KR20190116044A - Method and computer readable recording medium for actively analyzing a online chatting session between users and providing analysis results visually to a user, by a conversation understanding ai system - Google Patents

Abstract

Provided is a method performed by a computer, which controls a predetermined display related to a second user terminal by intervening in an online text chatting session established between a first user terminal and the second user terminal which are remotely arranged. According to the present invention, the method comprises the steps of: allowing a display, during a text chatting session, to display a first text input which is newly input by a first user on a first user terminal with previous text inputs which are exchanged between the first user terminal and a second user terminal on the text chatting session; displaying a text expression of a user intent of the first user, which is determined by performing natural language processing on the first text input in accordance with a predetermined knowledge base model; displaying an analysis result obtained as a result of analyzing the user intent of the first text input with a user intent corresponding to at least one of the previous text inputs on the basis of the predetermined knowledge base model; and displaying a text expression of at least one chatting response candidate which can be provided by the second user terminal in response to the first text input determined based on at least a part of the user intent and an analysis result.

Description

TECHNICAL AND COMPUTER READABLE RECORDING MEDIUM FOR ACTIVELY ANALYZING A ONLINE CHATTING SESSION BETWEEN USERS AND PROVIDING ANALYSIS RESULTS VISUALLY TO A USER, BY A CONVERSATION UNDERSTANDING AI SYSTEM}

The present disclosure relates to active monitoring and analysis of online chat sessions between users by a conversation understanding AI system, and more particularly, to actively monitoring and analyzing online chat sessions between users and visualizing the results. It relates to how to present.

In recent years, according to technological developments in the field of artificial intelligence, in particular, in the field of understanding natural language, development of a dialogue understanding AI system capable of receiving and understanding natural language input in the form of voice and / or text by a user and performing necessary actions accordingly, Usage is increasing gradually. On the other hand, in recent years, each company or institution operates a customer counseling center, and a dialogue understanding AI system is being built for such a customer counseling center.

Patent Document 1: Korean Patent Application No. 10-2016-7032134

Understanding conversations An AI system can be used as a way to facilitate conversations by intervening in conversations between human users (for example, between customers and counselors), although they may be the ones who lead the conversations themselves. . There is a need for a dialogue understanding AI system that can monitor and analyze the progress of conversations between human users, especially on online chats, and thus provide information for the parties to help in the direction of the conversation.

According to an aspect of the present disclosure, by a computer, intervening an online text chat session established between a first user terminal and a second user terminal, remotely disposed with each other, to control a predetermined display associated with the second user terminal. Provided is a method performed. The method of the present disclosure is performed by a computer, intervening an online text chat session established between a first user terminal and a second user terminal, remotely disposed with each other, to control a predetermined display associated with the second user terminal. The method may further comprise, during a text chat session, causing the display to display a newly input by the first user on the first user terminal, with previous text inputs exchanged between the first user terminal and the second user terminal on the text chat session. 1 displaying a text input; Causing the first text input to display a textual representation of a user intent of the first user, determined by natural language processing according to a predetermined knowledge base model; Displaying, together with a corresponding user intent of at least one of the previous text inputs, an analysis result obtained as a result of analyzing the user intent of the first text input based on a predetermined knowledge base model; And displaying a textual representation of one or more conversation response candidates that may be provided by the second user terminal as a response to the first text input, determined based on at least a portion of the user intent and analysis result. .

According to one embodiment of the present disclosure, a predetermined knowledge base model includes at least an ontology model and a user database for understanding conversations, wherein the user database includes a history of past conversations, a location, a language used, and vocabulary preferences associated with the first user. , Age, gender, or black consumer.

According to an embodiment of the present disclosure, the analysis result may include profile information of the first user, wherein the profile information includes at least one of gender, age group, and language of the first user.

According to an embodiment of the present disclosure, the analysis result includes emotion information indicating an emotional state of the first user, the emotion information includes value information allocated for each of the plurality of emotion types, and the plurality of emotion types May include at least one emotion type of satisfaction, calm, dissatisfaction, and anger.

According to one embodiment of the present disclosure, the computer is a conversation understanding AI server for a customer consultation center, which may be associated with a predetermined goal.

According to an embodiment of the present disclosure, the method may further include obtaining a sequential flow of intents including user intents corresponding to each of the previous text inputs and a user intent of the first text input; Sequential Flow of Intents and Conversation Understanding Conversation Template Prepared for AI Service Server-Conversation template includes a plurality of conversation patterns, each conversation pattern representing the flow of each corresponding intent, each conversation pattern to target Determining one or more conversation patterns among the conversation patterns corresponding to the sequential flow of intents, based on the number of goal achievement successes and goal achievement failures in relation to each other; Determining a prediction value of a target achievement probability based on each corresponding target achievement success count and target achievement failure count of the determined one or more conversation patterns; And causing the display to display a prediction of the target achievement probability.

According to an embodiment of the present disclosure, the predicted value of the target achievement probability may correspond to a ratio of the total sum of the target achievement success sums to the sum of the corresponding target achievement success counts and the target achievement failure counts of the determined one or more conversation patterns. have.

According to an embodiment of the present disclosure, the dialogue template may have a target achievement probability determined for each of the plurality of dialogue patterns based on the target achievement success count and the target achievement failure count of each conversation pattern.

According to one embodiment of the present disclosure, a goal may be associated with at least one of selling a certain product and / or service, signing up, and subscribing.

According to one embodiment of the present disclosure, the step of displaying each textual representation of one or more conversation response candidates is based on, for each of the determined one or more conversation patterns, based on the number of goal achievement successes and the number of goal achievement failures of each conversation pattern. Determining the determined probability of achieving each target; Determining a conversation pattern having the highest target achievement probability among the determined one or more conversation patterns as an optimal conversation pattern; And based on the determined optimal conversation pattern, generating each textual representation of one or more conversation response candidates.

According to another feature of the present disclosure, a computer readable recording medium having one or more instructions, wherein the one or more instructions, when executed, cause the computer device to perform any one of the methods described above. A possible recording medium is provided.

The method according to the embodiment of the present disclosure can assist the inexperienced counseling staff in order to provide a response direction of the counseling staff for each situation, when used in a chat consultation between the counseling staff of the customer consultation center and the customer. In addition to ensuring that customer consultations are conducted, this can have the effect of reducing the emotional labor fatigue of the counseling staff. The method according to the embodiment of the present disclosure, when used in the chat consultation between the counseling staff of the customer consultation center and the customer, the probability of achieving the target for each situation (for example, the If so, the customer may provide an estimate of the probability that a certain target task corresponding to each customer center will be achieved, such as a purchase of a corresponding product or service, for example, a purchase conversion rate, and the like. By providing the best response direction for counseling staff, you can ultimately improve the probability of meeting your target.

1 is a diagram schematically illustrating a system environment in which a conversation understanding AI system may be implemented, according to an embodiment of the present disclosure.
2 is a functional block diagram schematically illustrating the functional configuration of the conversation understanding AI service server 106 of FIG. 1, according to an embodiment of the disclosure.
3 is a functional block diagram schematically illustrating a functional configuration of the conversation understanding unit 204 of FIG. 2, according to an embodiment of the present disclosure.
4 is a functional block diagram schematically illustrating the functional configuration of the responding user terminals 108a-108m of FIG. 1, in accordance with an embodiment of the present disclosure.
FIG. 5 is a diagram illustrating an example of a screen configuration that may be presented on the screen display units of the response user terminals 108a-108m of FIG. 1, according to an embodiment of the present disclosure.
6 is a diagram conceptually illustrating an example of a template 600 for the dialog understanding AI service server 106.

Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. In the following, when it is determined that there is a risk of unnecessarily obscuring the subject matter of the present disclosure, a detailed description of already known functions and configurations will be omitted. In addition, it should be understood that what is described below is only related to an embodiment of the present disclosure, but the present disclosure is not limited thereto.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the present disclosure. For example, a component expressed in the singular should be understood as a concept including a plurality of components unless the context clearly indicates the singular. It is to be understood that the term "and / or" as used in this disclosure encompasses all possible combinations by one or more of the listed items. The terms 'comprise' or 'having' as used in the present disclosure are merely intended to designate that there exists a feature, number, step, operation, component, part, or combination thereof described on the present disclosure. It is not intended to exclude the possibility of the presence or the addition of one or more other features or numbers, steps, actions, components, parts or combinations thereof by use.

In the exemplary embodiment of the present disclosure, the module or the unit means a functional part that performs at least one function or operation, and may be implemented in hardware or software or in a combination of hardware and software. Also, a plurality of 'modules' or 'units' may be integrated by at least one software module and implemented by at least one processor, except for 'modules' or 'units', which need to be implemented by specific hardware. have.

In an embodiment of the present disclosure, the 'conversation understanding AI system' receives and interprets a natural language input (for example, a command, statement, request, question, etc. from a user in natural language) and an intent of the user. ) May refer to any information processing system capable of identifying and performing appropriate actions based on the user's intent, but is not limited to any particular form.

In addition, all terms used in this disclosure, including technical or scientific terms, unless otherwise defined, have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs. It is to be understood that the terms used in the dictionary, which are generally used, should be interpreted as having a meaning consistent with the contextual meaning of the related art, and shall not be construed as being excessively limited or extended unless clearly defined otherwise in the present disclosure. .

Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings.

1 is a diagram schematically illustrating a system environment 100 in which a conversation understanding AI system may be implemented, in accordance with an embodiment of the present disclosure. As shown, the system environment 100 includes a plurality of customer user terminals 102a-102n, a communication network 104, a conversation understanding AI service server 106, and a plurality of responding user terminals 108a-108m. .

According to one embodiment of the present disclosure, each of the plurality of customer user terminals 102a-102n may be any user electronic device having an online chat function by wired or wireless communication. It should be appreciated that each of the customer user terminals 102a-102n can be a variety of wired or wireless communication terminals, including, for example, smart phones, tablet PCs, desktops, laptops, PDAs, digital TVs, and the like, and is not limited to any particular form. According to one embodiment of the present disclosure, each of the customer user terminals 102a-102n may have a predetermined display device.

According to one embodiment of the present disclosure, each of the customer user terminals 102a-102n can communicate with the conversation understanding AI service server 106, that is, send and receive necessary information via the communication network 104. According to one embodiment of the present disclosure, each of the customer user terminals 102a-102n may receive a customer user input from outside, for example, in the form of a text. According to one embodiment of the present disclosure, each of the customer user terminals 102a-102n has an operation result corresponding to the above customer user input obtained through communication with the conversation understanding AI service server 106 through the communication network 104 ( For example, providing a specific conversation response and / or performing a specific task, etc.) to a corresponding customer user. According to one embodiment of the present disclosure, the conversation understanding AI service server 106 may be, for example, a customer consultation center server, and the customer user terminals 102a-102n may communicate with, for example, a customer's communication terminal. Can be.

According to one embodiment of the present disclosure, each of the customer user terminals 102a-102n communicates with one of the responding user terminals 108a-108m via the communication network 104 and the conversation understanding AI service server 106. Establish an online chat session. According to one embodiment of the present disclosure, through an online chat session established through communication network 104 and conversation understanding AI service server 106, each of the customer user terminals 102a-102n is responding user terminal 108a-108m. Chat with one of them, for example, a natural language conversation by text.

According to one embodiment of the present disclosure, communication network 104 may include any wired or wireless communication network, such as a TCP / IP communication network. According to one embodiment of the present disclosure, the communication network 104 may include, for example, a Wi-Fi network, a LAN network, a WAN network, an Internet network, and the like, but the present disclosure is not limited thereto. According to one embodiment of the present disclosure, the communication network 104 may include, for example, Ethernet, GSM, Enhanced Data GSM Environment (EDGE), CDMA, TDMA, OFDM, Bluetooth, VoIP, Wi-MAX, Wibro, and any other various wired or wireless. It may be implemented using a communication protocol.

According to one embodiment of the present disclosure, the conversation understanding AI service server 106 may communicate with the customer user terminals 102a-102n and the answering user terminals 108a-108m in accordance with a predetermined wired or wireless communication scheme. have. According to one embodiment of the present disclosure, the conversation understanding AI service server 106 receives various inputs from the customer user terminals 102a-102n, and the customer user terminals 102a-102n and the response user terminals (described later) ( 108a-108m) may establish an online chat session. According to one embodiment of the present disclosure, the conversation understanding AI service server 106 intervenes with each established user input from the customer user terminals 102a-102n and the responding user terminals 108a-108m, via an established chat session. For example, text input).

In this figure, an online chat session is shown between the customer user terminals 102a-102n and the response user terminals 108a-108m by the conversation understanding AI service server 106, although the present disclosure is limited thereto. It is not. According to another embodiment of the present disclosure, a chat session between customer user terminals 102a-102n and responding user terminals 108a-108m may be established by a separate chat service server (not shown). In this case, the conversation understanding AI service server 106 may intervene in the established chat session to perform monitoring and analysis of user input.

According to one embodiment of the present disclosure, the dialogue understanding AI service server 106 analyzes each user input received from the customer user terminals 102a-102n on the established chat session based on a previously prepared knowledgebase model or the like. And process to determine the corresponding user intent. According to one embodiment of the present disclosure, the conversation understanding AI service server 106 may include a user intent corresponding to a current user input, and a previous conversation history on the chat session (that is, a customer user terminal on the chat session). 102a-102n) and the corresponding flow of intents according to previous conversation histories between corresponding user terminals 108a-108m), and based on previously prepared knowledgebase models, the response user terminals on the corresponding chat session ( 108a-108m may determine an indicator (eg, a probabilistic indicator) indicating whether it is time to give an answer to the customer user terminals 102a-102n.

According to one embodiment of the present disclosure, the conversation understanding AI service server 106 is characterized by characteristic data about a customer user of each customer user terminal 102a-102n, such as a previous conversation (consultation) record for each user, the location of the user, Language of use, user's vocabulary preferences, age group, gender, black consumer status, and / or various other user characteristic information. According to an embodiment of the present disclosure, the characteristic data about the customer user described above may be, for example, information input by the customer user or information collected from the behavior of the customer user. According to one embodiment of the present disclosure, the conversation understanding AI service server 106 refers to the user characteristic data described above and based on previously prepared knowledgebase models, user input (and previous conversation) received on the above chat session. History) to generate analytical results regarding information about the customer's user, such as, but not limited to, user profile and / or current emotional state (e.g., but not limited to satisfaction, calmness, dissatisfaction, anger, etc.). Can be.

According to one embodiment of the present disclosure, the conversation understanding AI service server 106, as a service server of a customer service center, may be associated with a predetermined goal. For example, according to one embodiment of the present disclosure, the conversation understanding AI service server 106 is a server of a customer consultation center for selling a predetermined product and / or service, and each of the conversation understanding AI service server 106 has accessed. A goal may be achieved when the customer user terminals 102a-102n purchase corresponding products and / or services. According to another embodiment of the present disclosure, the conversation understanding AI service server 106 is a server of a customer consultation center for inducing subscription or subscription, and each customer user terminal that has accessed the conversation understanding AI service server 106. If 102a-102n applies for membership or subscription, the goal may be achieved. According to another embodiment of the present disclosure, the dialogue understanding AI service server 106 may be regarded as a case where the goal has been achieved by simply acquiring the intention of satisfaction of consultation from a customer as a server of a customer consultation center. Is not limited thereto.

According to the embodiment of the present disclosure, the dialogue understanding AI service server 106 is a server of a customer consultation center, which is between a customer and an agent (here between the customer user terminals 102a-102n and the response user terminals 108a-108m). Almost any type of conversational pattern that can occur in a conversation, where the conversational pattern includes a predetermined pattern represented by a sequential flow of intents corresponding to natural language inputs that are sequentially sent and received on any chat session. It can include a conversation template. According to an embodiment of the present disclosure, since there may be a plurality of natural language inputs corresponding to one same intent, a conversation composed of various different sentences may actually correspond to the same conversation pattern on a conversation template. Should know. For more details about the dialogue template, refer to FIG. 6 and the like.

According to an embodiment of the present disclosure, for each of the above-described conversation patterns, the conversation template may include a target of the corresponding customer consultation center, that is, the customer consultation center may have a relationship between the customer (ie, the customer user terminals 102a-102n) and the counseling staff. Goals to be achieved by conversation (eg, sales of certain products and / or services, subscriptions, subscriptions, and customer satisfaction), but may not be limited to this, but have specific purposes and indicate whether or not they have been achieved The probability of achievement of the various goals that can be identified as a target (ie, the number of successful goal achievements, the number of failed goal achievements, and the success rate determined by the number of successful goal achievements and the number of failed goal achievements, such as (target) The number of successful achievements) / (probability of achievement of target achievement + target failures)), and the like. Regarding the creation of such a conversation template and the like, it is beyond the scope of the present disclosure and detailed description thereof will be omitted. For further details, see the other patent applications Nos. 10-2018-0006435 and 10-2018-0009678, etc. of the applicant.

According to one embodiment of the present disclosure, the conversation understanding AI service server 106 is configured at each point in time during a conversation on the online chat session between the customer user terminals 102a-102n and the response user terminals 108a-108m (eg, At the point where a specific input has occurred in an ongoing conversation, where the conversation has not yet been concluded since the conversation has started, and whether or not the goal has been determined yet, the corresponding customer user terminals 102a-102n and the responding user terminal (up to that point) Identify a sequential flow of a series of intents (eg, intents corresponding to inputs / responses) between 108a-108m) (e.g., multi-sentence analysis) and, based on the dialogue template described above, Identify progress patterns that may develop in the future, that is, possible conversational patterns in the future. According to one embodiment of the present disclosure, the dialogue understanding AI service server 106 is also based on the number of successes and failures of achieving the goal for all of the future possible conversation patterns, which are identified based on the dialogue template described above. Thus, the predicted value of the target achievement probability at that time can be determined.

According to one embodiment of the present disclosure, at each point in time during which an online chat session is established between customer user terminals 102a-102n and responding user terminals 108a-108m and a conversation (consultation) is taking place on the chat session, Understand that the AI service server 106, based on the dialogue template described above, is based on the dialogue template described above, to determine the appropriate dialogue pattern after that point in time (e.g., among the conversation patterns that may occur in the future with reference to the conversation history so far). Optimal conversation patterns, etc., which increase the probability of achieving the target. According to one embodiment of the present disclosure, the conversation understanding AI service server 106 may also, at each point in the conversation on the chat session, an appropriate conversation pattern (and the like) that can increase the probability of achieving the goal based on a conversation template prepared in advance. Along with the response direction (intent) on the determined conversation pattern and the above-described analysis result, for example, user profile and / or emotional state, and the predetermined knowledge prepared in advance. Based on the base models, one or more conversation response candidates (suggestions) may be generated, at which point responding user terminals 108a-108m may refer to for responses to customer user terminals 102a-102n. .

According to one embodiment of the present disclosure, the conversation understanding AI service server 106 is configured to analyze user input from customer user terminals 102a-102n received through a chat session by analyzing and processing the received user input. Obtained results, for example, corresponding user intents, indicators indicating whether it is time to provide answers, analysis result information such as one or more conversation response candidates corresponding to the user intents, user profile characteristics, emotional characteristics, predictions of goal achievement probability, and the like. Together, it may transmit via the communication network 104 to the corresponding responding user terminals 108a-108m. According to one embodiment of the present disclosure, the dialogue understanding AI service server 106 may be various customer center servers in various fields such as finance, medical, law, shopping, etc., but the present disclosure is not limited thereto.

According to one embodiment of the present disclosure, each of the responding user terminals 108a-108m may establish a chat session with the aforementioned customer user terminals 102a-102n via the communication network 104. According to one embodiment of the present disclosure, each of the responding user terminals 108a-108m receives input from the customer user terminals 102a-102n received via the conversation understanding AI service server 106 on the established chat session. And results of analysis and processing of the user input, such as a user intent, an indicator indicating whether it is time to provide an answer, one or more conversation response candidates corresponding to the user intent, a user profile characteristic, an emotional characteristic, a prediction of a target achievement probability, and the like. Results of analysis and the like can be received. According to an embodiment of the present disclosure, the response user terminals 108a-108m may receive the received user input (text input), various analysis results, and the like through the screen output unit (display). May be presented to the responding user (eg, a counselor). According to one embodiment of the present disclosure, the responding user terminals 108a-108m also receive input from the responding user on that terminal and communicate this via the corresponding chat session. The AI service server 106 and the customer user terminal 102a- 102n).

2 is a functional block diagram schematically illustrating the functional configuration of the conversation understanding AI service server 106 of FIG. 1, according to an embodiment of the disclosure. As shown, the conversation understanding AI service server 106 includes a chat session establishment / management unit 202, a conversation understanding unit 204, and a communication unit 206.

According to one embodiment of the present disclosure, the chat session establishment / management unit 202 establishes and establishes a chat session between one of the customer user terminals 102a-102n and one of the response user terminals 108a-108m. It is possible to operate the conversation between the two via a chat session. According to one embodiment of the present disclosure, the chat session establishment / management unit 202 receives a user text input or the like coming from the customer user terminals 102a-102n according to a predetermined communication protocol and responds to the corresponding response user terminal 108a. -108m), and also inputs from the responding users, etc. that have been sent from one of the responding user terminals 108a-108m, to the corresponding customer user terminals 102a-102n.

According to an embodiment of the present disclosure, the conversation understanding unit 204 receives a user input received from the customer user terminals 102a-102n through the chat session establishment / management unit 202, and prepares a knowledge base model prepared in advance. Based on these, it is possible to determine a user intent corresponding to the received user input. According to an embodiment of the present disclosure, the conversation understanding unit 204 may generate a series of intents generated in the customer user terminals 102a-102n and the response user terminals 108a-108m through the corresponding chat session up to a specific time point. The pattern of the sequential flow of these fields can be grasped, and based on a dialogue template prepared in advance, a predicted value of a target achievement probability, etc. can be determined. According to an embodiment of the present disclosure, the conversation understanding unit 204 also analyzes the received user input based on user characteristic data regarding each of the user terminals 102a-102n prepared in advance, and relates to the corresponding user. Information such as user profile information and emotion information. According to an embodiment of the present disclosure, the dialogue understanding unit 204 may also determine a user intent determined in advance based on previously prepared knowledge base models (and / or dialogue templates), information about the analyzed user, and the like. One or more matching conversation response candidates may be generated.

According to one embodiment of the present disclosure, the communication unit 206 enables the conversation understanding AI service server 106 to communicate with each of the responding user terminals 108a-108m via the communication network 106 of FIG. 1. . According to an embodiment of the present disclosure, the communication unit 206 may include a user input signal received from the customer user terminals 102a-102n through the chat session establishment / management unit 202, and a conversation understanding unit from the user input signal. The user intent generated at 204, one or more conversation response candidates corresponding to the user intent, and various analysis results may be transmitted to one of the response user terminals 108a-108m according to a predetermined protocol.

3 is a functional block diagram schematically illustrating a functional configuration of the conversation understanding unit 204 of FIG. 2, according to an embodiment of the present disclosure. As shown, the conversation understanding unit 204 includes a natural language understanding (NLU) module 302, a conversation understanding knowledge base 304, a user database 306, a conversation template 308, and a user profile analysis. Module 310, emotion analysis module 312, conversation management module 314, and conversation generation module 316.

According to one embodiment of the present disclosure, the NLU module 302 receives text input from the customer user terminals 102a-102n through the communication network 104 of FIG. 1 and the chat session establishment / management unit 202 of FIG. 2. Can be received. In accordance with one embodiment of the present disclosure, the NLU module 302, based on a knowledge base model, e.g., conversation understanding knowledge base, etc. 304 described below, assigns the received text input to one or more user intents. Can match. The user intent here can be associated with a series of action (s) that can be understood and performed by the conversation understanding AI service server 106 in accordance with the user intent.

According to one embodiment of the present disclosure, conversation understanding knowledge base 304 may include, for example, a predefined ontology model. According to one embodiment of the present disclosure, the ontology model may be represented, for example, in a hierarchical structure between nodes, each node being a "intent" node or a child linked to an "intent" node corresponding to the user's intent. It may be one of the "property" nodes (child "nodes" nodes that are linked directly to "intent" nodes or relinked to "property" nodes of "intent" nodes). According to one embodiment of the present disclosure, an "intent" node and "attribute" nodes that are directly or indirectly linked to the "intent" node may constitute one domain, and the ontology may consist of a collection of such domains. Can be. According to one embodiment of the present disclosure, conversation understanding knowledge base 304 includes domains corresponding to all intents, for example, that conversation understanding AI service server 106 may understand and perform operations corresponding thereto. It can be configured to. According to one embodiment of the present disclosure, it should be understood that the ontology model may be dynamically changed by adding or deleting nodes, or modifying relationships between nodes.

According to one embodiment of the present disclosure, intent nodes and attribute nodes of each domain in the ontology model may be associated with words and / or phrases related to corresponding user intents or attributes, respectively. According to one embodiment of the present disclosure, the dialogue understanding knowledge base 304 comprises an ontology model consisting of a hierarchy of nodes and a set of words and / or phrases associated with each node, for example in the form of a lexical dictionary (specific (Not shown), the NLU module 302 may determine the user intent based on the ontology model implemented in the form of a lexical dictionary. For example, according to one embodiment of the present disclosure, when the NLU module 302 receives a text input or a sequence of words, the NLU module 302 may determine which words in which domain in the ontology model are associated with each word in the sequence, Based on such a determination, the corresponding domain, ie user intent, can be determined.

According to an embodiment of the present disclosure, the user database 306 may be a database that stores and manages characteristic data for each user. According to an embodiment of the present disclosure, the user database 306 is, for example, whether the user's previous conversation (counseling) record, the user's location, the language used, the user's vocabulary preferences, age group, gender, or black consumer for each user. And / or various other user characteristic information. According to one embodiment of the present disclosure, the NLU module 302 determines more accurate user intent by referring to each user characteristic data of the user database 306, such as a feature or context for each user, when determining the user intent. can do.

In this figure, a user database 306 for storing and managing characteristic data for each user is illustrated as being disposed in the conversation understanding AI service server 106, but the present disclosure is not limited thereto. According to another embodiment of the present disclosure, a user database for storing and managing characteristic data for each user may exist, for example, in the customer user terminals 102a-102n, and the customer user terminals 102a-102n and the conversation understanding AI service. It should be appreciated that it may be distributed across the servers 106.

According to one embodiment of the present disclosure, the conversation template 308 may occur between the customer user terminals 102a-102n and the responding user terminals 108a-108m on a chat session via the conversation understanding AI service server 106. It can include almost any type of conversational pattern (sequential flow of intents). According to an embodiment of the present disclosure, the conversation template may include a probability of achieving a goal (eg, a goal of a customer service center) associated with the corresponding conversation understanding AI service server 106 for each of the aforementioned conversation patterns.

According to an embodiment of the present disclosure, the user profile analysis module 310 may obtain or estimate profile information about the customer user of the customer user terminals 102a-102n. According to an embodiment of the present disclosure, the user profile analysis module 310 may obtain profile information about the corresponding user from the user characteristic data on the user database 306 described above. According to one embodiment of the present disclosure, the user profile analysis module 310 receives input from the customer user terminals 102a-102n received through the chat session establishment / management unit 202, and processes and analyzes the input. As a result, for example, one or more intents acquired by the NLU module 302, and / or user characteristic data on the user database 306 described above may be comprehensively analyzed to obtain or estimate information about the user profile. have. Although not specifically illustrated, according to an embodiment of the present disclosure, the user profile analysis module 310 may include a predetermined knowledge base model for user profile analysis.

According to one embodiment of the present disclosure, the emotion analysis module 312 analyzes the input from the customer user terminals 102a-102n received through the chat session establishment / management unit 202 of FIG. Information about emotions can be estimated. According to an embodiment of the present disclosure, the emotion analysis module 312 may include an input from the customer user terminals 102a-102n received through the chat session establishment / management unit 202, processing and analysis results regarding the input, For example, one or more intents obtained by the NLU module 302 described above and / or user profile information obtained by the user profile analysis module 310 (eg, user's gender, age group, language used, etc.), the corresponding chat session. It is possible to estimate the information on the current emotion of the user by comprehensively analyzing the previous conversation history on the screen. According to one embodiment of the present disclosure, the emotion analysis module 312 classifies a person's emotion type into a predetermined number of types (eg, satisfaction, calmness, dissatisfaction, anger, etc.), and performs a comprehensive analysis of the above information. Accordingly, a value may be assigned to each emotion type to indicate the current emotional state of the user through the assigned value. Although not specifically illustrated, according to an embodiment of the present disclosure, the emotion analysis module 310 may include a predetermined knowledge base model for analyzing the emotional state of the user.

  According to one embodiment of the present disclosure, the conversation management module 314 is based on a user intent determined by the NLU module 302, and includes a predetermined knowledge base model prepared for conversation management and / or the aforementioned conversation template. According to 308, a series of conversation flows corresponding to the user intent determined above may be generated. According to one embodiment of the present disclosure, the conversation management module 314 determines which conversation response should be made in response to a user intent received from the NLU module 302, for example, based on a predetermined conversation management knowledge base model. Can be determined and a detailed conversation / action flow generated accordingly. According to one embodiment of the present disclosure, the conversation management module 314 may, for example, receive a user intent received from the NLU module 302 and a series of intents (inputs / responses) generated so far on the current conversation session. Knowing the sequential flow of the corresponding intents, and based on the dialogue template 308, it is possible to determine which dialogue response to make in this step, and generate a detailed dialogue / action flow accordingly. According to an embodiment of the present disclosure, the conversation management module 314 may obtain conversation patterns that may occur in the future from the conversation template 308, and determine a prediction value of a target achievement probability at that time. According to one embodiment of the present disclosure, the conversation management module 314 obtains future possible conversation patterns from the conversation template 308, and appropriate conversation patterns (eg, after that point in time) to improve the probability of achieving the associated goal. With reference to the conversation history up to now, it is possible to generate an optimal conversation pattern, etc., which increases the probability of achieving a goal among conversation patterns that can occur in the future, and a conversation flow according thereto.

According to one embodiment of the present disclosure, the dialogue generation module 316 may generate one or more candidates suitable as a dialogue response to be provided to the user based on the dialogue flow generated by the dialogue management module 314. According to an embodiment of the present disclosure, the dialogue generation module 316 may be configured in the current dialogue in consideration of a value (ie, current emotion state information of the user) assigned to each emotion type by the emotion analysis module 312 described above. One or more conversation response candidates deemed appropriate may be generated. According to one embodiment of the present disclosure, the conversation creation module 316 is not only information about the emotional state of the user, but also one or more intents obtained by the NLU module 302 described above, and / or user profile analysis. With reference to user profile information (eg, user's gender, age group, etc.) obtained by module 310, one or more suitable candidates may be generated as a dialogue response to be provided to the user. Although not specifically illustrated, according to an embodiment of the present disclosure, the conversation generating module 316 may include a predetermined knowledge base model (eg, a predetermined response sentence model, etc.) for generating a specific conversation response candidate to be provided to the user. ) May be included.

4 is a functional block diagram schematically illustrating the functional configuration of the responding user terminals 108a-108m of FIG. 1, in accordance with an embodiment of the present disclosure. As shown, the response user terminals 108a-108m include a communication unit 402, a response user input receiver 404, and an information visualization / screen output unit 406.

According to one embodiment of the present disclosure, the communication unit 402 enables the responding user terminals 108a-108m to communicate with the conversation understanding AI service server 106 via the communication network 104. According to one embodiment of the present disclosure, the communication unit 402 is configured such that, for example, a signal obtained on the response user input receiving unit 404 is transmitted to the conversation understanding AI service server 106 through the communication network 104 according to a predetermined protocol. can do. According to one embodiment of the present disclosure, the communication unit 402 is a user generated from various signals received from the conversation understanding AI service server 106 via, for example, the communication network 104, for example, a user input signal, a user input signal thereof, and the like. An intent, one or more conversation response candidates corresponding to the user intent, various analysis results, and the like may be received, and appropriate processing may be performed according to a predetermined protocol.

According to one embodiment of the present disclosure, the response user input receiving unit 404 receives natural language input (eg, voice and / or text input) from a response user (eg, a counselor) on the response user terminals 108a-108m. can do. According to an embodiment of the present disclosure, the response user input receiver 404 may include, for example, a microphone and an audio circuit, and may acquire a user voice input signal through the microphone and convert the obtained signal into audio data. According to an embodiment of the present disclosure, the response user input receiver 404 may include various pointing devices such as a mouse, a joystick, a trackball, a keyboard, a touch panel, a touch screen, a stylus, and the like, and various types of input devices. The user may acquire text input and / or touch input signals input from the user through these input devices. According to one embodiment of the present disclosure, natural language input from the responding user is received from the conversation understanding AI service server 106, for example, via one or more of the conversation response candidates described above, namely, the communication network 104 and the communication unit 402. One of the one or more conversation response candidates that match the user intent may correspond to that selected by the responding user on the responding user terminals 108a-108m (however, this disclosure is not so limited). According to one embodiment of the present disclosure, the natural language input from the responding user is one of the one or more conversation response candidates corresponding to the user intent, for example, received from the conversation understanding AI service server 106, and the response user terminal 108a-. 108m) may be text input automatically generated on the online chat session by selection by the responding user.

According to an embodiment of the present disclosure, the information visualization / screen output unit 406 may be a user generated from various signals received from the dialogue understanding AI service server 106 via the communication network 104, for example, a user input signal. Tent, one or more conversation response candidates corresponding to the user intent, and various analysis results (eg, user profile information obtained by the user profile analysis module 310 of FIG. 3 described above, emotion analysis module 312 of FIG. 3). User's emotion information determined by, and a predicted value of the target achievement probability at the current time point) may be converted into visual information and displayed. According to one embodiment of the present disclosure, the information visualization / screen output unit 406 includes various display devices such as a touch screen based on technologies such as LCD, LED, OLED, QLED, etc. A visual response corresponding to the input may be presented to the user, such as text, symbols, video, images, hyperlinks, animations, various notes, and the like.

In the above-described embodiment of the present disclosure with reference to FIGS. 1 to 4, the dialogue understanding AI system has been described mainly in the case of being utilized in a customer service center or the like, but it should be understood that the present disclosure is not limited thereto. According to another embodiment of the present disclosure, it should be understood that the conversation understanding AI system may be applied to various types of online chat sessions between users via the conversation understanding service server in addition to the customer consultation center. In addition, in the embodiment of the present disclosure described above with reference to FIGS. 1 to 4, the visual information of the information analyzed by the conversation understanding AI service server 106 is presented through a screen output unit on the response user terminals 108a-108m. Although described with a focus on the case, the present disclosure is not limited thereto. According to another embodiment of the present disclosure, visual information of the information analyzed by the conversation understanding AI service server 106 may be presented on a manager terminal (not shown) managing the plurality of response user terminals 108a-108m. It should be noted that it may be.

In the embodiment of the present disclosure described above with reference to FIGS. 1 to 4, for convenience, a specific module has been described as performing certain operations, but the present disclosure is not limited thereto. According to another embodiment of the present disclosure, it should be understood that the operations described as performed by any particular module in the above description may be performed by a separate module from each other.

5 is a diagram illustrating an example of a screen configuration that may be presented on a screen output unit of the response user terminals 108a-108m according to an embodiment of the present disclosure. The illustrated example relates to, for example, a chat between a user who calls a shopping mall customer center (corresponding to one of the customer user terminals 102a-102n) and a counselor (corresponding to the response user terminals 108a-108m). Can be.

As shown, the right side of the screen shows a chat window 502 showing the chat content between the user (corresponding to one of the customer user terminals 102a-102n) and the counselor (corresponding to the response user terminals 108a-108m). It is. It should be appreciated that the chat window 502 is only one example and may have a variety of other forms that a person skilled in the art would think of. As shown, the chat window 502 shows a conversation history between the user and the counselor.

As shown in the upper right box 504, a user intent obtained by natural language processing of a newly received input sentence from a user, that is, "I need to make a wooden plank with holes for my son's Arkham", is a sentence. (Ie, the intent corresponding to the inner mind of the user who said the user input sentence above is represented, for example, in the form of "Do I really have to buy this?"). According to one embodiment of the present disclosure, although not shown in detail in the drawing, a manner (eg, visual) that a user on a corresponding terminal can easily recognize when a sentence representation of an intent of a user input sentence is displayed on a screen. It should be noted that the present invention can be expressed in various ways, including prominent color, highlight or bold type.

In the box 506 below the box 504, profile information of the gender, age, and language of the user on the customer user terminals 102a-102n is displayed, and in the box 508 below, the input by the user is displayed. It can be seen that the current emotional state of the user on the customer user terminals 102a-102n, which is obtained and analyzed from the sentence and the like, is displayed as a probability for each emotional marker. As shown, it can be seen that the emotional state of the current user is most prevalent in the state of tranquility.

As shown in FIG. 5, the box 510 at the bottom of the box 508 is a result of synthesizing input sentences from the above user user terminals 102a-102n and various other analysis results, and presenting them to the user at the present stage. One candidate response that can be provided (ie, "Wheat if we customize you a wooden plank with holes on it") is indicated. Although not specifically illustrated in the drawing, according to one embodiment of the present disclosure, a highly visible method (eg, visually significant color, highlighting) that is easily recognized by a user on a corresponding terminal when expressing a candidate response on a screen Or in various ways including bold type).

As shown, the box 512 at the bottom left shows a prediction of the current target achievement probability based on the history of the conversation so far between the user and the counselor. As described above, the predicted value of the target achievement probability may be determined based on, for example, a conversation template prepared in advance.

6 is a diagram conceptually illustrating an example of a conversation template 308 for the conversation understanding AI service server 106.

As shown, conversation patterns that may be performed on the conversation understanding AI service server 106, included in the conversation template 308, that is, the pattern of the sequential flow of intents may be classified into nine types. As shown, each of A, B, C, D, E, F, G, H, I, J, and K shown in the rows of the conversation pattern column represents each intent, and each row of the conversation pattern column represents an intent. To show any combination of them and the sequential flow between them. As shown, each conversation pattern is also associated with each success and failure count and thus success rate. For example, in the first row, the first row has a pattern of A-> B-> C-> D-> E among the conversation patterns, and the pattern has 70 target successes and 30 failures. Therefore, the success rate is indicated as 0.7. As shown, in the case of the conversation pattern of the last ninth row, that is, A-> D-> C-> K-> G, the success rate is 1 and the failure number is 0, and the success rate is 1. According to an embodiment of the present disclosure, for example, the template may be configured to include only conversation patterns whose occurrence number exceeds a predetermined reference value (for example, 30), in which case A-> D-> of the ninth row described above. Note that the pattern C-> K-> G can be excluded from the template.

According to one embodiment of the present disclosure, when the flow of intents shown on a given chat session is A-> B-> C so far, the conversation understanding AI service server 106 is based on the conversation template 308. In addition, two patterns, one row and five rows, can be obtained as conversation patterns that can occur in the future. According to one embodiment of the present disclosure, in this case, the conversation understanding AI service server 106 compares each success rate of two conversation patterns, e.g., row 1 and row 5, and selects one row indicating a higher success rate, The next step may be determined as the direction of the dialogue response D.

It is to be understood that the matters illustrated in FIG. 6 are merely conceptual representations of extremely simplified conversation patterns in order to facilitate understanding of embodiments of the present disclosure, and are not intended to limit the present disclosure. According to another embodiment of the present disclosure, it should be understood that various types of conversation patterns may appear in various forms.

As will be appreciated by those skilled in the art, the present invention is not limited to the examples described in the present disclosure but may be variously modified, reconfigured and replaced without departing from the scope of the present disclosure. It should be understood that the various techniques described herein may be implemented by hardware or software, or a combination of hardware and software.

A computer program according to an embodiment of the present disclosure may include a storage medium readable by a computer processor or the like, such as an EPROM, an EEPROM, a nonvolatile memory such as a flash memory device, a magnetic disk such as an internal hard disk and a removable disk, a magneto-optical disk, and It may be implemented in a form stored in various types of storage media, including a CDROM disk. In addition, the program code (s) may be implemented in assembly or machine language. All modifications and changes that fall within the true spirit and scope of this disclosure are intended to be covered by the following claims.

Claims (1)

And intervening in an online text chat session established between a first user terminal and a second user terminal remotely disposed with each other, to control a predetermined display associated with the second user terminal.