KR20220039339A - Method and apparatus for providing chatbot service based on slot filling - Google Patents

Abstract

Disclosed are a method and device for providing a chatbot service based on slot filling. According to one embodiment of the present invention, the method for providing a chatbot service includes the steps of: determining an intent corresponding to the user conversation; selecting an execution block corresponding to the intent using an intent mapping list; performing slot filling corresponding to the execution block; and performing conversation processing corresponding to the intent based on the slot filling. The intent mapping list is provided for each of the execution blocks, and the intent mapping list corresponds to at least one intent. Accordingly, the chatbot can effectively respond to various user conversation patterns.

Description

Slot filling-based chatbot service provision method and device {METHOD AND APPARATUS FOR PROVIDING CHATBOT SERVICE BASED ON SLOT FILLING}

The present invention relates to a method and apparatus for providing a chatbot service, in particular, to a slot filling-based chatbot service providing technology that allows slot filling for information that is difficult for a user to speak from the beginning or information added during a conversation with the user. it's about

Unless otherwise indicated herein, the material described in this section is not prior art to the claims of this application, and inclusion in this section is not an admission that it is prior art.

A chatbot is a computer program or artificial intelligence equipped with the ability to communicate through text or voice. There are chatbots of various levels, including chatbots to which .

Conversational AI agent system using chatbot identifies which service the user wants in response to user input and provides a response. In this case, when the user provides an input through a natural sentence, natural language processing technology may be applied to the input for processing. To this end, several utterance example sentences having the same user intent are collected and classified into one intent. It is possible to provide a chatbot service that can handle various spontaneous utterances of a user by mapping a bot response corresponding to each of the classified intents or defining an operation such as a service API call.

When one bot response unit (execution block) is mapped for each utterance unit (intent) of one user, as the service function increases, bot responses and corresponding intents also increase. Intents are allocated to intents, and when intent classification is performed, a plurality of intents having a slight score difference are generated, thereby increasing the possibility of an error in intention classification.

Even if the number of intents does not increase, if various utterance patterns are registered in an intent to recognize various natural utterances of the user, similar patterns are registered in several intents, so that the natural language processor accurately detects similar utterances between different intents. It is difficult to handle, so it is prone to errors in intent classification.

In other words, if utterances are classified according to the situation in which the intent and the bot response unit are mapped 1:1, the performance of the natural language processor decreases as the number of intents increases or the number of example utterance sentences registered in the intent increases. When utterances are classified to ensure the performance of the natural language processor, similar utterance patterns are all defined in one intent, which makes the implementation of the block mapped to the corresponding intent very complicated, and several functions actually provided are in one block. It becomes a mapping situation, and management difficulties are inevitable.

In addition, most chatbots process the bot's actions by searching for all intents when a user utters a utterance. A problem arises in which the configuration becomes complicated.

Therefore, there is an urgent need for a new chatbot service providing technology that can map intents and bot response units more freely.

Korean Patent Laid-Open Patent No. 10-2019-0006403, published on January 18, 2019 (Title: Voice processing method and system supporting the same)

An object of the present invention is to enable a chatbot to effectively respond to various user conversation patterns by setting it to operate differently based on the slot filling defined in the execution block when the same intent is used in different execution blocks.

In addition, it is an object of the present invention to perform slot filling not only for slots existing in the intent but also for completely new slots using multiple entities mapped to one slot, and through this, information that is difficult for a user to say from the beginning It is to enable slot filling for information added during conversation.

Another object of the present invention is to simplify a process for maintaining a context flow by providing a slot reuse function, and to more effectively manage slot values set for maintaining a context flow.

In addition, an object of the present invention is to enable N:N (N is a natural number) mapping between intents classified appropriately for a natural language processor and execution blocks that process intents, so that one It is to be able to execute a function and, conversely, to perform a completely different function with similar utterances.

Another object of the present invention is to enable various intent-execution block mapping relationships by using an intent mapping list including an intent identifier and a slot condition for each one or more intents.

In addition, the object of the present invention is not limited to the objects as described above, and it is obvious that other objects may be derived from the following description.

According to an embodiment of the present invention for achieving the above object, a method for providing a slot filling-based chatbot service includes determining an intent corresponding to a user conversation; selecting an execution block corresponding to the intent by using an intent mapping list; performing slot filling corresponding to the execution block; and performing dialog processing corresponding to the intent based on the slot filling, wherein the intent mapping list is provided for each of the execution blocks and corresponds to at least one intent.

In this case, performing slot filling may include performing slot filling based on at least one of an entity mapped to a slot, slots corresponding to the execution block, and intent mapping information of the slot.

In this case, the intent mapping information includes an intent name mapped to a slot, a slot name of an intent mapped to a slot, a slot value of an intent mapped to a slot, and a result value based on a slot value of an intent mapped to a slot. It may include at least one.

In this case, the slots corresponding to the execution block may include slots of intents connected to the execution block through the intent mapping list.

In this case, performing the slot filling may extract a slot value by performing synonym matching on the entity mapped to the slot.

In this case, one slot may be mappable to a plurality of entities.

In this case, the method may further include performing a slot reuse setting for maintaining a slot context in consideration of a scenario in which an execution block is included.

In this case, the slot values of the slots of the intents connected to the execution block in which the slot reuse setting has been performed may not be deleted even when the scenario is re-executed after the end of the scenario.

In this case, the intent mapping list may include an intent identifier and a slot condition for each intent.

In addition, the slot filling-based chatbot service providing apparatus according to an embodiment of the present invention, one or more processors; and an execution memory for storing at least one or more programs executed by the one or more processors; wherein the at least one program determines an intent corresponding to a user conversation, selects an execution block corresponding to the intent using an intent mapping list, and performs slot filling corresponding to the execution block. and performs dialog processing corresponding to the intent based on the slot filling, wherein the intent mapping list is provided for each of the execution blocks and corresponds to at least one intent.

In this case, the at least one program may perform slot filling based on at least one of an entity mapped to a slot, slots corresponding to the execution block, and intent mapping information of the slot.

In this case, the intent mapping information includes an intent name mapped to a slot, a slot name of an intent mapped to a slot, a slot value of an intent mapped to a slot, and a result value based on a slot value of an intent mapped to a slot. It may include at least one.

In this case, the slots corresponding to the execution block may include slots of intents connected to the execution block through the intent mapping list.

In this case, the at least one program may extract a slot value by performing synonym matching on the entity mapped to the slot.

In this case, one slot may be mappable to a plurality of entities.

In this case, the at least one program may perform slot reuse setting for maintaining a slot context in consideration of a scenario in which an execution block is included.

In this case, the slot values of the slots of the intents connected to the execution block in which the slot reuse setting has been performed may not be deleted even when the scenario is re-executed after the end of the scenario.

In this case, the intent mapping list may include an intent identifier and a slot condition for each intent.

According to the present invention, when the same intent is used in different execution blocks, the chatbot can effectively respond to various user conversation patterns by setting it to operate differently based on the slot filling defined in the execution blocks.

In addition, the present invention can perform slot filling not only for slots existing in the intent but also for completely new slots using multiple entities mapped to one slot, and through this, information or conversation that is difficult for a user to speak from the beginning can be performed. Slot filling may be enabled for information added during the process.

In addition, the present invention can simplify a process for maintaining a context flow by providing a slot reuse function, and more effectively manage slot values set for maintaining a context flow.

In addition, the present invention enables N:N (N is a natural number) mapping between intents classified appropriately for a natural language processor and execution blocks processing the intents, thereby executing one function with various different utterances. Conversely, similar utterances may perform completely different functions.

Also, according to the present invention, it is possible to define various intent-execution block mapping relationships by using an intent mapping list including an intent identifier and a slot condition for each one or more intents.

Effects of the present embodiments are not limited to the above-mentioned effects, and other effects not mentioned will be clearly understood by those skilled in the art from the description of the claims.

1 is an operation flowchart illustrating a method for providing a slot filling-based chatbot service according to an embodiment of the present invention.
2 is a diagram illustrating an example of an entity according to the present invention.
3 is a diagram illustrating an example of an intent to which the entity shown in FIG. 2 is applied.
FIG. 4 is a diagram illustrating an example of intent mapping information based on the intent shown in FIG. 3 .
5 is an operation flowchart illustrating an example of a slot filling process according to the present invention.
6 is a block diagram illustrating an apparatus for providing a slot filling-based chatbot service according to an embodiment of the present invention.
7 is a block diagram illustrating a chatbot service system according to an embodiment of the present invention.
8 is a diagram illustrating a block diagram of a 'coffee order' scenario according to an embodiment of the present invention.
9 is a diagram illustrating an example of intents used in the scenario shown in FIG. 8 .
10 is a block diagram illustrating a connection between a plurality of intents and a plurality of execution blocks.
11 is a view showing an example of the execution block coffee selection shown in FIG.
12 is a diagram illustrating an example of the execution block hot ice check shown in FIG.
13 is a diagram illustrating an example of execution block takeout confirmation shown in FIG. 9 .
9 is a block diagram illustrating a connection relationship between the execution blocks shown in FIGS. 9 to 13 .
10 is a diagram illustrating an example of connecting blocks by defining a user response standby state in an execution block.
16 is a diagram illustrating an example of mapping intents defined for each utterance pattern to a block.
17 is a diagram illustrating a case of 1:1 mapping of intents and execution blocks.
18 is a diagram illustrating an example of intent classification and block mapping according to an embodiment of the present invention.
19 is a diagram illustrating execution blocks corresponding to the responses shown in FIG. 27 .
20 is a view showing an example of prioritization according to the inclusion of the condition corresponding to the coffee order scenario illustrated in FIGS. 8 to 9 .
21 is a block diagram illustrating an example of modules required to execute a chatbot.
22 is a diagram illustrating a process of searching for an intent and a block while the chatbot is running.
23 is an operation flowchart illustrating a method for providing a chatbot service according to another embodiment of the present invention.
24 is a graph showing the performance of a natural language processor according to an increase in the number of utterance patterns.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. However, detailed descriptions of well-known functions or configurations that may obscure the gist of the present invention in the following description and accompanying drawings will be omitted. Also, it should be noted that, throughout the drawings, the same components are denoted by the same reference numerals as much as possible.

The terms or words used in the present specification and claims described below should not be construed as being limited to their ordinary or dictionary meanings, and the inventor is appropriate as a concept of terms for describing his invention in the best way. It should be interpreted as meaning and concept consistent with the technical idea of the present invention based on the principle that it can be defined in Therefore, the embodiments described in this specification and the configurations shown in the drawings are only the most preferred embodiment of the present invention, and do not represent all the technical spirit of the present invention, so at the time of the present application, various It should be understood that there may be equivalents and variations. In addition, terms such as 1st, 2nd, etc. are used to describe various components, and are used only for the purpose of distinguishing one component from other components, and are not used to limit the components.

In the conventional chatbot service, the slot value is extracted by using only the user's initial utterance or by performing synonym matching based on only one entity mapped to the slot. There was a disadvantage in that slot filling was difficult for the added information.

In addition, when performing slot filling, simply setting whether a slot dependent on an intent or a prompt message It can also lead to inefficient results.

Accordingly, the present invention intends to propose a slot filling-based chatbot service providing technology that can solve the above problems.

1 is an operation flowchart illustrating a method for providing a slot filling-based chatbot service according to an embodiment of the present invention.

Referring to FIG. 1 , in the method for providing a chatbot service based on slot filling according to an embodiment of the present invention, an intent corresponding to a user conversation is determined ( S110 ).

In this case, an intent corresponding to the user conversation may be selected from among intent candidates connected to one or more candidate blocks or main blocks connected to the current block through a graph.

In this case, the intent corresponding to the user conversation may be determined by considering the first intent candidate connected to the candidate block in preference to the second intent candidate connected to the main block.

Also, in the slot filling-based chatbot service providing method according to an embodiment of the present invention, an execution block corresponding to an intent is selected using an intent mapping list (S120).

In this case, the intent mapping list is provided for each of the execution blocks, and may correspond to at least one intent.

In this case, the intent mapping list may include an intent identifier and a slot condition for each intent.

In this case, the slot condition may correspond to a conditional statement related to one or more slots.

In this case, the slot condition may include a null condition.

In this case, the intent mapping list may include intent identifiers and slot conditions corresponding to a plurality of intents (N:1), and in the case of an intent, execution blocks different from each other have the same intent identifier. If the slot conditions are different, it can be connected (1:N) to different execution blocks.

In this case, the slots corresponding to the execution block may include slots of intents connected to the execution block through the intent mapping list.

In this case, the execution block corresponding to the intent is a first execution block having a slot condition inclusion relationship corresponding to the large set when the intent corresponding to the user conversation is connected to a plurality of execution blocks. A second execution block having an inclusion relationship of a slot condition corresponding to a small set to be included may be considered and selected with priority. In this case, the inclusion relationship of the slot condition may be the inclusion relationship of slots included in the slot condition. For example, an execution block corresponding to $level == "regular member" and $service == "subscription" may be considered and selected prior to an execution block corresponding to $service == "subscription".

In addition, the slot filling-based chatbot service providing method according to an embodiment of the present invention performs slot filling corresponding to the execution block (S130).

In this case, the slot filling may refer to a process of receiving an input of a slot value that can be inputted into the slot from the user.

In this case, slot filling may be performed based on at least one of an entity mapped to a slot, slots corresponding to an execution block, and intent mapping information of the slot.

For example, if the user correctly said the slot value through a conversation with the chatbot, slot filling can be performed based on the entity mapped to the slot, and the user's sentence corresponding to the intent corresponding to the current execution block. , slot filling can be performed based on slots corresponding to the execution block.

In this case, the slot value may be extracted by performing synonym matching on the entity mapped to the slot. That is, entities mapped to the slot may indicate which slot value can be input to the slot.

For example, assuming that the 'coffee' entity is mapped to the $coffee slot included in the 'order_coffee' intent, a slot value corresponding to the 'coffee' entity may be input to the slot value of the $coffee slot.

For another example, assuming that the 'juice' entity is mapped to the $juice slot included in the 'order_juice' intent, a slot value corresponding to the 'juice' entity may be input to the slot value of the $juice slot. .

In this case, one slot may be mappable to a plurality of entities.

In this case, in the present invention, by mapping a plurality of entities to one slot, it is possible to respond to various situations that may occur through user utterance.

For example, if you can order both coffee and juice, such as 'order a drink' instead of an intent with a set category, such as 'order coffee' or 'order juice', it is stored in the $drink slot included by the 'order_drink' intent. By mapping a plurality of 'coffee' entities, 'juice' entities, and 'drinks' entities, it is possible to define various beverages that the user utters.

At this time, according to an embodiment of the present invention, a slot to which an entity is not mapped may exist. In this case, the slot value may be freely set in the user's utterance based on a method other than entity synonym matching.

In this case, the synonym matching for the entity is matching whether a value set as a synonym of the entity in the user's speech content exists. If a synonym of the entity is detected in the user's speech content, a slot value corresponding to the entity can be entered.

For example, it may be assumed that the join entity, register entity, cancel entity, and withdraw entity shown in FIG. 2 are mapped to the $action slot included in the 'member_action' intent. At this time, if an utterance "I want to join" is input from the user, entity synonym matching is performed to input the representative word 'join' corresponding to the join entity as a slot value.

In this case, in the present invention, slot filling may be performed based on the intent mapping information of the slot in order to additionally perform slot filling even for intents that do not exist in the intent corresponding to the current execution block.

At this time, in the present invention, when performing slot filling, based on the intent mapping information, slot filling can be performed not only on slots existing in the intent but also on completely new slots. can Through this process, it is possible for the chatbot to respond flexibly to various situations that may occur due to user utterance.

For example, when the user says "HDMI connection is not working", the chatbot can respond "Please enter the serial number on the back to confirm the model".

For another example, when the user says "Please cancel the purchase", the chatbot may send a verification code to the user's mobile phone, and then respond "Please enter the verification code sent to the mobile phone to verify your identity."

That is, information that is difficult for the user to speak from the beginning or information added during a conversation may also be slot-filled.

In this case, the intent mapping information includes an intent name mapped to a slot, a slot name of an intent mapped to a slot, a slot value of an intent mapped to a slot, and a result value based on a slot value of an intent mapped to a slot. It may include at least one.

Hereinafter, a process of performing slot filling based on intent mapping information will be described in detail with reference to FIGS. 2 to 4 .

For example, assuming that an entity is defined as shown in FIG. 2 and an intent and a slot are defined as shown in FIG. 3, since the join entity and the cancel entity are mapped in the 'join_confirm' slot, ' Intent mapping information for the 'join_confirm' slot may be defined as shown in FIG. 4 .

For example, in the case of a 'join_confirm' intent located first among the intents shown in FIG. 4 , it may be mapped using an intent name mapped to a slot.

At this time, referring to FIG. 3 , since the 'join_confirm' slot is included in the 'join_confirm' intent, a slot value can be input corresponding to the 'join_confirm' intent.

For another example, in the case of a 'positive' intent located second among the intents shown in FIG. 4, mapping is performed using a result value based on the intent name mapped to the slot and the slot value of the intent mapped to the slot. it may have been

That is, if the user answers "yes" to the bot's "Do you want to join?" question, it can be classified as a 'positive' intent based on FIG. 3, and the result value of 'join' according to the intent mapping information It can be entered as the slot value of the 'join_confirm' slot.

As another example, in the case of the 'member_action' intent located at the third position among the intents shown in FIG. 4, it may be mapped using the intent name mapped to the slot and the slot name of the intent mapped to the slot. .

That is, if the user answers "Do you want to join?" to the bot's "Do you want to join?" question, it may be classified as a 'member_action' intent based on FIG. 3 . At this time, the 'action' slot of the 'member_action' intent is mapped with a 'join' entity that can match 'join' as a synonym, so 'join', which corresponds to the entity representative word, is the slot value of the 'join_confirm' slot. can be entered as

For another example, in the case of the 'member_action' intent located at the fourth position among the intents shown in FIG. 4 , the intent name mapped to the slot, the slot name of the intent mapped to the slot, and the intent mapped to the slot The mapping may be performed using both the slot value and the result value of the slot value of the intent mapped to the slot.

That is, if the user answers "Would you like to register for swimming" to the bot's "Do you want to join?" question, it may be classified as a 'member_action' intent based on FIG. 3 . At this time, the 'action' slot of the 'member_action' intent is mapped with a 'register' entity that can match 'registration' as a synonym, and the slot value of the 'action' slot of the 'member_action' intent is 'register' Since 'join' is mapped to the result value in the case of 'join', 'join' may also be input to the slot value of the 'join_confirm' slot.

Hereinafter, a slot filling process according to an embodiment of the present invention will be described in more detail with reference to FIG. 5 .

According to an embodiment of the present invention, it is determined whether or not slot filling is required for the slot designated by the slot component while sequentially processing the components in the execution block (S502), and when the slot value is empty and slot filling is required, the chatbot After requesting an input from the user through a prompt message (Bot Prompt) (S504), the user may wait for a response (Slot Listen) (S506).

Thereafter, when a user message is input by the user by utterance (S508), synonym matching may be performed with synonyms of an entity mapped to a target slot to perform slot filling on a sentence corresponding to the input message (S510) .

After that, it is determined whether there is a slot value extracted by entity synonym matching (S512), and if there is a slot value extracted by entity synonym matching as a result of the determination in step S512, action validation is performed to determine the slot value After performing verification for ( S522 ), if the verification is successful, the next block component can be executed.

At this time, if, as a result of determination in step S512, entity synonym matching fails and there is no slot value extracted by entity synonym matching, the slot value is extracted based on the intents declared in the execution block. It can be (S514).

That is, the slot value can be extracted based on the slots included in the intents declared in the execution block.

After that, it is determined whether there is a slot value extracted based on the block intents declared in the execution block (S516), and as a result of the determination in step S516, the block intents declared in the execution block are determined. intent), after performing action validation to verify the slot value (S522), if the verification is successful, the next block component can be executed.

At this time, as a result of the determination in step S516, extraction using the block intents declared in the execution block fails, and there is no slot value extracted based on the block intents declared in the execution block. Otherwise, the slot value may be extracted using the intent mapping information of the target slot (S518).

After that, it is determined whether there is a slot value extracted using the intent mapping information (S520), and if there is a slot value extracted using the intent mapping information as a result of the determination in step S520, action validation is performed After verifying the slot value ( S522 ), if the verification is successful, the next block component may be executed.

At this time, if it is determined in step S520 that the extraction using the intent mapping information fails to fill the slot value of the target slot, it may be determined whether or not to retry slot filling (S524).

If it is determined in step S524 that slot filling can be re-attempted, the user is guided to retry and requested to re-enter (S526), and the process can be performed again from step S504.

In this case, if the number of retries exceeds the number of retries as a result of the determination in step S524 or retry is impossible, the slot filling may be treated as failed.

In addition, according to an embodiment of the present invention, even if the slot value for the target slot is filled and action validation is performed according to step S522, but the verification of the slot value fails, whether to retry slot filling is determined. It can be (S528).

If it is determined in step S528 that slot filling can be re-attempted, the retry is guided and requested to be input again (S530), and the process can be performed again from step S504.

In this case, if the number of retries exceeds the number of retries as a result of the determination in step S528 or retry is impossible, the slot filling may be treated as failed.

In addition, although not shown in FIG. 1, the slot filling-based chatbot service provision method according to an embodiment of the present invention may perform slot reuse setting to maintain a slot context in consideration of a scenario including an execution block. .

That is, when the user utters a utterance, it is possible to determine whether to reuse the slot value of the corresponding slot or to receive a new input according to reuse setting information of the slot connected to the block to be executed.

In this case, the slot reuse setting may be performed in consideration of various situations according to the type of scenario.

For example, in the case of the 'coffee order' scenario, the 'coffee order change' scenario may operate after the scenario ends and all slot values for coffee order are input. In this case, you should not set up slot reuse to force you to re-query from scratch for filling the slot while running the 'change coffee order' scenario.

As a detailed example, in the 'coffee order' scenario, when the user's utterance of "coffee please" is first received, the chatbot may respond with "what coffee would you like?"

After that, when an utterance of “{Americano:coffee} please” is received from the user, the chatbot may respond with “Would you like some warm or iced?”

After that, when an utterance of "Please send to {ice: hotice}" is received from the user, the chatbot may respond with "Are you going to eat?"

After that, when an utterance "{takeout:takeout}" is received from the user, the chatbot uses the slot value entered in each slot ($coffee, $hotice, $takeout) to say "Iced Americano takeout order has been placed" ', the 'order coffee' scenario may end by executing an utterance confirming the order.

At this time, assuming that slot reuse is set for each slot ($coffee, $hotice, $takeout) and an utterance "I will change to {latte:coffee}" is received from the user, 'Coffee order change' 'In the scenario, the slot values of $hotice and $takeout, which are slots that the user did not say, are maintained, so that the chatbot can immediately respond with "Changed to ice latte takeout".

That is, in this case, since slot filling must be performed again on the slot values of $hotice and $takeout, it is possible to set not to reuse the slot.

Therefore, if slot reuse is not set for each of the slots ($coffee, $hotice, $takeout) and an utterance "{cappuccino:coffee} please" is received from the user again after the above order, the user The slot values of $hotice and $takeout, which are unspoken slots, are initialized and erased, and the chatbot can ask the user to refill the deleted slot values.

Conventionally, in order to maintain the slot context, input and output properties were set for each intent unit, and the set values had to be managed each time before/after the intent. Therefore, it is difficult to manage when whether or not to maintain the slot context is changed according to the scenario. there was

In the present invention, this problem can be solved through the slot reuse setting, and the context by the previous intent can be maintained more effectively.

In this case, the slot values of the slots of the intents connected to the execution block for which the slot reuse setting has been performed may not be deleted even when the scenario is executed again after the end of the scenario.

For example, according to the present invention, when extracting a slot from an intent mapped corresponding to a user conversation, in the case of slots that are set for slot reuse, if the user does not say a slot, the already filled slot value is stored without deleting it. can be processed In this case, since the slot context needs to be maintained even after the scenario ends, the slot values existing in the context can be maintained without being deleted.

In addition, in the slots for which slot reuse is not set, if the slot is not mentioned by the user when extracting the slot from the intent mapped to correspond to the user conversation, the slot value may be initialized.

Through this process, the context can be maintained without managing separate input and output properties, so that the service can be efficiently provided without burdening the chatbot server.

In this case, if necessary in consideration of the conditions set in the slot, the slot value may be directly changed by using an assign action or a reset action.

Also, in the method for providing a chatbot service based on slot filling according to an embodiment of the present invention, a conversation process corresponding to an intent is performed based on the slot filling ( S140 ).

As described above, according to the present invention, in the intent, slot information is defined only based on information necessary from the point of view of natural language processing, and slot filling information required to execute a function is defined in the execution block, so that other execution blocks using the same slot are separated from each other. It may operate differently according to the slot filling information defined in the execution block.

That is, the slot information defined in the intent may not be completely fixed, but may mean a default setting level, and may correspond to a concept that can be redefined by an execution block.

With the above configuration, in the chatbot service providing method according to an embodiment of the present invention, when the same intent is used in different execution blocks, various user dialogs are set to operate differently based on the slot filling defined in the execution blocks. You can make your chatbot respond effectively to patterns.

In addition, the chatbot service providing method according to an embodiment of the present invention can perform slot filling not only for slots existing in the intent but also for completely new slots by using a plurality of entities mapped to one slot, through which Slot filling may be enabled for information that is difficult for the user to say from the beginning or information added while conducting a conversation.

In addition, the chatbot service providing method according to an embodiment of the present invention can simplify the process for maintaining the context flow by providing a slot reuse function, and more effectively manage slot values set for maintaining the context flow.

6 is a block diagram illustrating an apparatus for providing a chatbot service according to an embodiment of the present invention.

Referring to FIG. 6 , an apparatus for providing a chatbot service according to an embodiment of the present invention includes a processor 620 and an execution memory 630 . According to an embodiment, the chatbot service providing apparatus may further include a communication unit 610 .

The communication unit 610 may serve to transmit information related to a chatbot service with a plurality of other devices through a communication network such as a network. In this case, the network provides a path for transferring data between devices, and is a concept that encompasses both an existing network and a network that can be developed in the future.

For example, the network is IP Network, which provides large-capacity data transmission and reception service and data service without interruption through Internet Protocol (IP), and All IP, which is an IP network structure that integrates different networks based on IP. ) network, etc., wired network, Wibro (Wireless Broadband) network, 3G mobile communication network including WCDMA, HSDPA (High Speed Downlink Packet Access) network and 3.5G mobile communication network including LTE network, 4 including LTE advanced It may be achieved by combining one or more of a generation mobile communication network, a satellite communication network, and a Wi-Fi network.

For example, a network is a wired/wireless local area network that provides communication of various information devices within a limited area, a mobile communication network that provides communication between and between mobile devices and between a mobile device and the outside of the mobile device, and provides communication between an earth station and an earth station using satellites. It may be either a satellite communication network or a wired/wireless communication network, or a combination of two or more. Meanwhile, the transmission method standard of the network is not limited to the existing transmission method standard, and may include all transmission method standards to be developed in the future.

The execution memory 630 stores at least one or more programs executed by the processor 620 .

In this case, the at least one program determines an intent corresponding to the user conversation, selects an execution block corresponding to the intent using the intent mapping list, performs slot filling corresponding to the execution block, and slot filling Conversation processing corresponding to the intent is performed based on

In this case, the intent mapping list is provided for each of the execution blocks, and may correspond to at least one intent.

In this case, the at least one program may perform slot filling based on at least one of an entity mapped to a slot, slots corresponding to the execution block, and intent mapping information of the slot.

In this case, the intent mapping information includes an intent name mapped to a slot, a slot name of an intent mapped to a slot, a slot value of an intent mapped to a slot, and a result value based on a slot value of an intent mapped to a slot. It may include at least one.

In this case, the slots corresponding to the execution block may include slots of intents connected to the execution block through the intent mapping list.

In this case, the at least one program may extract a slot value by performing synonym matching on the entity mapped to the slot.

In this case, one slot may be mappable to a plurality of entities.

In this case, the at least one program may perform slot reuse setting for maintaining a slot context in consideration of a scenario in which an execution block is included.

In this case, the slot values of the slots of the intents connected to the execution block in which the slot reuse setting has been performed may not be deleted even when the scenario is re-executed after the end of the scenario.

In this case, the intent mapping list may include an intent identifier and a slot condition for each intent.

The device illustrated in FIG. 6 may receive various information such as number and character information, and may transmit signals input in connection with setting various functions and controlling functions of the device to the processor 620 through the input unit. In addition, the input unit may be configured to include at least one of a keypad and a touch pad that generate an input signal according to a user's touch or manipulation. In this case, the input unit of the device may be configured in the form of a single touch panel (or touch screen) together with the display unit to simultaneously perform input and display functions. In addition, as the input unit, all types of input means that may be developed in the future may be used in addition to input devices such as a keyboard, a keypad, a mouse, and a joystick.

In addition, the display unit may display information about a series of operation states and operation results that occur while the function of the device is performed. In addition, the display unit may display a menu of the device, user data input by a user, and the like. Here, the display unit is a liquid crystal display (LCD, Liquid Crystal Display), an ultra-thin liquid crystal display (TFT-LCD, Thin Film Transistor LCD), a light emitting diode (LED, Light Emitting Diode), an organic light emitting diode (OLED, Organic LED), It may be composed of an active organic light emitting diode (AMOLED, Active Matrix OLED), a retina display, a flexible display, and a three-dimensional display. In this case, when the display unit is configured in the form of a touch screen, the display unit may perform some or all of the functions of the input unit.

In addition, the storage unit of the device is a device for storing data, and includes a main storage device and an auxiliary storage device, and may store an application program necessary for a functional operation of the device. The storage unit may largely include a program area and a data area. Here, when each function is activated in response to a user's request, the device executes corresponding application programs under the control of the processor 620 to provide each function.

Also, the communication unit 610 may perform a function for transmitting and receiving data through a network. Here, the communication unit 610 may include an RF transmitting unit for up-converting and amplifying the frequency of the transmitted signal, and an RF receiving unit for low-noise amplifying the received signal and down-converting the frequency. The communication unit 1810 may include at least one of a wireless communication module and a wired communication module. In addition, the wireless communication module is a configuration for transmitting and receiving data according to a wireless communication method, and when the device uses wireless communication, the device transmits data using any one of a wireless network communication module, a wireless LAN communication module, and a wireless fan communication module. can be sent and received. In addition, the wired communication module is for transmitting and receiving data by wire. The wired communication module may connect to a network through a wire and transmit/receive data to/from the device. That is, the device accesses the network using a wireless communication module or a wired communication module, and may transmit/receive data to and from other devices through the network.

Also, the processor 620 may be an operating system (OS) and a process device for driving each component.

With the above configuration, the apparatus for providing a chatbot service based on slot filling according to an embodiment of the present invention is set to operate differently based on the slot filling defined in the execution block when the same intent is used in different execution blocks This allows the chatbot to effectively respond to various user conversation patterns.

In addition, the slot filling-based chatbot service providing apparatus according to an embodiment of the present invention can perform slot filling not only for slots existing in the intent but also for completely new slots by using several entities mapped to one slot. Through this, slot filling can be made possible even for information that is difficult for the user to say from the beginning or information added while conducting a conversation.

In addition, the apparatus for providing a slot filling-based chatbot service according to an embodiment of the present invention provides a slot reuse function, thereby simplifying the process for maintaining the context flow, and more effectively managing the slot values set for maintaining the context flow. can

In addition, the present invention classifies intents corresponding to natural language input according to similar properties according to natural language processing performed by an interactive AI agent system, and maps a plurality of classified intents to blocks that are processing units of bots. It's about technology.

For a useful chatbot service, it is necessary to continuously improve the natural language processing ability for terms by business or service domain and the processing ability for the provided functions or tasks.

To this end, it is necessary to train the natural language processor so that various domain-specific terms and sentences can be processed by the natural language processor. In other words, it is necessary to define intents from the point of view of Natural Language Understanding (NLU) rather than defining intents only in units of functions or tasks.

On the other hand, in order to effectively process functions or tasks, detailed processing steps or flows need to be defined as execution blocks, and a connection relationship (precedence relationship) between the defined execution blocks needs to be defined. In this case, the connection relationship is preferably defined in the form of a graph for freely connecting various execution blocks rather than in the form of a list or tree. In this case, it is preferable that the execution blocks are defined appropriately for the function provided by the service rather than from the point of view of natural language processing.

In this way, it is necessary to be able to freely (N:N) map intents and blocks defined in a form suitable for each purpose, in order to efficiently provide a chatbot service that meets the user's intention. That is, one intent can be used for multiple blocks, and one block must be able to process multiple intents. For this, intents and blocks must be mapped N:N rather than 1:1 mapping .

When a large amount of utterance patterns process a large amount of data classified into various intents, collisions between similar utterances occur, and verification and processing are required.

In this case, similar utterance patterns can be grouped with the same intent, and conditions can be given based on slots to respond so that different bot responses are possible.

In addition, if it is possible to map a plurality of user utterance units to one bot response unit (execution block) so that it can be processed with the same response although it is not a similar utterance, it is possible to process the same response even if the utterance types are different This enables efficient intent management.

Hereinafter, the chatbot service is a concept that covers not only the voice service based on the user's voice but also the messenger service based on the text input by the user.

The design and implementation of a chatbot can be largely divided into three parts: a chatbot builder, a natural language processing (NLU), and a dialog management (DM). At this time, the chatbot builder structures and defines the data for the dialogue flow processing (DM) to operate, and the dialogue flow processing (DM) processes the bot response based on this data.

A chatbot process may need two work areas to run. One is the work area where the bot designer defines the bot driving process, and the other is the execution area where the defined process is executed.

7 is a block diagram illustrating a chatbot service system according to an embodiment of the present invention.

Referring to FIG. 7 , the chatbot service system includes a chatbot 710 , a natural language processor 720 , a chatbot resource database 730 , and a chatbot builder 740 .

The chatbot 710 receives an utterance sentence from a user, generates a response thereto, and provides a chatbot response to the user.

The natural language processor 720 receives the spoken sentence from the chatbot, receives the intent definition information from the chatbot resource database 730, generates an intent classification result and a slot extraction result, and provides it to the chatbot 710 .

The chatbot resource database 730 receives intent definition information, block information, response information and scenario information, etc. from the chatbot builder, and provides block information, response information and scenario information to the chatbot 710, and a natural language processor 720 to provide intent definition information.

The chatbot builder 740 receives an input about the chatbot resource from the bot designer, generates the chatbot 710, and provides intent definition information, block information, response information and scenario information to the chatbot resource database 730. .

In the chatbot builder 740, a slot, which is important information for learning speech sentences and bot processing, is defined as an intent, the defined intent is mapped to an execution block, and conditions are set in the execution block, and the action of the bot is performed accordingly. can be defined In addition, blocks that can be processed in the response waiting state in the chatbot builder 740 are set, and these blocks may be defined as a graph-type scenario.

At this time, the bot designer works to define the bot driving process through the chatbot builder 740 . First, a block diagram in which bot scenarios for main business processing are defined in a graph form may be provided.

8 is a diagram illustrating a block diagram of a 'coffee order' scenario according to an embodiment of the present invention.

Referring to FIG. 8 , in the block utterance scenario of the coffee ordering chatbot, the user's utterance "I will order coffee" is first received (811).

Also, in the block utterance scenario of the coffee ordering chatbot, the chatbot executes the utterance "What coffee would you like?" (821).

In response to the chatbot's utterance, in the block utterance scenario of the coffee ordering chatbot, the user utters the desired coffee name (812).

In response to the user's utterance, in the block utterance scenario of the coffee ordering chatbot, the chatbot executes the utterance "Would you like something hot or iced?" (822).

In response to the chatbot's utterance, in the block utterance scenario of the coffee ordering chatbot, the user selects hot or ice (813).

In response to the user's utterance, in the block utterance scenario of the coffee ordering chatbot, the chatbot executes the utterance "Would you like to eat and go?" (823).

In response to the chatbot's utterance, in the block utterance scenario of the coffee ordering chatbot, the user selects in-store dining or takeout (814).

When all three slots of $ (coffee name), $ (hot ice), and $ (takeout) are input, in the block utterance scenario of the coffee ordering chatbot, the chatbot executes an utterance confirming the order (1324).

In response to the chatbot's utterance, in the coffee order chatbot's block utterance scenario, the user utters 'yes' (815) or 'no' (816), and when the user says 'yes', the chatbot executes a coffee order Execute a custom action (825) to confirm the order details (826), and when the user says 'no', the chatbot executes the utterance "Cancel the order" ( 827), then the operation is terminated.

In FIG. 8 , “User” indicates a user’s utterance, “templete” indicates a chatbot utterance, and “Custom Action” indicates an action processing of the bot defined by an execution block.

In the example shown in FIG. 8 , $ (coffee name), $ (hot ice), $ (takeout), and $ (order number) correspond to required slots, and although not explicitly shown in FIG. 8 , coffee In the block firing scenario of the order chatbot, more option slots can be used. In this case, the mandatory slot and/or the optional slot may be generated from the user's utterance, or may be a slot indicating the status of other chatbot systems, or generated according to a condition in a block.

Based on the diagram shown in FIG. 8 , all utterance patterns that can be processed in each block may be collected, and the patterns may be grouped in a form that is easy for the natural language processor to classify. In addition, when an intent is classified based on the grouped information and an intent is set by tagging a slot for each utterance pattern, learning data of a natural language processor (NLU) according to the learning intention based on the grouped and classified intent can be used as

Scenarios can be viewed as connecting blocks in the form of a graph, and in each block, a list of intents that can be processed can be defined.

9 is a diagram illustrating an example of intents used in the scenario shown in FIG. 13 .

Referring to FIG. 9 , the intents used in the coffee order scenario include intent order_coffee 910 , intent order_hotice 920 , intent order_takeout 930 , intent order_togo 940 , and intent order_eatin 950 . It can be seen that it contains

At this time, the slot $coffee, the slot $hotice, and the slot $takeout are defined for the intent order_coffee 910, the slot $hotice is defined for the intent order_hotice 920, and the slot for the intent order_takeout 930 $takeout is defined, and slots may not be defined for intent order_togo 940 and intent order_eatin 950 .

When the bot processes the intents shown in FIG. 9, it is processed in a unit called a block (processing block), and a plurality of intents may be mapped to one block and processed. Also, one intent may be mapped to a plurality of blocks.

10 to 13 are diagrams illustrating examples of execution blocks used in the scenario shown in FIG. 9 .

10 is a block diagram illustrating a connection between a plurality of intents and a plurality of execution blocks.

Referring to FIG. 10 , it can be seen that the intent order_coffee 910 , the intent order_hotice 920 , and the intent order_takeout 930 are connected to the execution block coffee selection 1010 . In this case, the intent mapping list of the execution block coffee selection 1010 may include intent identifiers corresponding to the intent order_coffee 910 , the intent order_hotice 920 and the intent order_takeout 930 .

At this time, the intent order_hotice 920 and the intent order_takeout 930 are also connected to the execution block hot ice check 1020 .

At this time, the intent order_takeout 930 is also connected to the execution block takeout check 1030 .

The intent order_togo 940 , the intent order_eatin 950 , the intent followup.yes 960 , and the intent followup.no 970 are connected to the execution block takeout confirmation 1030 .

As shown in FIG. 10 , intents are mapped to blocks for execution, and conditions are grammatically defined based on a variable called slot for each mapped intent, so that the bot can only classify utterances by user utterance. Appropriate detailed processing is possible through conditions without processing.

As shown in FIG. 10 , a plurality of intents may be connected to one block (N: 1), and a plurality of blocks may be connected to one intent (1:N), so that there is free space between intents and blocks. Mapping (N:N) is possible.

Furthermore, one execution block may have a connection relationship with other executable blocks that can be executed later. For example, the execution block coffee selection 1010 may be connected to the execution block hot ice confirmation 1020 and the execution block takeout confirmation 1030, and the execution block hot ice confirmation 1020 is the execution block takeout confirmation 1030 ) can be associated with

11 is a view showing an example of the execution block coffee selection shown in FIG.

Referring to FIG. 11 , the execution block coffee selection 1010 may include a linked intent list, slot, slot filling, and action information.

In this case, the connected intent list may be included in the intent mapping list, and the intent mapping list may include an intent identifier and a slot condition for each intent. In this case, the slot condition may correspond to a conditional statement related to one or more slots, and may include a null condition.

The slots ($coffee, $hotice, $takeout) shown in FIG. 11 may include the union of all slots corresponding to the intents (order_coffee, order_hotice, order_takeout) to which the execution block coffee selection 1010 is connected.

The slot filling order information may include information on the filling order of the slots of the execution block coffee selection 1010 and utterance information necessary to fill the corresponding slot first.

The action information may consist of a condition and a response when the corresponding condition is satisfied. In this case, the condition may correspond to one or more conditional statements using slots, and the response may include a link that calls another block.

For example, the execution block coffee selection 1010 may connect the execution block hot ice confirmation 1020 based on condition determination using the slot $hotice.

For example, the execution block coffee selection 1010 may connect the execution block takeout check 1030 based on condition determination using the slot $takeout.

FIG. 12 is a diagram illustrating an example of the execution block hot ice check shown in FIG. 10 .

Referring to FIG. 12 , the execution block hot ice check 1020 may include connected intent list, slot, slot filling, and action information.

The slots ($hotice, $takeout) shown in FIG. 12 may include the union of all slots corresponding to the intents (order_hotice, order_takeout) to which the execution block hot ice check 1020 is connected.

The slot filling order information may include information on the filling order of the slots of the execution block hot ice check 1020 and utterance information required to fill the corresponding slot first.

The action information may consist of a condition and a response when the corresponding condition is satisfied. In this case, the condition may correspond to one or more conditional statements using slots, and the response may include a link that calls another block.

For example, the execution block hot ice check 1020 may connect the execution block takeout check 1030 based on the condition determination using the slot $takeout.

13 is a diagram illustrating an example of execution block takeout confirmation shown in FIG. 10 .

Referring to FIG. 13 , the execution block takeout check 1030 may include a linked intent list, slot, slot filling, and action information.

The slot $takeout shown in FIG. 13 may include the union of all slots corresponding to the intents (order_takeout, order_togo, order_eatin, followup.yes, followup.no) to which the execution block takeout check 1030 is connected. there is.

The slot filling order information may include information about the filling order of the slots of the execution block takeout check 1030 and utterance information necessary to fill the corresponding slot first.

The action information may consist of a condition and a response when the corresponding condition is satisfied. In this case, the condition may correspond to one or more conditional statements using slots, and the response may include a link that calls another block.

14 is a block diagram illustrating a connection relationship between the execution blocks shown in FIGS. 10 to 13 .

Referring to FIG. 14 , it can be seen that the execution block coffee selection 1010 , the execution block hot ice confirmation 1020 , and the execution block takeout confirmation 1030 are connected in a graph form.

In this case, the connection between the execution blocks may mean a connection with the next execution blocks capable of processing the user's utterance when entering the user utterance standby state in a specific current block. That is, when a user's utterance is received from a specific current block, possible situations are execution blocks connected to the current block, so only intents connected to these execution blocks can be searched as intents requiring user utterance processing. In the waiting state for user utterance, if only the intents included in the sub-blocks connected based on the current block are collected and the intent is searched through the natural language processor, it corresponds to the utterance much more quickly and accurately than when all intents are searched. Intent can be identified.

In the example shown in FIG. 14 , the action block coffee selection 1010 is connected with the action block hot ice selection 1020 and the action block takeout confirmation 1030 , so that the user utterance in the state of the action block coffee selection 1010 is When input, the next state may be the execution block hot ice selection 1020 or the execution block takeout confirmation 1030 .

In addition, when a user utterance is input in the state of the execution block hot ice selection 1020 , the next state may be the execution block takeout confirmation 1030 .

When the user intent is identified from the user utterance, the blocks may be searched by arranging them in the order of priority of intent conditions of blocks including the identified intent and/or a priority defined by a bot designer.

When a block matching the condition is found, it can move to the block, perform an action defined according to the block setting, and return to the user response standby state according to the condition of the action.

Connecting a block to a block may be defined as listing blocks to be processed next in the current block when entering the user response waiting state, and several user response waiting states may be defined in one block.

15 is a diagram illustrating an example of connecting blocks by defining a user response standby state in an execution block.

Referring to FIG. 15 , it can be seen that one block can be connected to a plurality of other blocks in a graph form.

Block 1 (Block 1) shown in FIG. 15 is connected to block 2 (Block 2), block 3 (Block 3), block 4 (Block 4), and block 5 (Block 5). At this time, block 1 (Block 1) is connected to block 2 (Block 2) and block 3 (Block 3) by condition 4 (Condition 4), and block 3 (Block 3) by condition 5 (Condition 5), It can be connected to block 4 (Block 4) and block 5 (Block 5). Accordingly, after block 1 (Block 1), the situations of block 2 (Block 2), block 3 (Block 3), block 4 (Block 4), and block 5 (Block 5) may occur.

Block 2 (Block 2) is connected to block 6 (Block 6). Accordingly, after block 2 (Block 2), the situation of block 6 (Block 6) may occur.

Block 3 (Block 3) is connected to block 6 (Block 6) and block 7 (Block 7). Accordingly, after block 3 (Block 3), the situation of block 6 (Block 6) and block 7 (Block 7) may occur.

Block 5 (Block 5) is connected to block 6 (Block 6) and block 7 (Block 7). Accordingly, after block 5 (Block 5), the situations of block 6 (Block 6) and block 7 (Block 7) may occur.

For example, if the current state is a state corresponding to Block 2, it is sufficient to search for an intent for only the intents corresponding to Block 6 for subsequent utterance by the user, and the possibility of occurrence is high. There is no need to perform user utterance matching on intents corresponding to the absent Block 7 .

When one intent and one execution block are mapped, one intent may have to include a plurality of utterance patterns.

For example, the intent (I1: signup intention) is the learning sentences "I want to try" (pattern classification P1), "How to write" (pattern classification P1), "I am curious about how to join" (pattern classification P3) and "Inquiry on how to join" (pattern classification P3), and the intent (I2: intention to withdraw) is the learning sentences "I want to stop writing" (pattern classification P2), "I don't want to use it anymore" (pattern classification P2), It can include "I want to know how to withdraw" (Pattern classification P3) and "Inquire about withdrawal method" (Pattern classification P3).

In this case, one block may be connected to the intent (I1: join intention) and one block may be connected to the intent (I2: leave intention), but the intents include a plurality of utterance patterns, so that the performance of the natural language processor is reduced. can be lowered

Accordingly, in order to prevent degradation of the performance of the natural language processor, it is desirable to avoid including a plurality of utterance patterns in one intent. For example, the intent (I1: intention to join) includes only the learning sentences "I want to try" (pattern classification P1) and "how to write" (pattern classification P1), and the intent (I2: intention to withdraw) is Include only the learning sentences "I want to stop writing" (pattern classification P2) and "I don't want to write anymore" (pattern classification P2), and the intent (I3: Inquiry) is the learning sentence "I am curious about how to <join>" ( Pattern classification P3), "Inquire about how to <join>" (Pattern classification P3), "I want to know how to <withdraw>" (Pattern classification P3), and "Inquire about <withdrawal> method" (Pattern classification P3) there is.

In the case of classifying intents so that they can be effectively processed by the natural language processor as described above, intents and processing blocks need to be connected in more various forms. To this end, each intent of the block may have a condition, and this condition may be defined as statements comparing slot values. That is, the block and intent mapping information may be defined in the block in the form of a list (intent mapping list) in which intents and slot conditions are paired.

16 is a diagram illustrating an example of mapping intents defined for each utterance pattern to a block.

Referring to FIG. 16 , it can be seen that three intents are connected to two blocks as a pair of intent and slot condition.

In this case, the slots used for the slot condition may include not only the slots defined in the current intent, but also slots of other blocks set while executing a previous conversation or slots provided by the system.

In the example shown in FIG. 16 , intent 1 corresponds to utterance pattern 1, intent 2 corresponds to utterance pattern 2, and intent 3 It can be seen that 3) corresponds to the firing pattern 3 (Pattern 3).

In the example shown in FIG. 16 , in the intent mapping list of block 1, a pair of intent 1 and condition 1 and intent 3 and condition 2 are included. ), there are two elements of the pair. Accordingly, Block 1 is connected to two intents Intent 1 and Intent 3 .

In the intent mapping list of block 2, two elements of a pair of intent 2 and condition 3 and a pair of intent 3 and condition 4 are exist. Accordingly, the block 2 (Block 2) is connected to the two intents (Intent 2, Intent 3).

17 is a diagram illustrating a case of 1:1 mapping of intents and execution blocks.

Referring to FIG. 17 , it can be seen that each of the three intents I1 , I2 , and I3 includes four utterance patterns.

That is, the intent I1 (join intent) includes the utterance patterns "I want to try", "How to write", "I want to know how to join", and "Inquire about how to join".

Intent I2 (withdrawal intention) includes utterance patterns "I want to stop using it", "I don't want to use it anymore", I want to know how to leave, and "I want to ask how to leave".

Intent I3 (intention to inquire about active members) includes the utterance patterns "I want to become a full member", "I want to be promoted to a full member", I want to know how to become a full member", and "How to become a full member".

If the intent (intent) and response (action block) are mapped 1:1, the natural language processor (NLU) is able to distinguish between these similar sentences when a sentence such as "*** I'm curious" appears in multiple intents. it's difficult. For example, a sentence such as "I am the first to ask how to join" may be processed by a natural language processor as a sentence similar to "I am asking how to unsubscribe" of the intention to leave, depending on the situation, resulting in incorrect intention recognition.

That is, the user's intention classification may be different from the intention classification that is easy to be processed by a natural language processor (NLU). The user's intentions may be classified as one intention that is easy to be processed by the natural language processor.

18 is a diagram illustrating an example of intent classification and block mapping according to an embodiment of the present invention.

Referring to FIG. 18 , it can be seen that sentences are classified in a form that is easy for the natural language processor (NLU) to process, and responses are mapped according to conditions by tagging necessary slots.

In the example shown in FIG. 18 , four intents I1 , I2 , I3 , I4 are connected to three responses R1 , R2 , R3 .

At this time, the intent 1 (I1) is connected to the subscription guidance response (R1). In this case, the slot condition may be a null condition.

At this time, Intent 4 (I4) is connected to the subscription guidance response (R1), the withdrawal guidance response (R2), and the regular member guidance response (R3). In this case, the intent 4 (I4) may be connected to the subscription guidance response (R1) by the slot condition $service == "subscription". In this case, the intent 4 (I4) may be connected to the withdrawal guidance response (R2) by the slot condition $service == "withdrawal". In this case, the intent 4 (I4) may be connected to the regular member guidance response R3 by the slot conditions $level == "regular member" and $service == "subscription".

At this time, the intent 2 (I2) is connected to the withdrawal guidance response (R2). In this case, the slot condition may be a null condition.

At this time, the intent 3 (I3) is connected to the regular member guidance response (R3). In this case, the slot condition may be a null condition.

Each of the responses shown in FIG. 18 may correspond to one execution block.

In this case, the execution block may be defined including intent and slot conditions and response processing of the chatbot.

19 is a diagram illustrating execution blocks corresponding to the responses shown in FIG. 18 .

Referring to FIG. 19 , each execution block includes an intent mapping list including a pair of at least one intent and a slot condition, and operation information.

19 , the slot condition may include a null condition corresponding to “no condition” and may include a conditional statement using one or more slots.

When the chatbot performs dialog processing, the natural language processor first classifies the intent from the sentence input by the user, extracts a slot, and then finds a matching block by examining the conditions listed in the block including the classified intent. .

In the example shown in FIGS. 18 and 19, when the intent is classified as "I4: Intent of inquiry", blocks including this intent are B1, B2, and B3, and blocks matched according to the condition are shown in the table below. equal to 1.

condition block Custom condition check order $service == "Subscribe" B1: Registration Information One $service == "Withdraw" B2: Notice of withdrawal 2 $level == "Full Member" and $service == "Sign Up" B3: Regular member information 3

The order of checking the condition can be determined by the user designating the order in which blocks are listed. If, as shown in Table 1, processing is simply performed in the order listed by the user, a sentence such as regular membership registration satisfies the conditions of $level == "regular member" and $service == "join", but at the same time logically $service = = "join" is also satisfied, and the B1 block having a high condition check rank may be executed, which may not be a desirable result.

In order to solve this problem, if priority is given again in consideration of the inclusion relationship of conditions, it can be determined as shown in Table 2 below.

condition block Custom condition check order Reordering ranking considering inclusion relationship $level == "Full Member" and $service == "Sign Up" B3: Regular member information 3 One $service == "Subscribe" B1: Registration Information One 2 $service == "Withdraw" B2: Notice of withdrawal 2 3

At this time, the inclusion relationship of the condition can be determined by the inclusion relationship of the set of slots used in the conditional expression, and the containing side (larger set) is treated with priority over the containing side (small set), and if there is no containment relation with each other, A user-specified priority may be maintained. In the example of Table 2, the condition of B3, which checks both level and service, has a higher condition inclusion relation order than B1 and B2, which checks only service, so that B3 can be processed preferentially over B1 or B2.

20 is a view showing an example of prioritization according to the inclusion relationship of the condition corresponding to the coffee order scenario illustrated in FIGS. 9 to 14 .

Referring to FIG. 20 , it can be seen that a condition corresponding to two coffee slots and two takeout slots has a higher priority than a condition corresponding to one coffee slot.

Therefore, when the user utters "I want to take out Americano", the slot coffee is filled with "Americano" and the slot takeout is filled with "Takeout", so the execution block coffee order is executed prior to the execution block takeout confirmation.

When the chatbot process is executed, intents may be collected from connected blocks in a current user response waiting state, and a suitable intent may be searched for among intents collected through a natural language processor.

When searching for a block by intent, a suitable block can be searched for by sorting in the order of the priority of intent conditions of the blocks including the intent and the priority defined by the user.

If a suitable block is not found because the condition is not met, the above method can be repeated using the next suitable intent.

When a block that meets the condition is found, it moves to the block, performs the action specified in the chatbot's response process, and enters the next user response waiting state according to the condition.

21 is a block diagram illustrating an example of modules required to execute a chatbot.

Referring to Figure 21, if the modules required for the chatbot to execute the resources defined by the bot designer are largely divided around the conversation flow processor, the part that controls the message input/output, the part that manages the block state, the part that processes natural language, and the block It can be seen as a part that processes conditions and a part that executes blocks.

In the example of FIG. 21 , the modules required to run the chatbot are a dialog flow processor, block state manager, message input gateway, message output gateway, response message generator, block process executor, NLU searcher, block searcher, intent classifier, slot extractor, natural language processor , an NLU priority manager, a condition interpreter and a block condition priority manager.

22 is a diagram illustrating a process of searching for an intent and a block while the chatbot is running.

Referring to FIG. 22 , when a user's utterance is input while the chatbot is waiting for user input, blocks executable at the current location are first searched. In this case, lower blocks connected to the current (execution) block may be searched first, and main blocks of the scenario may be searched. In this case, the main block refers to a list of blocks to be operated in the initial state of the conversation, and becomes the target blocks searched by the bot when the user first utters it. That is, if the intent is identified for only sub-blocks that are executable after the current block and the main blocks that are newly executable, efficient conversation flow processing is possible.

When a block is searched, the processing priority is checked by listing intents mapped to the searched block. In this case, intents defined in the searched blocks are listed, and a matching score of intents corresponding to the user's utterance is determined using a natural language processor. While sequentially searching the defined intent pattern threshold score list, comparing the names of intents and matching scores, the intent with the highest matching score among intents satisfying the pattern and threshold score is selected first. extract

If there is more than one block to which the matched intent belongs, processing priorities are sorted according to the defined intent mapping condition, and if the priorities of the conditions are the same, the priority of the scenario and block defined by the user is followed. In this case, the priority of the condition may be prioritized according to the inclusion relationship of the condition comparison phrase. That is, when a condition is configured based on a slot extracted from the user's utterance, the comparison syntax constituting the condition is mainly composed of the slots mentioned by the user and the unmentioned slots are omitted. It can be seen that this reflects the user's intention better. That is, when there are multiple conditions that are true in the conditional statement, the condition including more constraints or slots may be considered as having higher priority.

The sorted conditions are checked in order of priority, and the block containing the condition that is satisfied in the first order is treated as an execution block. If a block that satisfies the condition cannot be found, the block to which the intent belongs is found again with the intent of the next priority, and the process of determining the priority of the condition and checking the condition is repeated to find the block.

23 is an operation flowchart illustrating a method of providing a chatbot service according to another embodiment of the present invention.

Referring to FIG. 23 , in the chatbot service providing method according to another embodiment of the present invention, an intent corresponding to a user conversation is determined ( S2310 ).

In this case, an intent corresponding to the user conversation may be selected from among intent candidates connected to one or more candidate blocks or main blocks connected to the current block through a graph.

In this case, the intent corresponding to the user conversation may be determined by considering a first intent candidate connected to the candidate block in preference to a second intent candidate connected to the main block.

Also, in the chatbot service providing method according to another embodiment of the present invention, an execution block corresponding to the intent is selected using the intent mapping list (S2320).

In this case, the intent mapping list is provided for each of the execution blocks, and may correspond to at least one intent.

In this case, the intent mapping list may include an intent identifier and a slot condition for each intent.

In this case, the slot condition may correspond to a conditional statement related to one or more slots.

In this case, the slot condition may include a null condition.

In this case, the intent mapping list may include intent identifiers and slot conditions corresponding to a plurality of intents (N:1), and the intent indicates that different execution blocks have the same intent identifier. In this case, if the slot conditions are different, they may be connected (1:N) to different execution blocks.

In this case, the slots corresponding to the execution block may include slots of intents connected to the execution block through the intent mapping list.

At this time, when the intent corresponding to the user conversation is connected to a plurality of execution blocks, the execution block corresponding to the intent is the first execution block having the slot condition inclusion relationship corresponding to the large set. The second execution block having the inclusion relationship of the slot condition corresponding to the small set included in . In this case, the inclusion relationship of the slot condition may be the inclusion relationship of slots included in the slot condition. For example, an execution block corresponding to $level == "regular member" and $service == "subscription" may be considered and selected prior to an execution block corresponding to $service == "subscription".

In addition, the chatbot service providing method according to another embodiment of the present invention performs a conversation process corresponding to the intent based on the execution block (S2330).

With the above configuration, in the chatbot service providing method according to an embodiment of the present invention, N:N (N is a natural number) mapping between intents classified suitable for a natural language processor and execution blocks processing the intents. By making this possible, one function can be performed with several different utterances, and conversely, a completely different function can be performed with similar utterances.

Also, in the chatbot service providing method according to an embodiment of the present invention, it is possible to define various intent-execution block mapping relationships using an intent mapping list including an intent identifier and a slot condition for each intent.

In addition, the chatbot service providing method according to an embodiment of the present invention identifies candidate blocks that can be executed after the current execution block when identifying an intent corresponding to a user conversation, and uses only intents connected to the candidate blocks. It is possible to quickly and accurately identify an intent by using it as an identification target.

24 is a graph showing the performance of a natural language processor according to an increase in the number of speech patterns.

Referring to FIG. 24 , when an intent is divided by a function provided by the chatbot, when various speech sentences are added, a similar speech pattern frequently appears in several intents, and as shown in A, the intent of the NLU It can be seen that the classification performance deteriorates rapidly.

In order to solve this problem, by enabling the natural language processor to define an intent for each similar utterance pattern in a form suitable for processing, it is possible to maintain high performance of the natural language processor like B, even if the utterance patterns increase.

In addition, various utterance patterns can be efficiently processed by mapping different utterance patterns to units called 'blocks' so that the bot can properly process them according to conditions.

Blocks are connected in a graph form and multiple response waiting states can be defined depending on conditions, so that business processes with various processing flows or steps can be visually defined with a small number of blocks depending on conditions.

Implementations of the functional operations and subject matter described in this specification may be implemented as digital electronic circuits, implemented in computer software, firmware, or hardware including the structures disclosed herein and structural equivalents thereof, or a combination of one or more thereof. can be implemented Implementations of the subject matter described herein are one or more computer program products, ie one or more modules of computer program instructions encoded on a tangible program storage medium for controlling the operation of or for execution by a processing system. can be implemented.

The computer-readable medium may be a machine-readable storage device, a machine-readable storage substrate, a memory device, a composition of matter that affects a machine-readable radio wave signal, or a combination of one or more thereof.

As used herein, the term 'system' or 'device' encompasses all devices, devices and machines for processing data, including, for example, a programmable processor, a computer, or multiple processors or computers. A processing system may include, in addition to hardware, code that upon request forms an execution environment for a computer program, such as code constituting processor firmware, a protocol stack, a database management system, an operating system, or a combination of one or more thereof. .

A computer program (also known as a program, software, software application, script or code) may be written in any form of any programming language, including compiled or interpreted language or a priori or procedural language, and may be written as a stand-alone program or module; It can be deployed in any form, including components, subroutines, or other units suitable for use in a computer environment. A computer program does not necessarily correspond to a file in a file system. A program may be in a single file provided to the requested program, or in multiple interacting files (eg, files that store one or more modules, subprograms, or portions of code), or portions of files that hold other programs or data. (eg, one or more scripts stored within a markup language document). The computer program may be deployed to be executed on a single computer or multiple computers located at one site or distributed over a plurality of sites and interconnected by a communication network.

On the other hand, computer-readable media suitable for storing computer program instructions and data include, for example, semiconductor memory devices such as EPROMs, EEPROMs and flash memory devices, such as magnetic disks such as internal hard disks or external disks, magneto-optical disks and CDs. -Can include all types of non-volatile memory, media and memory devices, including ROM and DVD-ROM disks. The processor and memory may be supplemented by, or integrated into, special purpose logic circuitry.

An implementation of the subject matter described herein may include a backend component, such as a data server, or a middleware component, such as an application server, or a web browser or graphical user, such as a user capable of interacting with an implementation of the subject matter described herein. It may be implemented in a front-end component, such as a client computer having an interface, or in a computing system including any combination of one or more of such back-end, middleware, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication, such as, for example, a communication network.

While this specification contains numerous specific implementation details, they should not be construed as limitations on the scope of any invention or claim, but rather as descriptions of features that may be specific to particular embodiments of particular inventions. should be understood Likewise, certain features that are described herein in the context of separate embodiments may be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment may also be implemented in multiple embodiments, either individually or in any suitable subcombination. Further, although features operate in a particular combination and may be initially depicted as claimed as such, one or more features from a claimed combination may in some cases be excluded from the combination, the claimed combination being a sub-combination. or a variant of a sub-combination.

Also, although operations are depicted in the drawings in a specific order in this specification, it is not to be understood that such operations must be performed in the specific order or sequential order shown or that all illustrated operations must be performed in order to achieve desirable results. Can not be done. In certain cases, multitasking and parallel processing may be advantageous. Further, the separation of various system components of the above-described embodiments should not be construed as requiring such separation in all embodiments, and the program components and systems described may generally be integrated together into a single software product or packaged into multiple software products. It should be understood that

As such, this specification is not intended to limit the invention to the specific terminology presented. Accordingly, although the present invention has been described in detail with reference to the above-described examples, those skilled in the art can make modifications, changes, and modifications to the examples without departing from the scope of the present invention. The scope of the present invention is indicated by the following claims rather than the above detailed description, and all changes or modifications derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present invention. do.

As described above, in the method and apparatus for providing a slot filling-based chatbot service according to the present invention, the configuration and method of the embodiments described above are not limitedly applicable, but the embodiments are provided so that various modifications can be made. All or part of each embodiment may be selectively combined and configured.

According to the present invention, a method and apparatus for providing a slot filling-based chatbot service are disclosed. A chatbot service providing method according to an embodiment of the present invention determines an intent corresponding to a user conversation, selects an action block corresponding to the intent using an intent mapping list, and fills a slot corresponding to the action block , a dialog process corresponding to an intent is performed based on slot filling, and an intent mapping list is provided for each of the execution blocks and corresponds to at least one intent. Accordingly, the chatbot can effectively respond to various user conversation patterns, and it is possible to provide an efficient chatbot service.

610: communication unit 620: processor
630: memory 710: chatbot
720: natural language processor 730: chatbot resource DB
740: Chatbot Builder

Claims (10)

determining an intent corresponding to the user conversation;
selecting an execution block corresponding to the intent by using an intent mapping list;
performing slot filling corresponding to the execution block;
Comprising the step of performing dialog processing corresponding to the intent based on the slot filling,
The intent mapping list is provided for each of the execution blocks, and the slot filling-based chatbot service providing method, characterized in that it corresponds to at least one intent. The method according to claim 1,
The step of performing the slot filling is
A method for providing a slot filling-based chatbot service, characterized in that the slot filling is performed based on at least one of an entity mapped to a slot, slots corresponding to the execution block, and intent mapping information of the slot. 3. The method according to claim 2,
The intent mapping information is
characterized by including at least one of an intent name mapped to a slot, a slot name of an intent mapped to a slot, a slot value of an intent mapped to a slot, and a result value based on a slot value of an intent mapped to a slot A method of providing a slot filling-based chatbot service. 3. The method according to claim 2,
Slots corresponding to the execution block are
Slot filling-based chatbot service providing method comprising slots of intents connected to the execution block through the intent mapping list. 3. The method according to claim 2,
The step of performing the slot filling is
A method for providing a slot filling-based chatbot service, characterized in that the slot value is extracted by performing synonym matching on the entity mapped to the slot. 3. The method according to claim 2,
A slot filling-based chatbot service providing method, characterized in that one slot can be mapped with a plurality of entities. The method according to claim 1,
Slot filling-based chatbot service providing method, characterized in that it further comprises the step of performing the slot reuse setting to maintain the slot context in consideration of the scenario in which the execution block is included. 8. The method of claim 7,
Slot filling-based chatbot service providing method, characterized in that the slot values of the slots of the intents connected to the execution block for which the slot reuse setting is performed are not deleted even when the scenario is re-executed after the end of the scenario. The method according to claim 1,
The intent mapping list is
A method for providing a slot filling-based chatbot service, comprising: an intent identifier and a slot condition for each intent. one or more processors; and
an execution memory for storing at least one or more programs executed by the one or more processors; including,
The at least one or more programs,
An intent corresponding to a user conversation is determined, an action block corresponding to the intent is selected using an intent mapping list, a slot filling corresponding to the action block is performed, and the input is based on the slot filling. Conversation processing corresponding to the tent is performed,
The intent mapping list is provided for each of the execution blocks, and the slot filling-based chatbot service providing apparatus, characterized in that it corresponds to at least one intent.

Images