WO2018030672A1

WO2018030672A1 - Robot automation consultation method and system for consulting with customer according to predetermined scenario by using machine learning

Info

Publication number: WO2018030672A1
Application number: PCT/KR2017/007954
Authority: WO
Inventors: 김우섭
Original assignee: 주식회사 피노텍
Priority date: 2016-08-09
Filing date: 2017-07-24
Publication date: 2018-02-15
Also published as: KR101883185B1; KR20180017342A

Abstract

Disclosed is a robot automation consultation method and system for consulting with a customer according to a predetermined scenario using machine learning. A robot consultation system which performs a consultation in the form of an online chat with a customer terminal possessed by a customer who has joined a robot consultation service according to an embodiment of the present invention may comprise: a chatting server for sending or receiving a conversation content in the form of an online chat with the customer terminal; and a consultation server which analyzes a conversation content received from the chatting server by a machine learning technique and, on the basis of an analysis result, reproduces a predetermined consultation scenario to carry out a scenario consultation with the customer terminal when a corresponding intro-node exists and carries out a normal consultation when a corresponding intro-node does not exist.

Description

Robot automatic consultation method and system for consulting with customer in fixed scenario using machine learning

The present invention relates to a robot automatic consultation method and system for consulting with a customer in a predetermined scenario utilizing machine learning.

In the financial sector, a telephone call is generally used as a method for customer consultation. A call center system has been established and operated for efficient counseling and customer management.

However, in case of call center system, it is required to establish counseling system, open a telephone network, and operate a counselor.In case of a large number of counseling calls, a large number of telephone lines can be opened for simultaneous access, and a counselor who can handle telephone counseling can be used. Fixed costs of enormous expenses are incurred due to the operation of telecommunication facilities and counseling personnel. In addition, counseling users often have to wait for a long time to connect to the consultation telephone.

In addition, in the case of the call center system, counseling is performed according to the working hours of the counselor, so that the counseling is not possible when the consultation center is not a designated counseling time.

In order to overcome such inconveniences, a counseling system has been developed that uses a chat or a message transmission between mobile terminals to develop a counseling system. However, there is a limit that a counselor still needs to respond to a customer question.

In this regard, Korean Patent Registration No. 10-1339838 (December 4, 2013) discloses a financial counseling system and method using a mobile terminal.

In the present invention, in the case of automatic consultation with a customer based on the server-based robot chat method, if there is a predetermined scenario, the machine enters the scenario by using a machine learning technology so that the consultation is performed in a predetermined scenario. It is to provide a robot automatic consultation method and system for consulting with customers in a predetermined scenario using machine learning that enables robot consultation.

The present invention utilizes machine learning to make the current customer consultation more smoothly and efficiently by outputting a question-and-answer form on a separate display window for questions that may be of interest to the customer while proceeding with the automatic consultation in a predetermined scenario. It is to provide a robot automatic consultation method and system for consulting with a customer in a predetermined scenario.

Other objects of the present invention will become more apparent through the preferred embodiments described below.

According to an aspect of the present invention, a robot consultation system for consultation in the form of a chat with a customer terminal possessed by a customer subscribed to the robot consultation service, a chat server that exchanges the contents of the chat in the form of a chat with the customer terminal; And analyzing the conversation content received from the chat server using a machine learning technique, and if there is a corresponding intro node based on the analysis result, plays a prescribed counseling scenario to perform a scenario counseling with the customer terminal and a corresponding intro node exists. If not, a robot consultation system including a consultation server for performing general consultation is provided.

The consultation server may include a customer management module for managing a customer who subscribes to the robot consultation service; A counseling DB management module for managing learning data for entering one of the counseling scenarios for robotic counseling and one or more of the counseling scenarios in a database unit; It may include a chat management module for analyzing the contents of the conversation received from the chat server by a machine learning technique and selecting a predetermined consultation scenario from among the various consultation scenarios based on the analysis result.

The counseling DB management module manages a counseling scenario DB for storing one or more counseling scenarios, wherein the counseling scenario includes an intro node for initial entry, intro learning data used when entering the intro node, and an intro node. It can include one or more scenario nodes that make up the scenario as a child node of the connected hierarchy.

The intro node may be a mobile node designated in consideration of a node ID which is an ID for identifying the intro node, a query to be transmitted from the intro node to the customer terminal, and a number of cases that may occur in response to the query. An ID and multiple choice items or subjective learning data matched with the mobile node ID may be included.

The intro learning data includes one or more query statements that are expected to be able to enter the counseling scenario, and the chat management module uses one or more of keywords, nouns, words, and morphemes for one or more of the query statements in a natural language processing manner. The method may further include a machine learning unit for analyzing and matching a representative query having high similarity with a machine learning technique.

The scenario node is matched with a node ID for identifying a current node, a query statement for making a query, a mobile node ID for designating a number of cases that may occur in response to the query statement, and the mobile node ID. It may include multiple choice items or subjective learning data.

The counseling DB management module additionally manages a FAQ DB for storing frequently asked questions and answers (FAQ) data in the counseling process, and the chat management module includes a main chat window for playing the counseling scenario and the counseling scenario. The sub chat window for reproducing the query and answer data retrieved from the FAQ DB can be managed together.

The chat server may include a messenger module configured to open a messenger type chat window between the customer terminal and the counseling server to communicate with each other, and generate a web chat page capable of two-way conversation to invite the customer terminal to be customized. It may include one or more of the web chat module to facilitate the consultation.

The chat management module may include: a feature extractor configured to extract a vocabulary having a relatively high value by calculating a value of the vocabulary from a customer input sentence; Machine learning may be performed based on the feature to infer the most similar sentence among the customer expected queries previously registered in the database, and may include a machine learning unit for determining the existence of the intro node corresponding to the inferred sentence.

On the other hand, according to another aspect of the present invention, a robot consultation method in the robot consultation system that performs a consultation in the form of chat with the customer terminal possessed by the customer to join the robot consultation service, activating a chat window in which the customer terminal participates step; Receiving an initial query from the customer terminal; Determining an existence of an intro node based on learning data learned by a machine learning method according to a result of analyzing the first query; When the intro node exists, the robot counseling method comprising the step of playing a consultation scenario corresponding to the intro node in the chat window, and performing a general consultation if the intro node does not exist.

When reproducing the consultation scenario, the method may further include reproducing FAQ data related to the consultation scenario in a sub chat window.

Requesting feedback from the customer terminal and receiving a response when the consultation scenario is completed; And learning the learning data using a machine learning technique based on the response.

Other aspects, features, and advantages other than those described above will become apparent from the following drawings, claims, and detailed description of the invention.

According to the present invention, if there is a predetermined scenario in automatic consultation with a customer based on the server-based robot chat method, the machine enters the scenario using a machine learning technology so that the consultation is made in a predetermined scenario. It is effective to enable robot consultation with the doctor.

In addition, the automatic consultation in a predetermined scenario, the questions that can be wondered from the customer's point of view in a separate display window in the form of a query-response has the effect of making the current customer consultation more smooth and efficient.

1 is a view showing a robot consultation system and linkage system according to an embodiment of the present invention;

2 is a block diagram showing a detailed configuration of a consultation DB management module;

3 is an exemplary diagram for explaining machine learning feature extraction;

4 is an exemplary diagram for explaining a lexical value measurement;

5 is an exemplary diagram for explaining a feature distance measurement;

6 is an exemplary diagram for explaining a typo distance measurement;

7 is an exemplary diagram for explaining a spaced distance measurement;

8 is a view showing a structure of a consultation scenario;

9 is an exemplary diagram for a configuration of a scenario node;

10 is a flowchart of a robot counseling method according to an embodiment of the present invention;

11 is a view showing the structure of the chat window and an example screen.

As the present invention allows for various changes and numerous embodiments, particular embodiments will be illustrated in the drawings and described in detail in the written description. However, this is not intended to limit the present invention to specific embodiments, it should be understood to include all changes, equivalents, and substitutes included in the spirit and scope of the present invention.

When a component is referred to as being "connected" or "connected" to another component, it may be directly connected to or connected to that other component, but it may be understood that other components may be present in between. Should be. On the other hand, when a component is said to be "directly connected" or "directly connected" to another component, it should be understood that there is no other component in between.

Terms such as first and second may be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. As used herein, the terms "comprise" or "have" are intended to indicate that there is a feature, number, step, action, component, part, or combination thereof described on the specification, and one or more other features. It is to be understood that the present invention does not exclude the possibility of the presence or the addition of numbers, steps, operations, components, components, or a combination thereof.

In addition, the components of the embodiments described with reference to the drawings are not limited to the corresponding embodiments, and may be implemented to be included in other embodiments within the scope of the technical spirit of the present invention. Even if the description is omitted, it is obvious that a plurality of embodiments may be reimplemented into one integrated embodiment.

In addition, in the description with reference to the accompanying drawings, the same components regardless of reference numerals will be given the same or related reference numerals and redundant description thereof will be omitted. In the following description of the present invention, if it is determined that the detailed description of the related known technology may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted.

1 is a view showing a robot counseling system and the linkage system according to an embodiment of the present invention, Figure 2 is a block diagram showing the detailed configuration of the consultation DB management module, Figure 3 is an illustration for explaining the machine learning feature extraction 4 is an exemplary diagram for describing a lexical value measurement, FIG. 5 is an exemplary diagram for explaining a characteristic distance measurement, FIG. 6 is an exemplary diagram for explaining a typo distance measurement, and FIG. FIG. 8 is a diagram illustrating the structure of a counseling scenario, FIG. 9 is a diagram illustrating a configuration of a scenario node, and FIG. 10 is a flowchart of a robot counseling method according to an embodiment of the present invention. 11 is a view showing the structure of the chat window and an example screen.

Robot counseling system according to an embodiment of the present invention (counseling system through a robo advisor (or robo counselor)) performs a consultation via a chat via a messenger or web chat method with a customer terminal, according to the question input from the customer terminal General counseling will be conducted or robotic counseling based on robo chat will be performed through natural language processing based on a database on which case-by-case scenarios (conversation structures in which questions and answers are matched) are established. In addition, for financial counseling, necessary information may be obtained through API interworking with a financial institution's main system and provided to a customer terminal.

The robot consultation system 100 according to the present embodiment is connected to the customer terminal 10 through a network.

The customer terminal 10 is an electronic terminal on which the robot consultation application 12 having the consultation messenger library 14 can be installed. For example, the customer terminal 10 may be a smartphone, a mobile communication terminal, a PDA, a tablet PC, or a general PC equipped with an operating system. It can be one.

The customer terminal 10 installed with the robot counseling application 12 may separately designate a dedicated robot counselor (dedicated robo advisor) for counseling according to various fields, and may support a chat type counseling about a field to be consulted at any time. have.

The robot consultation system 100 analyzes the conversation contents input from the chat server 110 for conducting a consultation with the customer terminal 10 in a chat form, and the conversation contents input from the customer terminal 10 using a machine learning method, and plays back a predetermined consultation scenario. Or a counseling server 120 to proceed with general counseling so that appropriate counseling can be made.

The chat server 110 includes a messenger module 112 that opens a messenger type chat window between the customer terminal 10 and the counseling server 120 so as to communicate with each other. The messenger module 112 opens a chat window in which the chat subject customer and the robot counselor (hereinafter referred to as 'robo advisor') participate, and the customer message input by the customer terminal 10 in the chat window is consulted by the server 120. ) And a consultation message provided by the consultation server 120 to the customer terminal 10.

Here, the customer message may be a response to a question input by the customer or a request made by the consultation server 120. The consultation message may be an answer to a question entered by the customer or a request for a matter to be provided by the customer.

The chat server 110 includes a web chat module 114 that generates a web page (web chat page) capable of two-way conversation without a chat window being opened and invites the customer terminal 10 to proceed with a customized consultation. You may.

In the case of the web chat module 114, when a character is transmitted to a predetermined number from the customer terminal 10, the text content transmitted when the customer is a customer who has subscribed to the robot consultation service is a question capable of robot consultation. If it is determined, a web chat page for the customer is generated, and a URL accessible to the web chat page is transmitted to the customer terminal 10. The customer terminal 10 can be connected to the received URL to receive the robot consultation service.

In this case, the web chat page may be implemented to require authentication of the user, and it may be possible to check the contents displayed on the web page only for the user who has successfully authenticated the user.

The web chat module 114 may allow the advertisement content to be displayed during the process of accessing the web chat page from the customer terminal 10. In this case, by providing access to the web chat page only for the customer terminal 10 confirming the advertisement content, it is possible to obtain advertising revenue in providing the robot consultation service.

In addition, the chat server 110 pushes various messages related to the outbound consultation even when a chat window is not opened between the client terminal 10 for outbound consultation provided by the consultation server 120. It may further include a push server (not shown) to push. Alternatively, if a chat window is established between the client terminal 10 and the push server, the push server pushes a message related to the outbound consultation through the corresponding chat window or as a separate push message even if the customer message from the client terminal 10 is not transmitted. You may.

The counseling server 120 frequently occurs in the customer management module 122 for managing a customer who subscribes to the robot counseling service, various counseling scenarios for robotic counseling, learning data for entering various counseling scenarios, and / or counseling processes. Consulting DB management module 124 for managing the query and answer data (FAQ data) to be a database unit, and the customer message received from the chat server 110 to the general consultation or scenario consultation according to the results of the machine learning method In addition, when divided into scenario counseling, it includes a chat management module 126 to determine a suitable counseling scenario and to chat in accordance with a predetermined counseling scenario.

In addition, when a variety of information related to finance is required in the chat process according to the predetermined consultation scenario, the information request module 128 may be further included to request and obtain information for generating a consultation message in association with a financial institution's main system. have.

Referring to FIG. 2, the chat management module 126 analyzes a customer input sentence (customer message) transmitted from the chat server 110 through a messenger API and acquires it through natural language processing of analyzing keywords, nouns, words, and the like. It may include a sentence inference engine 210 for performing machine learning based on the results to find the nearest query stored in the database.

The sentence inference engine 210 performs inference on the customer input sentence using the machine learning result obtained by the machine learning tool 220.

The machine learning tool 220 may include a feature extractor 230 and a machine learning unit 240.

The feature extractor 230 extracts a feature from a customer input sentence transmitted from the customer terminal 10. For example, in the case of a natural language processing method, a feature may be a key keyword.

The feature extractor 230 may calculate a value of the vocabulary from the customer input sentence and extract a vocabulary having a relatively high value as a key keyword, that is, a feature. The value of the vocabulary can be calculated by analyzing the effect of each vocabulary on the intention of the question in the relevant sentence, and can be automatically analyzed into meaningful vocabulary and meaningless vocabulary.

Referring to FIG. 3, the vocabulary value is calculated in the example sentence, and two relatively high vocabulary are cards (83%) and lost (96.7%), and the words 'card' and 'lost' are inputted by the corresponding customer. It can be extracted as a feature from a sentence.

The vocabulary value can be determined by measuring the impact of the vocabulary on the question ID. Here, the question ID means an identification code prepared in advance to provide an appropriate answer to the customer. For example, if you prepared answers to 1000 kinds of customer input sentences, the number of question IDs would be 1000.

The value of each vocabulary may be calculated based on a result of determining how each keyword or vocabulary influences the selection of the question ID in the machine learning process. If the same keyword or vocabulary is used for different question IDs, the value is relatively low, and if it affects only a specific question ID, the value can be relatively high.

For example, referring to FIG. 4, the term 'lost' corresponds to a keyword having a high weight corresponding to question IDs such as 'card loss report', 'bankbook loss report', and 'wallet loss report'. In contrast, the vocabulary of 'what' corresponds to a keyword with a low weight corresponding to question IDs such as 'what is a card issuance document', 'what is a bank account loss report document', and 'what is a banquet fee'.

The feature extractor 230 includes a feature distance extractor 231, a synonym mapping unit 232, a keyword mapping unit 233, a noun mapping unit 234, a word mapping unit 235, and a typo distance measuring unit 236. It may include one or more of the spacing distance measurement unit 237.

The feature distance extractor 231 calculates a distance (error) between two features extracted from the customer input sentence. Referring to FIG. 5, a distance between a card and a loss, which are two features extracted from the sentence of FIG. 3, may be extracted as a feature distance from a map generated according to similarities of a plurality of vocabularies.

Vocabulary appearing in the same question ID when the map is generated according to the similarity of the vocabulary may be located relatively close to the distance map, and vocabularies not used in the same question ID may be located relatively far from the distance map.

The synonym mapping unit 232 finds and maps synonyms or synonyms with respect to the vocabulary classified in the sentence through the thesaurus.

The keyword mapping unit 233, the noun mapping unit 234, and the word mapping unit 235 each find and map keywords, nouns, and words analyzed through the morpheme analyzer.

The typo distance measuring unit 236 infers a vocabulary (or sentence) originally intended by measuring a typo distance when a typo exists in a customer input sentence.

Referring to FIG. 6, after the customer input sentences are separated into consonants, vowels, and support levels through keyboard unit analysis (separation of consonants and vowels, separation of double vowels into short vowels, etc.), distance analysis on the keyboard, typing habits The analytical process can measure the distance of a typo. When the measured distance of the typo falls within a predetermined threshold distance, a lexical candidate ranking calculated based on a feature in which a typo frequently occurs may be generated. For example, when the word "accounting account" is input, the phrase "account inquiry" is most similar among the vocabulary registered in the database, and thus it may be determined that it corresponds to a typo of the phrase.

The spacing distance measurement unit 237 infers a word (or sentence) originally intended by measuring a spacing distance when a spacing error exists in a customer input sentence.

Referring to FIG. 7, syllable unit analysis is performed on a customer input sentence, and if there is a spacing between syllables, it is determined whether proper spacing is present. If the spacing is not correct, stemming is impossible. Therefore, the optimal spacing is inferred while inputting and changing the number of various cases of spacing until morphological analysis is possible, and the spacing distance according to the deformation is obtained. For example, when input as "account_view_time", a spacing distance of "2" may be measured as compared with "account view".

As such, the feature extractor 230 may extract the feature by performing analysis on the customer input sentence.

The machine learning unit 240 performs machine learning based on the extracted feature and infers the most similar sentence among customer expected queries (statements identified by the question ID) registered in the database.

Based on the inferred sentence, it is determined whether a corresponding intro node exists, and if there is a counseling scenario, if there is a counseling scenario, the counseling server asks questions based on the determined scenario and allows the customer terminal to answer. Allow counseling to take place. In the case of general counseling in which a counseling scenario does not exist, the counseling server may provide an appropriate answer to the question input through the customer terminal.

Referring back to FIG. 1, the customer management module 122 constructs a customer database for a customer who has subscribed to the robot consultation service to manage customer status and individual customer information.

The counseling DB management module 124 manages a counseling database including a counseling scenario DB 22 and an intronode learning DB 24.

The counseling scenario DB 22 stores a plurality of counseling scenarios having a structure in which various questions and answers are matched.

The structure of the consultation scenario is shown in FIG. 8, and the configuration of the scenario node is shown in FIG.

The counseling scenario includes an intro node for entering the scenario, an intro learning data utilized for entering the intro node, and a scenario node constituting the scenario with subnodes of a hierarchy connected to the intro node.

An intro node is a mobile node ID designated by considering a node ID, which is an ID for identifying the node, a query to be transmitted from the node to the customer terminal 10, and a number of cases that can be generated in response to the query. It includes multiple choice items or short answer learning data matched with the node ID.

Intro learning data is data for entering an intro node, and includes one or more query statements that are expected to enter the scenario.

For example, suppose you have an intro node that corresponds to a "lost card". A number of hints suggest a card loss, such as "I lost my card", "I lost my wallet", "I lost my card, what should I do?" If the query is constructed with intro learning data to match the representative query with high similarity by machine learning, then the intro node called "card loss" is determined if the phrase means card loss based on the result of machine learning even if the customer makes a similar inquiry. Will be connected.

A scenario node is a node that is connected to the intro node and constitutes a scenario. A scenario node is a multiple choice item that matches a node ID for identifying the current node, a query statement for querying, a mobile node ID that specifies the number of cases that can occur in response to the query, and a mobile node ID. Or subjective learning data.

Sanirio nodes have a hierarchical structure in which intro nodes are arranged at the top level. As the customer progresses to the lower tier, the customer's requirements can be further subdivided to understand them in detail.

The counseling DB management module 124 may include a machine learning unit for learning intro learning data for entering an intro node for various counseling scenarios.

The machine learning unit performs machine learning to find the nearest query based on the result of analyzing keywords, nouns, words or morphemes for one or more queries expected to enter the intro node. The result of machine learning is to find keywords, nouns, words or morphemes that are frequently used above the predetermined criteria for the intro node, and to study the correlations (eg word order, appearance rate, etc.) with intro learning data. Can be stored in the DB (24).

The machine learning unit may perform the learning through the machine learning on the subjective learning data in addition to the intro learning data.

At this time, the FAQ DB (not shown) may store frequently asked question and answer data (FAQ data).

When the chat proceeds according to the prescribed consultation scenario, the customer provides answers to the questions provided in the scenario. In this case, the customer manually chats in the direction specified in the scenario, and in this process, the customer may want to inquire.

Therefore, in the present embodiment, by having a sub chat window in addition to the main chat window, the FAQ data associated with the counseling scenario currently being played in the main chat window can be played in a chat format, thereby resolving customer's curiosity.

The chat management module 126 activates the chat window when the robot consultation request is received from the customer terminal 10 (step S300). The chat window may be implemented as a messenger type or a web chat type.

When receiving an initial query (customer input sentence) from the customer through the activated chat window (step S305), it is determined whether there is a corresponding intro node (step S310). The determination of the existence of an intro node may be performed based on whether the intro node corresponding to the representative query is found based on finding the closest representative query based on the intro learning data described above.

If there is no corresponding intronode, there is no predetermined counseling scenario. In this case, general counseling is performed (step S315). In the case of general consultation about customer query, it is performed by searching and providing the answers registered in the database.

If there is a corresponding intro node, the chat management module 126 reproduces the counseling scenario corresponding to the selected intro node (step S320). When the counseling scenario is played, it can be played from the intro node.

Here, the chat management module 126 may include a scenario player (not shown) that finds a suitable counseling scenario for a customer query and plays it in a chat window.

The scenario player understands the relationship between the nodes constituting the counseling scenario composed of many nodes, and moves to the next node determined according to the customer's response transmitted from the customer terminal 10 to the question transmitted from the counseling server 120. Play the player.

In order to move to the next node ID, the scenario node may have a multiple choice item or a short answer item.

In the case of multiple choice, as illustrated in FIG. 9, an item corresponding to each answer is determined. When a specific answer is selected, the mobile node ID is determined. When present in a subjective form, one or more possible answers are arranged to determine which node to move next through machine learning.

The chat management module 126 may reproduce the FAQ data as well as the prescribed consultation scenario (step S330). In this case, the chat management module 126 may further include a FAQ player (not shown) for reproducing the FAQ data.

Referring to FIG. 11, when a consultation scenario to be played in the main chat window 410 is determined, if answers to frequently asked questions from customers who have consulted according to the corresponding consultation scenario are arranged as FAQ data, The FAQ data corresponding to the consultation scenario may be found and reproduced in the sub chat window 420.

For example, in the main chat window 410, when the customer wants to receive a loan consultation, the counseling may be started while asking about the annual income of the customer according to the prescribed consultation scenario. At this time, the FAQ about credit loan qualification requirements, which may be a question from the customer's point of view of the loan consultation, is expressed in a chat form on the sub chat window 420, so that the customer may miss related to the loan consultation. It is also possible to confirm the information, so that sufficient information can be obtained by one consultation.

Referring back to FIG. 10, while performing a robot consultation in accordance with a predetermined consultation scenario, the customer terminal 10 may receive feedback on a robot consultation currently in progress (step S340). Feedback requested to the customer may be, for example, whether it was suitable as a consultation scenario for the customer's initial query or whether there was an additional item of consultation content that the user wanted to know.

The machine learning unit may perform machine learning on the intro learning data and the subjective learning data based on the customer feedback to update the learning data (step S345). For example, if the customer is satisfied with the current counseling scenario, the query sent by the client is regarded as having keywords, words, morphemes, etc. suitable for the counseling scenario, and additional learning about the intro learning data is performed. Could be.

The robot consultation system 100 according to the present embodiment opens a chat window between the customer terminal 10 and the consultation server 120 through the chat server 110 so that free consultation is performed in a messenger or web chat format. In addition, the customer message input from the customer terminal 10 automatically finds and reproduces the appropriate counseling scenario in the counseling DB through natural language processing (sentence recognition), so that even if the counseling personnel are not directly involved, smooth counseling This can be done to provide differentiated services to customers. In addition to replaying the counseling scenario, it also plays back FAQ data to provide information about what customers can miss during the counseling process.

The robot counseling method according to the present embodiment described above can be embodied as computer readable codes on a computer readable recording medium. Computer-readable recording media include all kinds of recording media having data stored thereon that can be decrypted by a computer system. For example, there may be a read only memory (ROM), a random access memory (RAM), a magnetic tape, a magnetic disk, a flash memory, an optical data storage device, and the like. The computer readable recording medium can also be distributed over computer systems connected over a computer network, stored and executed as readable code in a distributed fashion.

Although the above has been described with reference to a preferred embodiment of the present invention, those skilled in the art to which the present invention pertains without departing from the spirit and scope of the present invention as set forth in the claims below It will be appreciated that modifications and variations can be made.

Claims

It is a robot consultation system that conducts consultations in the form of chats with customer terminals possessed by customers who subscribe to the robot consultation service.

A chat server that exchanges chat contents with the customer terminal in a chat format; And

If the chat contents received from the chat server are analyzed by machine learning, and there is a corresponding intronode based on the analysis result, a predetermined counseling scenario is played to perform a scenario counseling with the client terminal and a corresponding intronode does not exist. Robot consultation system comprising a consultation server to perform a general consultation if not.
The method of claim 1,

The consultation server,

A customer management module for managing a customer subscribed to the robot consultation service;

A counseling DB management module for managing training data for entering one of the counseling scenarios for robotic counseling and one or more of the counseling scenarios in a database unit;

And a chat management module that analyzes the conversation contents received by the chat server using a machine learning technique and selects a predetermined counseling scenario from among the various counseling scenarios based on the analysis result.
The method of claim 2,

The counseling DB management module manages a counseling scenario DB for storing one or more counseling scenarios,

The counseling scenario may include one or more scenario nodes constituting the scenario as an intro node for initial entry, intro learning data used when entering the intro node, and lower nodes of a hierarchical structure connected to the intro node. Robot consultation system.
The method of claim 3,

The intro node may be a mobile node designated in consideration of a node ID which is an ID for identifying the intro node, a query to be transmitted from the intro node to the customer terminal, and a number of cases that may occur in response to the query. And a multiple choice item or subjective learning data matched with the ID and the mobile node ID.
The method of claim 3,

The intro learning data includes one or more questions that are expected to be able to enter the consultation scenario,

The chat management module may further include a machine learning unit configured to analyze one or more of keywords, nouns, words, and morphemes with respect to one or more of the query statements in a natural language processing manner and match the representative query with high similarity with a machine learning technique. .
The method of claim 3,

The scenario node is matched with a node ID for identifying a current node, a query statement for making a query, a mobile node ID specifying a number of cases that may occur in response to the query statement, and the mobile node ID. Robot consultation system comprising a multiple choice item or subjective learning data.
The method of claim 3,

The counseling DB management module additionally manages a FAQ DB that stores frequently asked questions and answers (FAQ) data in the counseling process.

And the chat management module manages a main chat window for playing the counseling scenario and a sub chat window for playing the question and answer data retrieved from the FAQ DB in relation to the counseling scenario.
The method of claim 1,

The chat server,

A messenger module configured to open a messenger type chat window between the client terminal and the counseling server to communicate with each other;

Robot consultation system comprising one or more of the web chat module for generating a web chat page capable of two-way conversation to invite the customer terminal to proceed with a customized consultation.
The method of claim 2,

The chat management module,

A feature extraction unit for extracting a vocabulary having a relatively high value by calculating a value of the vocabulary from a customer input sentence;

And a machine learning unit configured to perform machine learning based on the feature to infer the most similar sentence among customer expected queries previously registered in the database, and to determine the existence of the intro node corresponding to the inferred sentence. Consultation system.
As a robot counseling method in a robot counseling system where a counseling service is conducted in a chat format with a customer terminal possessed by a robot counseling service.

Activating a chat window in which the client terminal participates;

Receiving an initial query from the customer terminal;

Determining an existence of an intro node based on learning data learned by a machine learning method according to a result of analyzing the first query;

Playing the counseling scenario corresponding to the intro node in the chat window when the intro node exists, and performing general counseling when the intro node does not exist.
The method of claim 10,

And reproducing the FAQ data related to the consultation scenario in a sub chat window when reproducing the consultation scenario.
The method of claim 10,

Requesting feedback from the customer terminal and receiving a response when the consultation scenario is completed; And

And learning the learning data using a machine learning method based on the response.