JP7113047B2 - AI-based automatic response method and system - Google Patents

Description

The following description relates to an artificial intelligence (AI) based automated response system (ARS).

The development of information communication technology has led to the development of an information society, and the Internet is becoming an important and indispensable medium in all fields such as society, culture, and economy.

In order to make inquiries, reservations, deliveries, etc. to various companies, there is an offline method in which a user calls the company directly or transmits the request through an agent company.

A method that is better than the offline method is the method of connecting to the customer center of related companies from the website based on the development of Internet technology, and the method of using the ARS method based on the development of mobile technology such as smartphones. , By installing and running the customer center application (App, Application), it became possible to process necessary items.

For example, Patent Literature 1 (published on March 29, 2019) discloses a technique for providing a customer center connection service on a screen to a customer based on artificial intelligence.

Korean Patent Publication No. 10-2019-0033138

To provide a method and system capable of quickly providing a reply by grasping the meaning of the utterance from mid results to the user's utterance and preparing the reply in advance.

To provide a method and system capable of interrupting the transmission of a reply when a voice signal by a user's utterance is received while the reply is being transmitted.

To provide a method and system capable of providing different response speeds according to the speed of user's speech.

To provide a method and system capable of providing a reply by translating into a language spoken by a user using real-time translation.

1. An artificial intelligence-based automated response method performed by a computer system, said computer system including at least one processor configured to execute computer readable instructions contained in a memory, said artificial intelligence-based automated response method comprising: The automatic response method comprises the step of: receiving the user's uttered voice by establishing a call with the user by the at least one processor; and using the mid results of the uttered voice by the at least one processor. An artificial intelligence-based automatic response method is provided, comprising: generating a response after analyzing an intent; and sending the response to the spoken voice by the at least one processor.

According to one aspect, the step of generating includes extracting sample sentences from a conversation log of an automatic response service, and performing conversation learning using sentences from which endings are removed from the sample sentences as learning data. It may include analyzing intent.

According to another aspect, the generating step may include analyzing the intent on a syllable-by-syllable basis obtained as the intermediate result.

According to another aspect, the generating step may include determining when to generate the reply based on confidence in an intent analysis result using the intermediate result.

According to another aspect, in the step of generating, when the confidence of the intention analysis result using the intermediate result reaches a threshold determined by conversation learning using sentences excluding the endings, the reply is generated in advance. generating.

According to yet another aspect, the step of sending may send the reply as a voice signal if an end point is detected from the spoken voice.

1. An artificial intelligence-based automated response method performed by a computer system, said computer system including at least one processor configured to execute computer readable instructions contained in a memory, said artificial intelligence-based automated response method comprising: The automatic response method comprises the steps of: receiving a speech voice of the user by establishing a call with the user by the at least one processor; sending a reply to the speech speech by the at least one processor; Provided is an artificial intelligence-based automatic response method, comprising the step of interrupting transmission of the response if at least one processor receives the user's uttered voice while the response is being transmitted.

According to one aspect, the suspending step may suspend sending the reply immediately or after applying a fade-out if the user's speech is received.

According to another aspect, the step of interrupting includes classifying and learning non-verbal expressions, and speech received while the reply is being sent is maintaining the sending of the reply if it corresponds to the grammatical expression, and suspending the sending of the reply if it does not correspond to the learned non-verbal expression.

According to another aspect of the artificial intelligence-based automatic response method, when the at least one processor lacks information necessary for a response as a result of recognition of the uttered voice, the question is returned to guide the information. It may further include providing a question.

According to another aspect, in the artificial intelligence-based automatic response method, when a plurality of intentions are recognized by the recognition result of the uttered voice, the at least one processor responds to each intention according to the recognition order of the intentions. The step of sequentially providing may also be included.

According to another aspect, the artificial intelligence-based automatic response method comprises: recognizing the speech rate of the user by the at least one processor; may further include determining the speech rate of the .

According to yet another aspect, the artificial intelligence-based auto-response method comprises: recognizing a spoken language of the user with the at least one processor; to a language model corresponding to the user's spoken language.

A non-temporary computer-readable recording medium is provided in which a program for causing a computer to execute the artificial intelligence-based automatic response method is recorded.

1. A computer system, comprising at least one processor configured to execute computer readable instructions contained in a memory, wherein the at least one processor is configured to communicate with a user to reproduce the user's speech. A computer system is provided for processing the steps of receiving, generating a reply after analyzing intent using intermediate results of said speech, and sending said reply to said speech.

1. A computer system, comprising at least one processor configured to execute computer readable instructions contained in a memory, wherein the at least one processor is configured to communicate with a user to reproduce the user's speech. A computer system for processing the steps of receiving, sending a reply to the spoken voice, and interrupting the sending of the reply if the user's spoken voice is received while the reply is being sent. I will provide a.

According to the embodiment of the present invention, the meaning of the utterance is grasped from the intermediate results of the user's utterance, and the reply is prepared before the end of the utterance, so that the reply can be quickly provided.

According to the embodiment of the present invention, by interrupting the transmission of the response when the voice signal of the user's utterance is received while the response is being transmitted, it is possible to realize a form of actually talking to a person. .

According to an embodiment of the present invention, different response speeds are provided according to the user's utterance speed, thereby providing a suitable service through an interaction adapted to the user's utterance speed.

According to embodiments of the present invention, real-time translation can be used to translate and provide responses in the language the user is speaking, thereby increasing the accessibility and convenience of services regardless of language. can.

1 is a diagram showing an example of a network environment in one embodiment of the present invention; FIG. 1 is a block diagram for explaining internal configurations of an electronic device and a server in one embodiment of the present invention; FIG. 1 is an exemplary diagram for explaining an AI automatic response system in one embodiment of the present invention; FIG. 1 is a diagram illustrating examples of components that an AI automated response system may include, in one embodiment of the present invention; FIG. 4 is a flow chart showing an example of an AI automatic response method in one embodiment of the present invention; FIG. 4 is a diagram for explaining examples of learning sentences in one embodiment of the present invention; FIG. 6 is an exemplary diagram illustrating a process of analyzing user's intention using an intermediate result of user's uttered voice according to an embodiment of the present invention; FIG. 6 is an exemplary diagram illustrating a process of analyzing user's intention using an intermediate result of user's uttered voice according to an embodiment of the present invention; FIG. 4 is a diagram for explaining an example of a process of interrupting transmission of a reply in one embodiment of the present invention; FIG. 12 is a diagram for explaining another example of the process of interrupting sending of a reply in one embodiment of the present invention; FIG. 4 is a diagram for explaining an exception situation that does not interrupt the sending of replies in one embodiment of the present invention;

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

Embodiments of the present invention relate to artificial intelligence (AI) based automated response systems (ARS).

Embodiments including the matters specifically disclosed in this specification can provide an automatic response system realized in a form of actually conversing with a human being based on artificial intelligence, thereby communicating with a user. Inquiries, reservations, delivery orders, and more can be handled quickly and conveniently while making calls more natural.

FIG. 1 is a diagram showing an example of a network environment in one embodiment of the present invention. The network environment of FIG. 1 illustrates an example including multiple electronic devices 110 , 120 , 130 , 140 , multiple servers 150 , 160 , and a network 170 . Such FIG. 1 is merely an example for explaining the invention, and the number of electronic devices and the number of servers are not limited as in FIG.

The plurality of electronic devices 110, 120, 130, 140 may be fixed terminals or mobile terminals implemented by a computer system. Examples of the plurality of electronic devices 110, 120, 130, and 140 include AI speakers, smart phones, mobile phones, navigation systems, PCs (personal computers), notebook PCs, digital broadcasting terminals, PDAs (Personal Digital Assistants), PMPs ( portable multimedia players), tablets, game consoles, wearable devices, IoT (internet of things) devices, VR (virtual reality) devices, AR (augmented reality) devices, and the like. As an example, although FIG. 1 shows an AI speaker as the electronic device 110, in embodiments of the present invention, the electronic device 110 substantially utilizes a wireless or wired communication scheme and communicates with other users via the network 170. It may refer to one of a variety of physical computer systems capable of communicating with electronic devices 120, 130, 140 and/or servers 150, 160.

The communication method is not limited, and not only the communication method using the communication network that can be included in the network 170 (eg, mobile communication network, wired Internet, wireless Internet, broadcast network, satellite network, etc.), but also the device It may also include short-range wireless communication between For example, the network 170 includes a PAN (personal area network), a LAN (local area network), a CAN (campus area network), a MAN (metropolitan area network), a WAN (wide area network), a BBN (broadband network), and the Internet. Any one or more of the networks may be included. Additionally, network 170 may include any one or more of network topologies including, but not limited to, bus networks, star networks, ring networks, mesh networks, star-bus networks, tree or hierarchical networks, and the like. will not be

Servers 150, 160 are each configured by one or more computing devices that communicate with a plurality of electronic devices 110, 120, 130, 140 over network 170 to provide instructions, code, files, content, services, etc. may be realized. For example, the server 150 may be a system that provides a first service to a plurality of electronic devices 110, 120, 130, 140 connected via the network 170, and the server 160 may also be a system that provides a plurality of electronic devices connected via the network 170. It may be a system that provides the second service to the electronic devices 110 , 120 , 130 , 140 . As a more specific example, the server 150 provides a service intended by the application (for example, automatic response service, etc.) may be provided to a plurality of electronic devices 110, 120, 130, 140 as a first service. As another example, the server 160 may provide, as a second service, a service of distributing files for installing and executing the applications described above to the plurality of electronic devices 110 , 120 , 130 , 140 .

FIG. 2 is a block diagram for explaining internal configurations of an electronic device and a server in one embodiment of the present invention. In FIG. 2, the internal configuration of the electronic device 110 and the internal configuration of the server 150 will be described as an example of the electronic device. Other electronic devices 120, 130, 140 and server 160 may also have the same or similar internal configurations as electronic device 110 or server 150 described above.

Electronic device 110 and server 150 may include memory 211 , 221 , processors 212 , 222 , communication modules 213 , 223 , and input/output interfaces 214 , 224 . The memories 211 and 221 may be non-temporary computer-readable recording media such as RAM (random access memory), ROM (read only memory), disk drive, SSD (solid state drive), flash memory. ), etc. Here, non-temporary large-capacity storage devices such as ROMs, SSDs, flash memories, and disk drives are installed in the electronic device 110 and the server 150 as non-temporary storage devices separate from the memories 211 and 221. may be included. The memories 211 and 221 also store an operating system and at least one program code (for example, a browser installed and executed in the electronic device 110 or a browser installed in the electronic device 110 to provide a specific service). code for applications, etc.) may be recorded. Such software components may be loaded from a computer-readable medium separate from memories 211,221. Such other computer-readable recording media may include computer-readable recording media such as floppy drives, disks, tapes, DVD/CD-ROM drives, memory cards, and the like. In other embodiments, software components may be loaded into memory 211, 221 through communication modules 213, 223 that are not computer-readable media. For example, at least one program is a computer program (as an example, the above-mentioned may be loaded into the memory 211, 221 based on the application).

Processors 212, 222 may be configured to process computer program instructions by performing basic arithmetic, logic, and input/output operations. Instructions may be provided to processors 212 , 222 by memory 211 , 221 or communication modules 213 , 223 . For example, processors 212 , 222 may be configured to execute instructions received according to program code stored in storage devices, such as memories 211 , 221 .

Communication modules 213 and 223 may provide functions for electronic device 110 and server 150 to communicate with each other via network 170, and electronic device 110 and/or server 150 may communicate with other electronic devices (for example, , electronics 120) or other servers (eg, server 160). As an example, requests generated by the processor 212 of the electronic device 110 according to program code recorded in a recording device such as the memory 211 may be communicated to the server 150 via the network 170 under the control of the communication module 213 . Conversely, control signals, instructions, contents, files, etc. provided under the control of the processor 222 of the server 150 are received by the electronic device 110 through the communication module 213 of the electronic device 110 via the communication module 223 and the network 170. may be For example, control signals, instructions, content, files, etc. of the server 150 received through the communication module 213 may be transferred to the processor 212 and the memory 211, and the content, files, etc. may be further stored in the electronic device 110. It may be recorded on a medium (the non-temporary recording device described above).

Input/output interface 214 may be a means for interfacing with input/output devices 215 . For example, input devices may include devices such as keyboards, mice, microphones, cameras, and output devices may include devices such as displays, speakers, tactile feedback devices, and the like. As another example, input/output interface 214 may be a means for interfacing with a device that integrates functionality for input and output, such as a touch screen. Input/output device 215 may be one device with electronic device 110 . Input/output interface 224 of server 150 may also be a means for interfacing with devices (not shown) for input or output that are connected to server 150 or that server 150 may include. As a more specific example, when the processor 212 of the electronic device 110 processes the instructions of the computer program loaded in the memory 211, service screens and content configured using data provided by the server 150 and the electronic device 120 are displayed. may be displayed on the display through input/output interface 214 .

Also, in other embodiments, electronic device 110 and server 150 may include more components than those in FIG. However, most prior art components need not be explicitly shown in the figures. For example, electronic device 110 may be implemented to include at least some of the input/output devices 215 described above, and may also include other components such as transceivers, cameras, various sensors, databases, and the like. It's okay. As a more specific example, if the electronic device 110 is an AI speaker, various sensors, camera modules, physical buttons, touch panel-based buttons, input/output ports, vibration Various components, such as a vibrator for, may be implemented to be further included in electronic device 110 .

The auto-response system provides an auto-response service platform that transmits various information regarding restaurants, accommodations, airline tickets, movies, performances, hospitals (medical care), travel, etc., based on conversations with users.

In the following, as an example, processing an inquiry to a restaurant, a reservation, a delivery order, etc. while naturally talking with a user on behalf of a restaurant clerk will be described, but this is only an example and is limited to this. It should not be used and is applicable to all companies and sectors where the use of automated response systems is possible.

FIG. 3 is an exemplary diagram for explaining the AI automatic response system in one embodiment of the present invention.

For example, the AI automated response system 300 according to embodiments of the present invention may be implemented on the server 150 described with reference to FIGS. 1 and 2. FIG.

Referring to FIG. 3, AI autoresponder system 300 may provide an autoresponder service for processing inquiries, reservations, delivery orders, etc. for multiple businesses 31-33.

The AI automatic response system 300 may provide an artificial intelligence conversation with the user 301 through a phone call or chatbot, and provides information desired by the user 301 based on the conversation with the user 301. or communicate the request of the user 301 to the companies 31-33.

The AI automatic response system 300 is realized by a database system (not shown) containing company information of each company 31-33 and a single system in order to provide an automatic response service related to the companies 31-33. Alternatively, it may be implemented in another system that can be linked. The AI automatic response system 300 may provide information desired by the user 301 or process the request of the user 301 based on company information.

For example, the AI automatic response system 300 receives a speech input by the user 301 "Can you place an order with Company A?" do. Accordingly, the AI automatic response system 300 recognizes and analyzes the voice input of the user 301 received from the electronic device 110, "Can you place an order with company A?" Please tell me the menu.", and the generated reply may be sent to the electronic device 110 by means of an audio signal. Based on the conversation with the user 301, the AI automatic response system 300 summarizes the request of the user 301, such as the selected menu and quantity, and then sends the summarized information to the relevant company (one of 31 to 33 ).

Hereinafter, specific embodiments of the AI-based automatic response method and system will be described.

FIG. 4 is a diagram illustrating examples of components that an AI automated response system may include, in one embodiment of the present invention.

The server 150 according to the present embodiment serves as a platform that provides automatic response services related to companies. In particular, the server 150 may include an AI automatic response system 300 that is implemented in a form that actually converses with humans based on artificial intelligence.

AI automated attendant system 300 may include gateway 401, speech recognizer 410, dialog manager 420, reply generator 430, speech synthesizer 440, sensing unit 450, and translator 460, as shown in FIG.

The gateway 401 is a receiving end (RX) that receives voice input from a user of the electronic device 110 through a phone function or a chatbot of an application installed in the electronic device 110, and a response voice to the electronic device 110 through an ARS response. and a transmitting end (TX), which serves to transmit. Gateway 401 may also include a socket controller for controlling the sending of the reply voice.

Speech recognizer 410 , dialog manager 420 , reply generator 430 , speech synthesizer 440 , detector 450 and translator 460 may be included as components of processor 222 of server 150 . Depending on the embodiment, components of processor 222 may be selectively included or excluded from processor 222 . Also, depending on the embodiment, the components of processor 222 may be separated or merged to represent the functionality of processor 222 .

Such processor 222 and components of processor 222 may directly process operations according to control instructions or may control server 150 for execution of the AI auto-response methods described below. . For example, processor 222 and components of processor 222 may be implemented to execute instructions according to the code of an operating system and the code of at least one program contained in memory 221 .

Here, the components of processor 222 may be representations of different functions of processor 222 executed by processor 222 according to instructions provided by program code stored on server 150 . For example, speech recognizer 410 may be utilized as a functional representation of processor 222 that controls server 150 according to the instructions described above such that server 150 recognizes speech input received from electronic device 110 .

Processor 222 may read the necessary instructions from memory 221 loaded with instructions associated with the control of server 150 . In this case, the read instructions may include instructions for controlling processor 222 to perform the AI auto-response method described below.

The AI auto-response method below does not have to occur in the order shown, some steps may be omitted, or additional steps may be included.

FIG. 5 is a flow chart showing an example of an AI automatic response method in one embodiment of the present invention.

At step 510, the processor 222 may perform conversation learning using the training data set containing the sentences with the endings removed and the responses to the corresponding sentences. For the conversation between the AI automatic response system 300 and the user, after labeling the sentences excluding the endings to the learning data level, the labeled learning data is learned by deep learning or machine learning to build a conversation learning model. you can At this time, for each of the learning sentences, a high-probability reply may also be configured together as a learning data set and used for conversation learning.

FIG. 6 is a diagram for explaining examples of learning sentences in one embodiment of the present invention.

For example, referring to FIG. 6, sample sentences 601 may be extracted from a conversation log that appears in an automatic response service, and sentences excluding the endings of the extracted sample sentences may be used as learning sentences 602 . For example, the sample sentence "Is it possible to make a restaurant reservation this Saturday?" good. By learning the remaining sentences excluding the endings of words, it is possible to learn in advance the correct answer set indicating what kind of response should be given before the end of a certain word in the middle of the conversation. Not only ending endings that end sentences, but also non-ending endings such as connecting endings in connecting clauses, inversion endings in embedded clauses, and remaining constituents other than core keywords. It is also possible to utilize sentences as learning sentences.

In other words, the AI automatic response system 300 includes a conversation learning model built in advance by learning sentences excluding endings.

Referring back to FIG. 5, at step 520, speech recognizer 410 may receive a real-time stream of the user's speech from electronic device 110 via gateway 401 once a call is established with the user of electronic device 110. At this time, the speech recognizer 410 may convert the received uttered speech into text in real time by STT (speech to text) and transmit the text to the dialog manager 420 .

At step 530, the dialog manager 420 may analyze the text communicated from the speech recognizer 410 for user intent based on natural language understanding (NLU) techniques and conversational learning. In particular, dialog manager 420 may capture the user's intent on a syllable-by-syllable basis of the text. That is, the dialog manager 420 may grasp the user's intention by dividing it into syllable units obtained as intermediate results through real-time text conversion instead of sentence units obtained as final results after the user finishes speaking. Dialog manager 420 may pre-know the meaning of user utterances at the stage of intermediate results before the end of the utterance.

At step 540, the reply generator 430 may pre-generate a reply based on the intent analysis results. In particular, the response generator 430 may determine the response generation time point based on the confidence of the intention analysis result using the intermediate result. As an example, the reply generator 430 generates a reply corresponding to the user's intention at that point in time when the confidence of the intention analysis result is equal to or greater than a predetermined threshold at the intermediate result stage before the end of the word. may be generated. Confidence for determining the time to generate a reply may be determined by conversational learning using sentences with endings removed, for example, repeating the process of finding a numerical value where the confidence for the intermediate result is within the error range compared to the final result. A confidence threshold for determining when to generate a response may then be determined based on the statistics of values obtained by the iterative process.

7 and 8 are exemplary diagrams for explaining the process of analyzing the user's intention using an intermediate result of the user's uttered voice according to an embodiment of the present invention.

Referring to FIG. 7, the user's intention is analyzed based on the intermediate result before the end of the user's utterance voice "I am going to take my parents. Is there a recommended menu for elderly people?" If so, there may emerge a threshold confidence that can generate a response when ascertaining "parents", "elderly", and even "menu". Before the end of the user's speech, that is, at the time when the user utters the ending "Are you?", it is possible to generate a reply in advance.

As another example, referring to FIG. 8, in the case of an automatic response service for making a reservation, the user's utterance "Please make a reservation for three people this Saturday." At this time, the intent analysis results indicate that if all the pre-determined required information (information slots) related to the reservation are satisfied, the confidence is reached that a response can be generated. It is possible to judge and generate a response in advance before the user finishes speaking, that is, at the time when the user utters the ending "Please."

5, at step 550, the voice synthesizer 440 may synthesize the response generated at step 540 as a voice signal, and then send it to the electronic device 110 via the gateway 401 when the user finishes speaking. . For example, the speech synthesizer 440 may determine that the user's speech is over when an end point is detected from the user's utterance, and at this time, a state in which the utterance is not detected is predetermined. It may be recognized as an endpoint if it is maintained for a certain period of time or more. The voice synthesizer 440 may synthesize a reply generated by the intermediate result as a voice signal in advance before the user finishes speaking, and may send the reply voice to the electronic device 110 after the user finishes speaking.

Therefore, the AI automatic response system 300 grasps the user's intention from the intermediate result before the user's speech ends, generates and synthesizes a reply in advance at the time when the user utters the ending, and then prepares in advance after the user's speech ends. By providing a saved reply, a response can be provided more quickly.

FIG. 9 is a flow chart showing another example of the AI automatic response method in one embodiment of the present invention. The following AI auto-response methods may be included in step 550 described above.

At step 901 , the dialog manager 420 may continuously check whether the user's spoken voice is received while the reply voice is being sent from the voice synthesizer 440 to the electronic device 110 .

In step 902, the voice synthesizer 440 may stop sending the reply voice if the user's uttered voice is received while the reply voice is being sent.

The AI automatic response system 300 may basically include a structure in which a receiving end (RX) and a transmitting end (TX) coexist as a gateway 401. At this time, the gateway 401 receives and receives the user's uttered voice. A socket controller may be included to control the delivery of the reply voice. AI automatic response system 300, if a voice signal of the user's uttered voice is input to the receiving end (RX) while the transmitting end (TX) is transmitting the response voice, the voice sent from the transmitting end (TX) You can interrupt the signal.

As an example, the voice synthesizer 440 may immediately interrupt the response voice being sent when the user's uttered voice is input. For another example, if the voice uttered by the user is input, the voice synthesizer 440 applies fade-out of a predetermined length to the response voice being sent, and then outputs the response voice. may be interrupted.

Rather than suspending the response voice for all speech responses received during the transmission of the response voice, a non-verbal filter may be used to selectively suspend the response voice.

FIG. 10 is a diagram for explaining another example of the process of interrupting transmission of a reply in one embodiment of the present invention.

Referring to FIG. 10, in step 1001, the sensing unit 450 detects the non-verbal expression of the speech received during the transmission of the reply speech in order to filter the non-verbal expression as speech that should be ignored. You can judge whether it is The sensing unit 450 may determine the non-verbal expression as a continuer that does not need to interrupt the sending of the reply during the sending of the reply, and furthermore, the non-verbal expression should not be used to grasp the user's intention. may be judged as good.

FIG. 11 is a diagram for explaining an exception situation that does not interrupt reply sending in one embodiment of the present invention.

For example, referring to FIG. 11, expressions of agreement or agreement such as [Yes, yes, yes, yes, ...], or various exclamations such as [Yeah, ah, ah, ...] Classified as verbal expressions, such non-verbal expressions may be learned. In addition, as an example of non-verbal expressions, speech containing emotional expressions may be learned. You can grasp up to the height of . The sensing unit 450 may determine whether the uttered voice received during the transmission of the response voice corresponds to the non-verbal expression based on the learning result of the non-verbal expression.

Referring back to FIG. 10, in step 1002, if the received uttered voice corresponds to a non-verbal expression, the voice synthesizer 440 determines that the received voice should be ignored and continues to transmit the response voice. If a speech voice that does not correspond to a non-verbal expression is received, it may be determined as a meaningful voice that should not be ignored, and transmission of the response voice may be interrupted.

Therefore, if the AI automatic response system 300 receives a voice signal by the user's utterance while the reply is being sent, it interrupts the sending of the reply, thereby realizing a form of actually talking with a human being. can be done.

Dialog manager 420 may provide appropriate responses to failure intervals to support natural conversations, such as those experienced with humans in automated attendant services. As an example, if the amount of result information for speech recognition is insufficient, e.g. Interrogative questions may be provided to induce utterances about the information.

As another example, if the resulting amount of information for speech recognition is too large, e.g., if more than one intent is recognized at once, the dialog manager 420 may use the A question may be provided to the user to confirm the intent in question. For example, if the user's utterance "If there is a recommended menu for elderly people, I would like to make a restaurant reservation for this Saturday." You may provide a question for user confirmation, such as "Can I make a reservation after showing you the recommended menu?"

In addition, if the question contains different intentions, for example, from the user utterance such as "Do you have a chair for children?" you can ask.

As yet another example, dialog manager 420 may provide responses in order by intent when the amount of resulting information for speech recognition is too large, eg, when more than one intent is recognized at once. For example, a numbering (first, second, etc.) may be applied to each intent according to the order of intent recognition and responses may be provided in sequence. In response to the user's utterance, "If there is a menu recommended for elderly people, I would like to make a reservation at a restaurant this Saturday." Second, what time would you like to make a reservation for this Saturday?”

User utterances such as "I would like to make a reservation. Do you have a parking lot?" However, it can also be taken as meaning that you want to make a reservation. In the present invention, it is possible to provide responses while considering such various user intentions.

As another example, the dialog manager 420 may process the same or similar responses repeatedly within a unit time or more than a predetermined number of times as a response to the failed section according to the amount of information. . As yet another example, dialog manager 420 may provide responses based on a multi-turn conversation approach that reflects previous conversation context. In addition to this, by providing appropriate responses to system utterance definition problems, natural language understanding (NLU) errors, etc., realize a natural conversation form that actually talks to humans in any situation. can be done.

In addition, the dialog manager 420 may recognize the speech rate of the user through the speech recognizer 410, and may apply different response speeds according to the speech rate of the user. As an example, the dialog manager 420 may determine the speech rate of the reply voice in proportion to the speech rate of the user, eg, a user who asks questions quickly may respond quickly to match the question rate.

Therefore, the AI automatic response system 300 provides different response speeds according to the user's speech speed, thereby providing a suitable service through interaction suitable for the user's speech speed.

が認識されれば韓国語モデルとして、「ｈｅｌｌｏ（もしもしに該当する英語）」が認識されれば英語モデルとして、「もしもし」が認識されれば日本語モデルとして、翻訳機４６０の言語モデルを分岐して自動応答サービスを提供してよい。言い換えれば、ダイアログマネージャ４２０は、外国人と電話が繋がった場合でも、最初の発話の言語を感知し、該当の言語モデルを利用してレストランへの問い合わせや予約、配達注文などのための自動応答サービスを提供することができる。 Additionally, the dialog manager 420 may utilize real-time translation to translate into the language the user is speaking to provide an automated response service. The dialog manager 420 includes a translator 460 that includes at least one language model, or may be configured in a manner that can be linked, and switches to a language model corresponding to the user's spoken language to provide an automatic response service with real-time translation. you can For example, the translator 460 may maintain a Korean model, an English model, and a Japanese model, and the sensing unit 450 may select one of the user's uttered voices as a wakeup word for setting the language. The language of the beginning part may be sensed. At this time, the dialog manager 420 may recognize the user's spoken language based on the wake-up word, and branch to a language model corresponding to the user's spoken language to provide an automatic response service. For example, the dialog manager 420 may use (outside 1) as the first word of user utterance.

is recognized, the language model of the translator 460 is branched into a Korean model, "hello (English corresponding to hello)" is recognized as an English model, and if "moshi moshi" is recognized as a Japanese model. may provide an automated response service. In other words, the dialog manager 420 senses the language of the first utterance even when a call is made with a foreigner, and automatically responds to restaurant inquiries, reservations, delivery orders, etc. using the corresponding language model. can provide services.

Therefore, the AI automatic response system 300 provides a response by translating into the language spoken by the user using real-time translation, thereby increasing the accessibility and convenience of services without being limited by language. be able to.

The apparatus described above may be realized by hardware components, software components, and/or a combination of hardware and software components. For example, the devices and components described in the embodiments include processors, controllers, arithmetic logic units (ALUs), digital signal processors, microcomputers, field programmable gate arrays (FPGAs), programmable logic units (PLUs), microprocessors, Or may be implemented using one or more general purpose or special purpose computers, such as various devices capable of executing and responding to instructions. The processing unit may run an operating system (OS) and one or more software applications that run on the OS. The processor may also access, record, manipulate, process, and generate data in response to executing software. For convenience of understanding, one processing device may be described as being used, but those skilled in the art will appreciate that the processing device may include multiple processing elements and/or multiple types of processing elements. You can understand. For example, a processing unit may include multiple processors or a processor and a controller. Other processing configurations are also possible, such as parallel processors.

Software may include computer programs, code, instructions, or a combination of one or more of these, to configure a processor to operate at its discretion or to independently or collectively instruct a processor. You can Software and/or data may be embodied in any kind of machine, component, physical device, computer storage medium, or device for interpretation by, or for providing instructions or data to, a processing device. good. The software may be stored and executed in a distributed fashion over computer systems linked by a network. Software and data may be recorded on one or more computer-readable recording media.

The method according to the embodiments may be embodied in the form of program instructions executable by various computer means and recorded on a computer-readable medium. Here, the medium may record the computer-executable program continuously or temporarily record it for execution or download. In addition, the medium may be various recording means or storage means in the form of a combination of single or multiple hardware, and is not limited to a medium that is directly connected to a computer system, but is distributed over a network. It may exist in Examples of media are magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, and It may include ROM, RAM, flash memory, etc., and may be configured to store program instructions. Other examples of media include recording media or storage media managed by application stores that distribute applications, sites that supply or distribute various software, and servers.

As described above, the embodiments have been described based on the limited embodiments and drawings, but those skilled in the art will be able to make various modifications and variations based on the above description. For example, the techniques described may be performed in a different order than in the manner described and/or components such as systems, structures, devices, circuits, etc. described may be performed in a manner different from the manner described. Appropriate results may be achieved when combined or combined, opposed or substituted by other elements or equivalents.

Accordingly, different embodiments that are equivalent to the claims should still fall within the scope of the appended claims.

300: AI Automated Response System 401: Gateway 410: Speech Recognizer 420: Dialog Manager 430: Answer Generator 440: Speech Synthesizer 450: Sensor 460: Translator

Claims (16)

An artificial intelligence-based automatic response method executed by a computer system, comprising:
The computer system includes at least one processor configured to execute computer readable instructions contained in memory;
The AI-based automatic response method includes:
receiving, by the at least one processor, a voice uttered by the user by establishing a call with the user;
pre- generating, by the at least one processor, a reply after analyzing intent using intermediate results of the spoken voice; and sending a pre-generated reply;
AI-based automatic response method. The generating step includes:
Extracting sample sentences from a conversation log of an automatic response service; and Analyzing the intention from the intermediate results by conversation learning using sentences with the endings removed from the sample sentences as learning data.
The artificial intelligence-based automatic response method according to claim 1. The generating step includes:
analyzing the intention on a syllable-by-syllable basis obtained as the intermediate result;
The artificial intelligence-based automatic response method according to claim 1. The generating step includes:
determining when to generate the reply based on the confidence of the intention analysis result using the intermediate result;
The artificial intelligence-based automatic response method according to claim 1. The generating step includes:
Determining a point in time at which the confidence in the intention analysis result using the intermediate result reaches a threshold as the point in time at which the response is generated; and
pre-generating the reply at the determined generation time before the spoken voice ends;
The threshold is determined by conversational learning using sentences excluding the endings, and after repeating the process of finding a numerical value whose confidence in the intermediate result is within an error range compared to the final result, the repeating process determined as a statistic of the values obtained by
The artificial intelligence-based automatic response method according to claim 1. The delivering step includes:
characterized in that, if an endpoint is detected from the uttered voice, the reply is sent as a voice signal,
The artificial intelligence-based automatic response method according to claim 1. An artificial intelligence-based automatic response method executed by a computer system, comprising:
The computer system includes at least one processor configured to execute computer readable instructions contained in memory;
The AI-based automatic response method includes:
receiving, by the at least one processor, the user's uttered voice by connecting the user's call;
pre-generating, by the at least one processor, a reply after analyzing intent using intermediate results of the spoken voice;
sending, by the at least one processor, a pre-generated reply to the spoken speech after the user finishes speaking; and while the reply is being sent, by the at least one processor, the user's spoken speech. is received, suspending sending said reply;
AI-based automatic response method. The suspending step includes:
interrupting the sending of the reply immediately or after applying a fade-out when the user's uttered voice is received;
The artificial intelligence-based automatic response method according to claim 7. The suspending step includes:
classifying and learning a non-verbal expression; and maintaining sending the reply if the speech received during the sending of the reply corresponds to the learned non-verbal expression; interrupting sending the reply if it does not correspond to the learned non-verbal expression;
The artificial intelligence-based automatic response method according to claim 7. The AI-based automatic response method includes:
further comprising providing, by the at least one processor, an interrogative question for guiding the information when the recognition result of the spoken voice lacks information necessary for a reply;
The artificial intelligence-based automatic response method according to claim 1 or 7. The AI-based automatic response method includes:
further comprising, when the at least one processor recognizes a plurality of intentions from the recognition result of the uttered voice, sequentially providing a reply to each intention according to an intention recognition order;
The artificial intelligence-based automatic response method according to claim 1 or 7. The AI-based automatic response method includes:
recognizing, by the at least one processor, the user's speech rate; and determining, by the at least one processor, the speech rate of the reply according to the user's speech rate.
The artificial intelligence-based automatic response method according to claim 1 or 7. The AI-based automatic response method includes:
recognizing, by the at least one processor, the user's spoken language; and switching, by the at least one processor, a language model for an automated response service to a language model corresponding to the user's spoken language. ,
The artificial intelligence-based automatic response method according to claim 1 or 7. A non-temporary computer-readable recording medium in which a program for causing a computer to execute the artificial intelligence-based automatic response method according to any one of claims 1 to 9 is recorded. a computer system,
at least one processor configured to execute computer readable instructions contained in memory;
The at least one processor
A process of receiving the user's uttered voice by connecting a call with the user;
pre- generating a reply after analyzing intent using intermediate results of said spoken voice; and
sending the pre-generated response to the spoken voice after the user finishes speaking;
computer system. a computer system,
at least one processor configured to execute computer readable instructions contained in memory;
The at least one processor
A process of receiving the user's uttered voice by connecting a call with the user;
Pre-generating a reply after analyzing intent using the intermediate result of the spoken voice;
sending out the pre-generated reply to the uttered voice after the user finishes speaking; and interrupting the sending of the reply if the user's uttered voice is received while the reply is being sent. to process
computer system.

Images