execution unit 112 outputs execution operation information indicating the content of the command executed in step ST1012 to the response generation unit 113.

  The response generation unit 113 generates response data corresponding to the command executed by the command processing unit 112 based on the execution operation information output from the command execution unit 112 in step ST1012 (step ST1013). The response generation unit 113 outputs the generated response data to the notification control unit 114.

  The notification control unit 114 causes, for example, a speaker provided in the navigation device to output a voice based on the response data output from the response generation unit 113 in step ST1013 (step ST1014). As a result, as shown in “S3” in FIG. 9, voices such as “I added Δ to the via point” and “I made the route a priority on the expressway” are output, and the user receives , Can be notified of the executed command.

  As described above, according to the first embodiment, the intention estimation device 1 estimates the number of intentions for the character string, and the morpheme analysis unit 103 that analyzes morphemes included in the character string based on the acquired character string. Depending on the estimated number of intentions, the character string is a single intention character string (single intention utterance) including only one intention or a multiple intention character string including multiple intentions (multiple intention utterance) If the intention number estimation unit 106 determines whether the character string is a single intention character string, the intention number estimation unit 106 determines whether the character string is a single intention character string, the relation with the morpheme for each intention based on the morpheme analyzed by the morpheme analysis unit 103. A single intention estimation unit 108 that estimates the intention for the single intention character string as a single intention using the single intention estimation model with which the degree is associated, the intention number estimation unit 106, and the character string is a multiple intention character string If it is determined that the morphological analysis unit And a composite intention estimation unit 110 configured to estimate a plurality of intentions for the multiple intention character string using a composite intention estimation information model in which the association degree with the morpheme is associated with each of the plurality of intentions based on the morpheme analyzed by 03. And the estimation result integration unit 111 that integrates the plurality of intentions estimated by the composite intention estimation unit 110 as a composite intention. In this way, it is possible to estimate the intention with high accuracy even when the acquired character string may be either a single intention character string or a multiple intention character string.

Second Embodiment
In the first embodiment, when it is estimated from the user's speech that the user's intention is two or more, the estimation result integration unit 111 integrates the compound intention estimation result estimated by the compound intention estimation unit 110, and the command execution unit 112 Is configured to cause the navigation device to execute a command corresponding to the integrated composite intention estimation result.
In the second embodiment, an embodiment will be described in which an upper limit is set to the number of intentions of the composite intention estimation result estimated by the composite intention estimation unit 110.
The second embodiment of the present invention will be described below with reference to the drawings.

FIG. 18 is a diagram showing a configuration example of the intention estimation device 1B according to the second embodiment.
The intention estimation device 1B of the second embodiment differs from the intention estimation device 1 described with reference to FIG. 1 in the first embodiment in that the estimation result selection unit 117 is provided. The other configuration of intention estimation apparatus 1B is the same as the configuration of intention estimation apparatus 1 described with reference to FIG. 1 in the first embodiment, so the same configuration as that of intention estimation apparatus 1 is the same as FIG. The sign of is attached and the duplicate explanation is omitted.
In the second embodiment, estimation result integration unit 111 outputs the combined intention estimation result, which is the integration result of the estimated intentions, to estimation result selection unit 117. At this time, the estimation result integration unit 111 also outputs the intention estimation score to the estimation result selection unit 117 by including it in the combined intention estimation result.
Further, in the second embodiment, the intention number estimation unit 106 outputs information on the estimated intention number to the estimation result selection unit 117.

  The estimation result selection unit 117 determines the intention as the estimation result with the intention number output from the intention number estimation unit 106 as the intention output upper limit for the compound intention estimation result output from the estimation result integration unit 111 Select from the top of the intention estimation score of the result. A specific method for selecting an estimation intention will be described later.

The operation of the intention estimation device 1B in the second embodiment will be described.
Here, FIG. 19 is a diagram showing an example of interaction performed between the user and the navigation device in the second embodiment.
FIG. 20 is a flow chart for explaining the operation of the intention estimation device 1B in the second embodiment.

First, as shown in FIG. 19, the navigation device outputs, for example, a voice "Please talk when it sounds" from a speaker provided in the navigation device (S01). Specifically, the voice control unit (not shown) of the intention estimation device 1B causes the navigation device to output a voice saying "Please talk when it pops."
When the navigation device outputs a voice saying "Please talk when it puzzes", the user utters the voice "Do you want to go to ○○, there is a convenience store nearby?" (U01). Here, as shown in FIG. 19, the voice that the navigation device receives an instruction from the intention estimation device 1B and outputs is represented as "S", and the speech from the user is represented as "U".

  Hereinafter, although it demonstrates along the flowchart of FIG. 20, the specific operation | movement of step ST2001-step ST2011 of FIG. 20 and step ST2013-step ST2015 is respectively step ST1001-step of FIG. 10 demonstrated in Embodiment 1. It is the same as the specific operation of ST1014.

First, the voice receiving unit 101 receives a voice according to the user's speech, and the voice recognition unit 102 performs voice recognition processing on the received voice to convert it into a character string, and the morpheme analysis unit 103 analyzes the morpheme for the character string A process is performed (steps ST2001 and ST2002). For example, the morphological analysis unit 103, the morphological analysis unit 103, "○ 、", "ha", "don't care", "don't", "te", "good", "near", "ni", "convenience store" And obtains the morpheme “a”, and outputs the information of the morpheme to the dependency analysis unit 104 and the intention number estimation unit 106 as the morpheme analysis result.
Next, the dependency analysis unit 104 performs dependency analysis processing on the character string (step ST2003). For example, "○○" is the target of the motion of "don't care", the target of the motion of "combine" is "present", and the motions "good" and "present" are "parallel". The dependency analysis unit 104 outputs the analysis results of “operation target _2” and “parallel relationship _1” as dependency information to the intention number estimation unit 106.
Then, using the dependency information output from the dependency analysis unit 104, the intention number estimation unit 106 estimates the number of intentions (step ST2004). Here, since the number of intentions estimated by the intention number estimation unit 106 is “2” (see step ST1104 of FIG. 11 described in the first embodiment) and the estimated number of intentions is larger than “1” (step ST2005) (In the case of “YES” in step), and proceeds to processing after step ST2010. Up to this point is the same as steps ST1001 to 1005 in FIG. 10 described in the first embodiment.

  In step ST2010, the intention number estimation unit 106 outputs, to the combined intention estimation unit 110, a character string that is the result of the morphological analysis of the morphological analysis unit 103. Then, compound intention estimating section 110 estimates the user's intention for the multiple intention utterance sentence.

Here, FIG. 21 is an example of the determination result of the user's intention determined by the compound intention estimation unit 110 in the second embodiment.
In FIG. 21, in order to facilitate the explanation, as the composite intention estimation model stored in the composite intention estimation model storage unit 109, the determination preparation diagram estimation model of the intention “delete via route [facility = ○○]”, the intention “ Description will be made on the assumption that there are three models of the judgment preparation diagram estimation model of peripheral search [facility type = convenience store] and the intention preparation diagram estimation model of intention “route deletion”. As in the first embodiment, the intention number estimation unit 106 determines that the intention estimation score is 0 when the intention estimation score for the intention determined using the three determination preparation drawing estimation models exceeds 0.5. .5 The intention determined to exceed 5 shall be determined to be the corresponding intention.

In FIG. 21, FIG. 21A shows the determination result of the preparation preparation chart estimation model with the intention “deletion of via point [facility = ○○]”. The combined intention estimation unit 110 obtains an intention estimation score of 0.65 for the intention “delete via route [facility = ○○]”. In this case, since the intention estimation score exceeds 0.5, the combined intention estimation unit 110 determines that the intention “delete via route [facility = ○○]” is the corresponding intention of the character string “U01”.
In FIG. 21, FIG. 21B shows the determination result of the intention “periphery search [facility type = convenience store]” determination preparation diagram estimation model, and FIG. 21C shows the determination result of the intention “route deletion” determination preparation diagram estimation model. Since the intention estimation score of the combined intention estimation unit 110 is 0.7 and exceeds 0.5 (see FIG. 21B), the intention “periphery search [facility type = convenience store]” is also the corresponding intention of the character string “U01” It is determined that Further, since the intention estimation score is 0.55 and exceeds 0.5 (see FIG. 21C), the composite intention estimation unit 110 determines that “route deletion” is also the corresponding intention of the character string “U01”. .
The combined intention estimation unit 110 is the corresponding intentions obtained by the three intention estimation models shown in FIGS. 21A to 21C, “Delete via site [facility = ○○]”, “periphery search [facility type = convenience store]”, And, “route deletion” is output to the estimation result integration unit 111.

  The estimation result integration unit 111 adds the corresponding intention other than the “other intention” to the integration result among the plurality of corresponding intentions output as the intention estimation result from the compound intention estimation unit 110 in step ST2010. Are integrated (step ST2011).

  As shown in FIG. 21A, the determination result of the prepared preparation diagram estimation model of the intention “delete via route [facility = ○○]” is the intention “intermediate place delete [facilities = ○○]”, so the estimation result integration unit 111 adds the intention "Deletion place deletion [facility = ○○]" to the integration result. Also, as shown in FIG. 21B and FIG. 21C, the determination result of the judgment preparation prepared map estimation model of the intention "nearby search [facility type = convenience store]" is "nearby search [facility type = convenience store]" Since the judgment result of the judgment preparation diagram estimation model is “route deletion”, the estimation result integration unit 111 similarly adds “periphery search [facility type = convenience store]” and “route deletion” to the integration result. At this time, in the second embodiment, the estimation result integration unit 111 also adds the intention estimation score to the integration result.

FIG. 22 is a diagram showing an example of the integration result of the intention integrated by the estimation result integration unit 111 in the second embodiment.
The estimation result integration unit 111 outputs the integration result of the estimated intentions to the estimation result selection unit 117 as a combined intention estimation result.

The estimation result selection unit 117 sets the number of intentions output from the intention number estimation unit 106 in step ST2004 as the intention output upper limit, with respect to the composite intention estimation result output from the estimation result integration unit 111 in step ST2011. The intention is selected from the top of the intention estimation score of the composite intention estimation result, and the selected estimation intention is set as the final intention estimation result (step ST2012).
Specifically, the estimation result selection unit 117 sets the number of intentions output from the intention number estimation unit 106 as the intention output upper limit, and selects only the high-order estimation intention of the intention estimation score using the intention estimation score as a determination criterion. .

Here, in step ST2004, the intention number estimation unit 106 estimates that the number of intentions is "2". Therefore, the estimation result selection unit 117 sets the number of final intention estimation results to “2” or less. The estimation integration result by the estimation result integration unit 111 is three of “route deletion [facility = ○○]”, “periphery search [facility type = convenience store]” and “route deletion”.
In addition, as shown in FIG. 22, the intention estimation score is “0.65” for “via route deletion [facility = ○○]” and “0.7” for “peripheral search [facility type = convenience store]”. Route deletion is "0.55".
The estimation result selection unit 117 sets the intention number output from the intention number estimation unit 106 as the intention output upper limit, selects the top two intention estimation scores of the composite intention estimation result, and outputs the selected result as the final intention estimation result. The estimation result selection unit 117 selects “via route deletion [facility = ○○]” and “peripheral search [facility type = convenience store]” as final intention estimation results.

Thus, in the intention estimation apparatus 1B, the output of unnecessary intention estimation results is suppressed by deleting "root deletion" from the compound intention estimation results by the estimation result selection unit 117, and an upper limit is set for the compound intention estimation results. The accuracy of intention estimation can be further improved as compared to the case where there is no such condition. As a result, a more appropriate final intention estimation result can be obtained.
FIG. 23 is a diagram showing an example of the content of the final intention estimation result generated by the estimation result selection unit 117 in the second embodiment.
The estimation result selection unit 117 outputs the final intention estimation result to the command execution unit 112.

The command execution unit 112 causes the command processing unit of the navigation device to execute a command corresponding to the final intention estimation result output from the estimation result selection unit 117 in step ST2012 (step ST2013). For example, the command execution unit 112 causes the command processing unit of the navigation device to execute a command for deleting a via point and a command for searching for a convenience store in the vicinity.
Further, the response generation unit 113 generates response data corresponding to the command executed by the command processing unit 112 by the command execution unit 112 (step ST2014), and the notification control unit 114 generates the response data generated by the response generation unit 113. , And output from a speaker provided in the navigation device (step ST2015). As a result, as shown in “S02” in FIG. 19, voices such as “Deleted via xxx” and “Search nearby convenience stores. Please select from a list.” Are output, and the user Can be notified of executed commands. The specific operation is the same as step ST1012 to step ST1014 of FIG. 10 described in the first embodiment.

  As described above, according to the second embodiment, in addition to the configuration of the intention estimation apparatus 1 according to the first embodiment, the estimation result integration unit 111 integrates the number of intentions estimated by the intention number estimation unit 106 as the upper limit. Among the plurality of intentions, the intention number estimation unit 106 selects the intention higher in the intention estimation score calculated when the intention number is estimated, and is configured to include the estimation result selection unit 117 as a combined intention. Thereby, using the intention number result obtained by the intention number estimation unit 106, an output upper limit is set for the composite intention estimation result obtained by the estimation result integration unit 111, and the output of the inappropriate intention estimation result is suppressed. Accuracy of the final integration result is further improved.

  Note that part of the functions of the intention estimation devices 1 and 1B described above may be performed by another device. For example, some of the functions may be executed by an externally provided server or a portable terminal such as a smartphone or a tablet.

  In the first and second embodiments described above, the intention estimation devices 1 and 1B estimate the user's intention based on the speech of the user's speech. Information is not limited to this. For example, the intention estimation devices 1 and 1B can also receive a character string input by the user using an input device such as a keyboard, and estimate the user's intention based on the character string.

  In the scope of the invention, the present invention allows free combination of each embodiment, or modification of any component of each embodiment, or omission of any component in each embodiment. .

  Since the intention estimation apparatus according to the present invention is configured to improve the accuracy of estimating the intention of a character string, it is applied to an intention estimation apparatus or the like that recognizes an input character string and estimates the user's intention. be able to.

  1, 1 B intention estimation device, 2 intention number estimation model generation device, 101 speech reception unit, 102 speech recognition unit, 103 morpheme analysis unit, 104 dependency analysis unit, 105 intention number estimation model storage unit, 106 intention number estimation unit, 107 single intention estimation model storage unit 108 single intention estimation unit 109 composite intention estimation model storage unit 110 composite intention estimation unit 111 estimation result integration unit 112 command execution unit 113 response generation unit 114 notification control unit 115 Data storage unit for learning, 116 intention number estimation model generation unit, 117 estimation result selection unit, 501 processing circuit, 502 HDD, 503 input interface device, 504 output interface device, 505 memory, 506 CPU.

Claims (8)

A morphological analysis unit that analyzes morphemes included in the character string based on the acquired character string;
If the number of intentions for the character string is estimated, and the character string is a single intention string including only one intention or is a multiple intention string including multiple intentions, according to the estimated number of intentions An intention number estimation unit that determines
When the intention number estimation unit determines that the character string is a single intention character string, a single intention estimation model in which the association degree with the morpheme is associated with each intention based on the morpheme analyzed by the morpheme analysis unit. A single intention estimation unit that estimates the intention for the single intention string as a single intention using
When the intention number estimation unit determines that the character string is a double intention character string, a compound intention in which the association degree with the morpheme is associated with each of a plurality of intentions based on the morpheme analyzed by the morpheme analysis unit. A combined intention estimating unit that estimates a plurality of intentions for the multiple intention character string using an estimation model;
An intention estimation apparatus comprising: an estimation result integration unit which integrates a plurality of intentions estimated by the composite intention estimation unit as a composite intention. A dependency analysis unit configured to analyze a relationship between morphemes included in the character string based on the morphemes analyzed by the morpheme analysis unit, and generate dependency information;
The intention estimation device according to claim 1, wherein the intention number estimation unit estimates the number of intentions for the character string based on the dependency information generated by the dependency analysis unit. The intention number estimation unit
The intention number for the character string is estimated using the intention number estimation model in which the dependency information is a feature amount and the correspondence relationship between the dependency information and the intention number is learned. Intention estimation device. Among a plurality of intentions integrated by the estimation result integration unit with the intention number estimated by the intention number estimation unit as an upper limit, the intention higher in intention estimation score calculated by the intention number estimation unit when the intention number is estimated The intention estimation apparatus according to claim 1, further comprising: an estimation result selection unit configured to select the combined intention as the combined intention. The morphological analysis unit analyzes morphemes included in the character string based on the acquired character string;
The intention number estimation unit estimates the number of intentions for the character string, and depending on the estimated intention number, the character string may be a single intention character string including only one intention, or a plurality of intentions. Determining whether it is an intended character string;
When the single intention estimation unit determines that the intention number estimation unit determines that the character string is a single intention character string, the degree of association with the morpheme is associated with each intention based on the morpheme analyzed by the morpheme analysis unit. Estimating the intention for the single intention string as a single intention using the single intention estimation model
When the combined intention estimation unit determines that the intention number estimation unit determines that the character string is a multiple intention character string, the degree of association with the morpheme for each of a plurality of intentions is based on the morpheme analyzed by the morpheme analysis unit. Estimating a plurality of intentions for the compound intention string using the associated compound intention estimation model;
The estimation result integration unit integrates the plurality of intentions estimated by the combined intention estimation unit as a combined intention. The dependency analysis unit analyzes the relationship between morphemes included in the character string based on the morphemes analyzed by the morpheme analysis unit, and generates dependency information.
6. The intention estimation method according to claim 5, wherein the intention number estimation unit estimates the number of intentions for the character string based on the dependency information generated by the dependency analysis unit. The intention number estimation unit
The method further includes the step of estimating the number of intentions for the character string using the intention number estimation model in which the dependency information is a feature amount and the correspondence relationship between the dependency information and the number of intentions has been learned. The intention estimation method described in 6. Among the plurality of intentions integrated by the estimation result integration unit, the estimation unit selected by the estimation result selection unit estimates the intention number when the intention number estimation unit estimates the number of intentions, with the intention number estimated by the intention number estimation unit as an upper limit. The intention estimation method according to claim 5, further comprising the step of selecting the upper intention of the estimated score and setting the intention as the combined intention.

Images