KR101719196B1 - Apparatus for forecasting purpose of customer counsel and computer-readable medium thereof - Google Patents

Abstract

Disclosed is a computer-readable medium for analyzing big data to classify a customer's counseling intent in an ongoing customer consultation and provide feedback based thereon.
The computer-readable medium stores instructions for causing a computer to perform the following operations: receiving an ongoing customer consultation data, recognizing a keyword in received customer consultation data, recognizing the recognized keyword And classifying the customer consultation intention based at least in part on the at least partially filtered customer consultation intent and providing corresponding feedback based at least in part on the classified customer consultation intent.

Description

[0001] APPARATUS FOR FORECASTING PURPOSE OF CUSTOMER COUNSEL AND COMPUTER READABLE MEDIUM THEREOF [0002]

The present invention relates to a customer consultation of a call center. More specifically, the present invention is an invention for predicting and responding to a customer's counseling intent by using big data analysis.

With the growth of the Internet and mobile industries, big data environments are emerging, meaning large, fixed or unstructured data beyond the scale of being able to be collected, managed, stored and analyzed with existing database management tools. Big data is a vast amount of data of various types generated in the digital environment, and it is considered as a core resource that can secure future competitiveness of companies and the like.

Also, extracting economically necessary value from big data can be referred to as big data technology. Utilizing Big Data can create various forms of value, such as establishing marketing strategies, optimizing processes, improving productivity and preventing fraud. Therefore, big data technology for effectively utilizing big data can be considered as a means of securing corporate competitiveness that should be considered first in the infinite competition age regardless of the type of enterprise.

The insurance industry can also use Big Data to secure its competitiveness. For example, an insurance company can use Big Data to serve customers at its call center. In the case of telephone counseling at the call center, there may be silent periods where both the counselor and the client are silent. Reducing the silence time allows the company to more efficiently operate its call center, and reduces the waiting time for customers, resulting in higher customer satisfaction.

This silence period is mainly caused by the waiting time for the agent to access the computer system at the agent terminal in order to process the consultation item according to the customer request. The silent time may be caused by a difference in skill of the agent. In the case of a skilled agent, the waiting time can be reduced by listening to the customer's consultation contents and accessing the computer system in advance based on the expected consulting type.

As described above, if the counselor can anticipate and respond to the client's consultation regardless of proficiency, it can reduce the silence time and utilize the counselor more efficiently. Since this improved call center consultation method can also increase the satisfaction of the customer's consultation, there is a demand in the insurance industry for such a customer consultation device.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned problems, and an object of the present invention is to provide a device and a computer-readable medium for efficiently performing customer consultation in a call center by utilizing big data.

It is also intended to provide an apparatus and a computer-readable medium for providing a constant quality of consultation regardless of the skill of the agent.

The present invention also provides a device and a computer-readable medium for more efficient use of call center personnel for customer consultation by reducing silence time in consultation.

Furthermore, the present invention is intended to provide a device and a computer-readable medium for efficiently utilizing company resources by utilizing big data.

A computer-readable medium according to an embodiment of the present invention for realizing the above-mentioned problems is disclosed. The computer-readable medium having stored thereon instructions for causing a computer to perform the following operations: recognizing a keyword in the ongoing customer consultation data received, determining, based at least in part on the recognized keyword Classifying customer consultation intentions, and providing corresponding feedback based at least in part on the classified customer consultation intentions.

According to an embodiment of the present invention, a consultation server for providing a consulting interface predicted from customer's consultation data is disclosed. The server includes a receiving module for receiving customer consultation data in real time, a keyword recognizing unit for classifying the customer consulting intention to be consulted by the customer on the basis of the at least one keyword, And a transmission module for transmitting the generated feedback data to the agent terminal.

According to another embodiment of the present invention, a computer-readable medium is disclosed for increasing the accuracy of a customer's counseling intent prediction. The computer-readable medium stores instructions for causing a computer to perform the following operations: grouping customer consultation data into counseling provision data units that include counseling history that has been provided from an agent; Extracting a keyword from the grouped customer consultation data, and storing the correlation between the extracted keyword and the consultation providing data.

According to another embodiment of the present invention, a consultation server for providing a consulting interface predicted from customer's consultation data is disclosed. The server includes a receiving module for receiving consultation providing data including a consultation history that has been provided from an agent, a control module for matching the consultation providing data with the customer consultation data, and extracting a keyword from the customer consultation data based on the matching result And a database for storing a correlation between the extracted keyword and consultation providing data.

In accordance with a further embodiment of the present invention, an agent terminal is provided that provides a consulting interface that is predicted from customer ' s consultation data. The agent terminal includes a transmission module for transmitting the on-going customer consultation data to the server, a receiving module for receiving feedback on the basis of the customer consultation data from the server, and an on- And a display module for outputting an evaluation information input interface for the evaluation information.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-described problems, and it is possible to more efficiently perform customer consultation in a call center by utilizing big data.

Further, the present invention can provide a constant consulting quality irrespective of the proficiency of the counselor.

Further, the present invention can reduce the silence time in consultation and utilize the call center personnel for customer consultation more efficiently.

Furthermore, the present invention can efficiently utilize company resources by utilizing big data.

The effects of the present invention are not limited to those mentioned above, and other effects not mentioned can be clearly understood by those skilled in the art from the following description.

Various aspects are now described with reference to the drawings, wherein like reference numerals are used to refer to like elements throughout. In the following examples, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of one or more aspects. However, it will be apparent that such aspect (s) may be practiced without these specific details. In other instances, well-known structures and devices are shown in block diagram form in order to facilitate describing one or more aspects.
Figure 1 illustrates a consultation system in accordance with an embodiment of the present invention.
Figure 2 illustrates server components in accordance with an embodiment of the present invention.
Figure 3 illustrates information or data stored in a database in accordance with one embodiment of the present invention.
4 schematically illustrates operation of a server according to an embodiment of the present invention.
Figure 5A illustrates an example of a probabilistic mapping of keywords, customer consultation intentions, and feedback in accordance with one embodiment of the present invention.
FIG. 5B illustrates an example of a correlation between a keyword, a customer consultation intention, and feedback according to an embodiment of the present invention.
FIG. 6 illustrates an example of a correlation between a keyword, a customer consultation intention, and feedback according to another embodiment of the present invention.
7 shows an example of a method of extracting keywords from consultation providing data according to another embodiment of the present invention.
FIG. 8 illustrates a method for predicting a customer consultation intention and generating feedback information according to an embodiment of the present invention.
FIG. 9 shows a method of extracting keywords from consultation providing data according to another embodiment of the present invention.

Various embodiments and / or aspects are now described with reference to the drawings. In the following description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of one or more aspects. However, it will also be appreciated by those of ordinary skill in the art that such aspect (s) may be practiced without these specific details. The following description and the annexed drawings set forth in detail certain illustrative aspects of one or more aspects. It is to be understood, however, that such aspects are illustrative and that some of the various ways of practicing various aspects of the principles of various aspects may be utilized, and that the description set forth is intended to include all such aspects and their equivalents.

In addition, various aspects and features will be presented by a system that may include multiple devices, components and / or modules, and so forth. It should be understood that the various systems may include additional devices, components and / or modules, etc., and / or may not include all of the devices, components, modules, etc. discussed in connection with the drawings Must be understood and understood.

As used herein, the terms "an embodiment," "an embodiment," " an embodiment, "" an embodiment ", etc. are intended to indicate that any aspect or design described is better or worse than other aspects or designs. . As used herein, the terms 'component,' 'module,' 'system,' 'interface,' and the like generally refer to a computer-related entity and include, for example, hardware, It can mean software.

In addition, the term "or" is intended to mean " exclusive or " That is, it is intended to mean one of the natural inclusive substitutions "X uses A or B ", unless otherwise specified or unclear in context. That is, X uses A; X uses B; Or when X uses both A and B, "X uses A or B" can be applied to either of these cases. It should also be understood that the term "and / or" as used herein refers to and includes all possible combinations of one or more of the listed related items.

It is also to be understood that the term " comprises "and / or" comprising " means that the feature and / or component is present, but does not exclude the presence or addition of one or more other features, components and / It should be understood that it does not. Also, unless the context clearly dictates otherwise or to the contrary, the singular forms in this specification and claims should generally be construed to mean "one or more. &Quot;

As used herein, the terms "information" and "data" are often used interchangeably.

Figure 1 illustrates a consultation system in accordance with an embodiment of the present invention.

1, the consulting system 100 can be largely composed of a server 110, a user terminal 120, an agent terminal 130, and a network 140 .

Also, the components that make up the consulting system 100 shown in FIG. 1 are exemplary and only some of the components may constitute the consulting system 100, or additional components (s) Or may be included in the system 100. For example, although not shown, the server 110 may include an Enterprise Application Integration (EAI) server, a customer identification information server, a call center server, a voice of customer (VOC), a database, a customer complaint management server and / . For example, server 110 in accordance with an aspect of the present invention may include a call center server (not shown) so that the information and / or data collected from the server may include information and / or data collected from the call center server . In addition, the server can collect information and / or data collected from the call center server by converting from voice to a string through a Speech to Text (STT) engine. In one aspect of the present invention, "Big Data" may mean information that is collected and generated by the server 110.

The server 110 may be configured to communicate with the user terminal 120 and the agent terminal 130, receive ongoing customer consultation data, recognize, extract and store keywords in the customer consultation data, Generating the feedback information based at least in part on the counseling intention prediction and classification, the classified customer counseling intention, and the keyword and customer counseling intention or customer consulting intention and feedback action based on the evaluation information on the feedback received at the agent terminal 130 And updating the correlation of the correlation. In addition, the server 110 groups the customer consultation data into consultation providing data units including consultation details provided from the agent, extracts keywords from the grouped customer consultation data, and extracts the extracted keywords and the consultation providing data The server may convert the customer's voice received from the call center into a character string, store the customer identification information, the contract information, the contract information, the product information and the insurance payment information, And to determine the insurance products to be designed, to decide which insurance products to sell, to generate financial consultants and call center agent training information, to control the intensity of insurance claims review, and to insure And generating recommendation information of the product.

The user terminal 120 can inquire about communication with the server 110 and / or the agent terminal 130, various identification information (for example, customer identification information, customer insurance registration information, customer use product information, etc.) Can be performed.

The agent terminal 130 can perform communication with the server 110 and the user terminal 120, for example. In addition, the agent terminal 130 may perform the operations performed in the server 110 described above. For example, the agent terminal 130 converts the voice of the calling customer into a string to generate customer consultation data, recognize the keyword in the customer consultation data, predict and classify the customer consulting intention based at least in part on the recognized keyword And so on. In addition, the agent terminal 130 transmits the on-going customer consultation data and the consultation providing data to the server, receives the feedback based on the customer consultation data from the server, and transmits the consultation information, the customer consultation intention, And an evaluation information input interface for the evaluation information input interface.

In addition, the network 140 may form a channel for data communication between the server 110, the user terminal 120, and the agent terminal 130.

The terms "system" and "network ", as used herein, are often used interchangeably.

The systems presented here are public switched telephone network (PSTN), x digital subscriber line (xDSL), rate adaptive DSL (RADSL), multi rate DSL (MDSL), very high speed DSL (VDSL) Asymmetric DSL), High Bit Rate DSL (DSL), and Local Area Network (LAN).

In addition, the system described herein can be applied to a mobile communication system such as Code Division Multi Access (CDMA), Time Division Multi Access (TDMA), Frequency Division Multi Access (FDMA), Orthogonal Frequency Division Multiple Access (OFDMA), Single Carrier- Various wireless communication systems such as other systems may be used.

The techniques described herein may be used in other systems as well as the systems mentioned above.

The network 140 according to one aspect of the present invention may be configured without discriminating the communication modes such as wired and wireless, and may be connected to various communication networks such as a personal area network (PAN), a wide area network (WAN) . Also, the network 140 may be a known World Wide Web (WWW) or may use a wireless transmission technology used for short-distance communication such as infrared (IrDA) or Bluetooth have.

1, the user terminal 120 communicates with the server 110 and / or the agent terminal 130 to consult at a call center or the like, pay a fee, obtain customer identification information, insurance subscription information, Quot ;, " purchase ", " purchase ", etc., and the like.

Such a user terminal 120 may be a user equipment, a mobile, a PC, a mobile phone, a kiosk, a cellular phone, a cellular, a cellular terminal, a subscriber unit, a subscriber station, a mobile station, a terminal, a remote station, a PDA, Any device that uses a wireless connection mechanism, such as a terminal, user agent, cellular telephone, wireless telephone, Session Initiation Protocol (SIP) telephone, wireless local loop (WLL) station, And the like, but are not limited thereto.

Such a user terminal 120 may also be referred to as any device capable of using a wired connection mechanism such as a wired facsimile, a PC with a wired modem, a wired telephone, a terminal capable of wired communication, It does not.

In addition, the agent terminal 130 may include devices capable of using a wireless connection mechanism or a wired connection mechanism of the same or similar type as the user terminal described above.

In one aspect of the invention, the agent terminal 130 may refer to a terminal that can be used by those associated with the company (e.g., an insurance company employee and / or a company call center agent). That is, according to one aspect of the present invention, a particular terminal that is relevant to an insurance company (e.g., server 110 of an insurance company) may be referred to as an agent terminal 130.

Such an agent terminal 130 may also be a personal computer, a mobile phone, a kiosk, a cellular phone, a cellular phone, a cellular terminal, a subscriber unit, a subscriber station, a mobile station, a terminal, Such as a cellular telephone, a wireless telephone, a Session Initiation Protocol (SIP) telephone, a wireless local loop (WLL) station, a handheld device having a wireless connection capability, a wireless modem, Or any device using a wireless connection mechanism, and the like. Further, the agent terminal 130 may be referred to as, but not limited to, a wired facsimile, a PC with a wired modem, a wired telephone, any device capable of using a wired connection mechanism such as a wired communication enabled terminal .

In one aspect of the present invention, the server 110 may provide different services for each of the user terminals by the various users. That is, the user terminal used by the customer and the agent terminal 130 used by the financial company employee are allowed to access the specific server among the plurality of insurance company servers, respectively, in accordance with the policies and / Or may be limited. Alternatively, the user terminal 120 may be permitted or restricted to access specific data of the insurance company server according to the policies and / or user settings of the insurance company.

In an aspect of the present invention, the server 110, the user terminal 120, and the agent terminal 130 may send and receive various information through the network 140. [

For example, according to the policy of the insurance company, the agent terminal can provide various information such as insurance design, confirmation and registration of financial products, confirmation of the customer's identity, insurance claim, insurance payment, insurance payment, Determination of the product, determination of the insurance product to be sold, generation of the agent training information, and the like. According to an aspect of the present invention, the agent terminal 130 can transmit the on-going client consultation data to the server 110 through the network 140 and receive the server's feedback on the transmitted client consultation data And evaluation information on the feedback can be generated and transmitted to the server.

The above-described information may be input via the user terminal 120 and / or the agent terminal 130, or may be generated by the server 110. [ The above-described information may be stored in a database of the server 110 or a computer-readable storage medium. In accordance with an aspect of the present invention, a computer-readable storage medium may be included within the user terminal 120 and / or the agent terminal 130. [

Such storage media may include any type of storage medium in which programs and data are stored so that they can be read by a computer system. According to one aspect of the present invention, such medium may be a ROM (Read Only Memory), a RAM (Random Access Memory), a CD (Compact Disk) -ROM, a DVD Storage devices, and the like. Additionally, such media may be distributed over a system coupled to the network 140 and may store computer readable codes and / or instructions in a distributed manner.

In an embodiment of the present invention, a customer, a financial company employee, and / or an insurance agent (FC), when a particular application distributed by a particular insurance company is installed and executed in the user terminal 120 and / The server 110 can communicate with the user terminal 120 and / or the agent terminal 130 via the network 140 so that the user can enjoy various services.

Figure 2 illustrates components of a server in accordance with an embodiment of the present invention.

According to an aspect of the present invention, the server 110 may include a receiving module 201, a transmitting module 203, a controlling module 205, and a database 207.

The components of server 110 in FIG. 2 are exemplary and only some of the components may constitute server 110, or additional components (s) may be included in server 110 in addition to the components have.

Additionally, in accordance with aspects of the present disclosure, the components shown in FIG. 2 may be components of the agent terminal 130. In such a case, the control module and database of the agent terminal 130 may perform at least a part of the operations and functions of the server's control module and database, which will be described below. Alternatively, the receiving module of the agent terminal 130 may receive information and data required from the server 110. [ The display of the agent terminal 130 may display information received from the server and / or generated by its control module. Alternatively, at least a portion of the components of the server 110 described below may be replaced by components of the agent terminal 130.

The receiving module 201 may receive information or data transmitted from the user terminal 120 or the agent terminal 130 via the network 140. [ According to an aspect of the present invention, the receiving module 201 may receive the customer's consultation data in progress at the call center. The consultant data of the on-going customer is collected at the time between when the customer and the customer are on the call, when the customer is on the call, when the customer requests the consultation, Or customer consultation data generated. For example, when a customer requests a consultation through communication with an agent of a call center, the receiving module 201 may receive the ongoing customer consultation data. At this time, the customer's consultation data may include at least one of the customer's voice or the character string when the customer's voice is converted into a character string. The received customer consultation data can be stored in the database. Customer consultation data can be classified as big data as unstructured data in a general sense. Further, in one aspect of the present invention, the receiving module 201 may receive the evaluation information of the agent on the feedback generated from the server. The control module may update the database through the evaluation information of the agent. Further, in another aspect of the present invention, the receiving module 201 may receive consultation providing data including a consultation history that has been provided from the consultant.

The sending module 203 may transmit information or data from the server 110 to the user terminal 120 or the agent terminal 130 via the network 140. [ The sending module 203 may transmit a response message to the user terminal 120 or the agent terminal 130 in response to the request message from the user terminal 120 or the agent terminal 130. For example, the sending module 203 sends customer consultation data, customer identification information, customer use product information, a counseling code, a counseling interface, a customer counseling intention data, and a customer's insurance related information requested by the agent terminal 130 to an agent terminal (130).

The information transmitted by the sending module 203 may include information retrieved from the database 207 if necessary. Further, the sending module 203 may distribute the application including the user interface related to the insurance company to the user terminal 120 and / or the agent terminal 130. [

The transmission module 203 and the reception module 201 of the server 110 can communicate with the user terminal 120 or the agent terminal 130 via the network 140 as described above.

In an aspect of the present invention, the sending module 203 and the receiving module 201 may communicate with other servers and the like via the network 140. [

The transmitting module 203 and the receiving module 201 may include a wired / wireless Internet module for network connection. For example, WLAN (Wireless LAN) (Wi-Fi), Wibro (Wireless broadband), Wimax (World Interoperability for Microwave Access), HSDPA (High Speed Downlink Packet Access) For example, a wired Internet technology may be a digital subscriber line (XDSL), a fiber to the home (FTTH), or a power line communication (PLC).

The transmitting module 203 and the receiving module 201 may transmit data to and receive data from an electronic device including a local communication module located relatively close to the server 110 including the local communication module. Bluetooth, Radio Frequency Identification (RFID), infrared data association (IrDA), Ultra Wideband (UWB), ZigBee, and the like can be used as a short range communication technology.

In one aspect of the invention, the data or information collected or received via the receiving module 201 may be stored in the database 207 or transmitted to other electronic devices in the vicinity via the local communication module.

The control module 205 can control the overall operations of the server 110. [ The control module 205 may store the data received by the receiving module 201 in the database 207. [ In addition, the control module 205 may parse the data received by the receiving module 201 to generate feedback information and the like. Generating such feedback data may be performed by retrieving the necessary information from the database 207.

In one aspect of the invention, the control module 205 is configured to recognize keywords from the ongoing customer consultation data, predict and classify customer consulting intentions based at least in part on the recognized keywords, Lt; RTI ID = 0.0 > at least < / RTI >

For example, a customer can talk to an agent at a call center to talk to an insurance planner, to pay fees, or to claim insurance. The agent can cause the agent terminal 130 to display an interface for solving the consultation case according to the intention of requesting the client's consultation. For example, when the customer requests the consultation about the insurance claim, the agent can check the consultation terminal 130 with the information on the affiliated product of the customer, the expected amount of the insurance money received for the product, and respond to the customer's consultation request. At this time, in case of a skilled agent, prior to hearing the intention of consulting the customer directly from the customer during the conversation with the customer, it is possible to know in advance the consultation intention "insurance claim" based on the context of the conversation with the customer. The skilled agent can know the customer's intention in consultation with the customer in the context of the conversation with the customer in view of the agent's own experience while listening to the customer's voice while the call with the customer is in progress. Therefore, in the case of a skilled agent, the user can execute an interface related to the consultation intention before calling the "insurance claim" through the agent terminal 130. However, in the case of a counselor having a relatively low proficiency, since the counselor terminal 130 executes the relevant interface after hearing the counseling intention of the client, a waiting time for executing the interface related to the counselor terminal 130 is required. At this time, the wait time can be a silence period in which neither the customer nor the agent speaks. If the role of the skilled agent can be performed by the server 110, it is possible to save the silence time due to the difference of skill level of the agent, and the processing method of the server can be used for the agent training.

In one aspect of the invention, the control module 205 can recognize keywords in the received customer consultation data and predict and classify customer consulting intentions based at least in part on the recognized keywords. The control module 205 may predict and classify customer consulting intentions based on statistical methods in the keywords recognized in the customer consultation data. In this case, the control module 205 may, for example, determine that if the keywords associated with the "insurance claim" are recognized in the received customer consultation data, the control module 205 determines the customer consulting intention based at least in part on the recognized keyword "Claims" can be predicted and classified. When the keywords related to "insurance claim" are recognized, the control module 205 can predict and classify the customer consultation intention as "insurance claim" since "insurance claim" Also, the control module 205 can predict and classify the customer's counseling intention in consideration of the weight given to each keyword. For example, the received customer consultation data is "I was hospitalized at the hospital because I was injured yesterday, and my left arm bone was broken, but I saw that my mother got a premium when I broke a bone in the past. , The keywords can be "hurt", "hospitalized", "bone", "broken", "premium", "received". The control module 205 may know in advance that the customer's counseling intent is "insurance claim" based, at least in part, on the recognized keywords. The control module 205 can predict the customer's intention in consideration of the weight given to each keyword. Since the keywords "insurance premium" and "received" are more likely to be classified as intentions for customer consultation called "premium charging" than other keywords, the control module 205 assigns weights to the keywords, "Can be predicted and categorized by customer intention. Accordingly, the control module 205 can predict and classify the received customer consultation data into customer consultation data with "insurance claims" as a customer consultation intention. The above-mentioned customer consultation data, customer consultation intention prediction and classification are merely examples, and various customer consultation data can be predicted and classified by various customer consulting intentions.

In addition, the control module 205 can predict and classify the customer consulting intent to be consulted by the customer based on the time-series analysis of the plurality of keywords. The control module 205 can more accurately predict the customer consulting intention of the consultation case through time-series recognition of the keyword. At this time, the prediction of customer counseling intent can be based on probability. The control module 205 can predict and classify the customer consulting intent of the consultation case with the customer consulting intention having the highest probability of matching the recognized keyword. For example, as a result of analyzing existing customer consultation data statistically, keywords such as "MRI", "CT", "inspection", "broken", "Gibbs" can be recognized after the keyword "bone". In this case, when keywords such as "MRI" and "examination" are recognized after the keyword "bone", the consultation case can be predicted as a case where there is a high possibility of "actual cost insurance charge" for the fracture. Therefore, the control module 205 can predict and classify the customer consultation intention of the consultation proposal as "requesting the actual cost insurance inspection fee ". Further, in the above-mentioned example, when the keyword "broken" is recognized after the keyword "bone ", the consultation can be predicted to be a" insurance claim " Accordingly, the control module 205 can predict and classify the customer consultation intention of the consultation request as "insurance claim ". The above-described prediction and classification operations are merely examples, and various customer consultation data can be classified into various customer counseling intentions.

In one aspect of the invention, the control module 205 may provide corresponding feedback based at least in part on the classified customer consultation intent. Such feedback may include information for providing a counseling code to the counseling terminal, information for providing a counseling interface to the counseling terminal, information for providing counseling intention data to the counseling terminal, and insurance-related information of the client to the counseling terminal And the like. For example, if the customer consultation data is predicted and categorized as "claiming " as a customer consultation intention, the control module 205 agent can generate the information needed to process the customer ' have. The information necessary for processing the customer's consultation request of "insurance claim" includes information for providing the customer identification information to the agent terminal, information such as " Information for providing the customer's counseling intention data called "insurance claim " which is the customer counseling intention predicted and classified by the counseling terminal, information for providing the insurance-related information of the customer to the counseling terminal, And information for providing a consultation code for allowing the agent terminal to provide the consultation interface, insurance related information of the customer, and the like. The aforementioned customer consultation intention and feedback are merely examples, and various feedbacks corresponding to various customer consultation intentions can be provided.

In one aspect of the invention, the control module 205 may update the correlation of the keyword with the customer consultation intention or the customer consultation intention and the feedback operation based on the evaluation information on the feedback received from the agent terminal. For example, the control module 205 predicts and classifies "insurance claim" with the intention of customer consultation, and provides the corresponding feedback to the agent terminal 130. However, negative evaluation information on the feedback from the agent terminal 130 If received, the control module 205 may have predicted an incorrect customer counseling intention for the recognized keyword. Accordingly, the control module 205 can update the correlation between the recognized keyword and the customer consultation intention. The control module 205 can update the correlation by correcting the probability that the consultation case is predicted and classified with a specific customer consultation intention when a specific keyword is recognized. In addition, the control module 205 may update the correlation by recording the number of negative feedbacks on the customer's counseling intention by the recognized keyword. The update method described above is merely an example, and an update method capable of supplementing the accuracy of the prediction of the control module 205 in various ways by the evaluation information may be used.

In another aspect of the present invention, the control module 205 groups customer consultation data into counseling provision data units that include counseling history that has been provided from the counselor, extracts keywords from the group customer consultation data, The correlation between the keyword and the consultation providing data can be stored. For example, the control module 205 compares and analyzes the consultation providing data including the consultation record provided by the consultant in response to the consultation request of the customer, The correlation can be extracted. This correlation implies what kind of counseling the counselor provided in response to the client's counseling request. These correlations can be used as training material for counselors, and the accuracy of predictions can be increased in predicting customer counseling intentions. The above-mentioned utilization is merely an example, and it is possible to utilize it by analyzing the correlation between the record of the consultation history provided by the agent and the record of the client's consultation request.

In another aspect of the present invention, the control module 205 groups customer consultation data into consultation data units provided from an agent, extracts keywords from the grouped customer consultation data, and correlates the extracted keywords with consultation providing data You can save the relationship. For example, if an agent talks to a customer and uses an interface related to "pay insurance", this record can be counseling data. The control module 205 may group the customer consultation data of the consultation cases in which the interface related to "insurance payment" has been used into the customer consultation data related to "insurance payment ". The keyword can be extracted from the grouped customer consultation data. For example, in the customer consultation data relating to "insurance payment", these words may be keywords if a plurality of words such as "hurt", "fracture", "insurance money", "examination document" appear repeatedly. The control module 205 can store the correlation between the keyword and the consultation providing data. The word "injured", "fracture", "insurance money", "examination document" and the like are words related to "insurance claim", and the control module 205 determines the correlation between the keyword and the consultation providing data " Can be stored. The correlation between the above-described keyword extraction, the keyword and the consultation providing data is merely an example, and a suitable method of classifying the keywords extracted from the customer consultation data may be included in the present invention.

In addition, the control module 205 may perform control related to the retrieval and storage of the big data in the receiving module 201 and the database 207. [ Such big data may include, for example, various data stored in the call center.

In addition, the control module 205 may determine the collection level of the big data. For example, the collection level of the big data may be classified into three types, i.e., classification of external data and internal data, classification of fixed data and unstructured data, number of predetermined keywords, capacity of data to be collected, Etc.), and the like. In addition, the control module 205 can determine the level of analysis of the big data.

In a further aspect of the present invention, the control module 205 of the server 110 may generate a program associated with an application service covering insurance products. The user terminal 120 or the agent terminal 130 can receive and execute the corresponding application generated by the control module 205 of the server 110. [ The control module 205 may generate a message to install the application if the corresponding application is not installed in the corresponding user terminal or agent terminal and transmit the message through the transmission module 203. [ In addition, the control module 3205 may generate a message to update to a new version of the application if the corresponding application of the user equipment is an old version. In one aspect of the invention, the functions of the control module 205 described above may be implemented by an independent server (e.g., a service platform server and / or an update server, etc.) separate from the server 110.

In one aspect of the present invention, the database 207 may be coupled with other components of the server 110. In addition, the database 207 may include one or a plurality of databases. Optionally or alternatively, a database containing part or all of the information may be located within the user terminal 120 or the agent terminal 130. In addition, the database 207 may exist independently from the server 110. In this case, the database 207 may be capable of communicating with the server 110.

In further aspects of the present invention, the database 207 stores various information needed to provide to the user terminal 120 or the agent terminal 130, .

In an aspect of the present invention, the database 207 may be a flash memory type, a hard disk type, a multimedia card micro type, a card type memory (for example, SD Or XD memory), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read- Memory, a magnetic memory, a magnetic disk, and an optical disk. In a further aspect of the present invention, the server 110 may operate in association with web storage that performs the storage function of the database 207 on the Internet.

The various embodiments described herein may be embodied in a recording medium or storage medium readable by a computer or similar device using, for example, software, hardware, or a combination thereof.

For example, in accordance with a hardware implementation, the embodiments described herein may be implemented as application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs) may be implemented using at least one of field programmable gate arrays, processors, controllers, micro-controllers, microprocessors, and electrical units for performing other functions. The embodiments described herein may be implemented by the control module 205 itself.

In another example, according to a software implementation, embodiments such as the procedures and functions described herein may be implemented with separate software modules. Each of the software modules may perform one or more of the functions and operations described herein. Software code may be implemented in a software application written in a suitable programming language. The software code may be stored in the database 207 and executed by the control module 205.

3 illustrates information or data stored in the database 207 in accordance with one embodiment of the present invention.

3, the database 207 according to an aspect of the present invention includes customer consultation data 310, a keyword 320, a customer consultation intention 330, feedback data 350, consultation provision data 370 ), And customer identification information 390 can be stored. The information or data shown in FIG. 3 is exemplary only, and additional information or data may also be stored in the database 207.

Additionally, in accordance with an aspect of the present invention, the database shown in FIG. 3 may be included in the agent terminal 130. In this case, the above-described data and information may be generated through the control module of the agent terminal 130, or data and information may be received from the server through the receiving module.

The database 207 according to one aspect of the present invention may collect information or data from the external server, the user terminal 120 and / or the agent terminal 130 via the receiving module 201 or directly. As described above, the database 207 (or the receiving module 201 may collect various types of data in various ways).

The information or data stored in the database 207 may include both formatted data, semi-structured data, and unstructured data.

Here, the formatted data refers to formatted data, and may include, for example, data stored in a fixed field. Also, unstructured data refers to unformatted data, and may include various data that can be obtained through a search such as, for example, an image, a moving picture, a text file, and the like. In addition, semi-structured data is not a fixed field but may include XML or HTML text as data including metadata, schema, and the like. Further, the internal data refers to data used by the users of the user terminal 120 or the agent terminal 130, and the external data may refer to data used by users other than the terminals 120 and 130.

The information or data stored in the database 207 may include both document data and non-document data.

Here, the document data may mean data in which the data itself can be understood by human cognitive ability. In addition, non-document data may be meaningful in a single unit, and may mean data that may have hidden meaning through a particular pattern when clustered.

The database 207 in accordance with an aspect of the present invention may be comprised of one or more subcomponents. For example, although not shown, the database 207 may include a form data collection unit, a semi-structured data collection unit, an unstructured data collection unit, an internal data collection unit, an external data collection unit, a document data collection unit, And / or a data integration unit.

In one aspect of the present invention, information or data to be secured among data collected and stored in the database 207 may be encrypted and stored.

Referring to FIG. 3, the customer consultation data 310 may include an ongoing call content in a call center. At this time, the customer consultation data may include the character strings of the customer's voice or the customer's voice in the case of conversation in progress in the call center. For example, when the customer A encodes the voice of the customer A or the voice of the customer A recorded during consultation with the agent through the call center, the character string may be included in the customer consultation data. A customer speaks with a counselor through a call center, saying, "I was hospitalized yesterday and I was hospitalized, but my left arm was broken, but my mom had a premium when she broke a bone in the past. Quot ;, "I have joined ", etc., can be the customer consultation data. In another aspect of the present invention, the consultation contents collected from a voice of customer (VOC) bulletin board or the like may also be customer consultation data. The above-mentioned customer consultation data is merely an example, and may include customer consultation data collected through all roots for which a customer such as the Internet, a telephone, and a face route requests a consultation.

The keyword 320 data may include words related to tasks performed by the call center. For example, in the insurance industry, keywords can be "hurt," "hospitalized," or "claims." These keywords may be stored in advance or may be added by analyzing the customer consultation data 310 in the control module 205 of the server. The control module of the server can extract keywords based at least in part on the number of repetitive occurrence of a specific word or the like in the customer consultation data 310 grouped by the consultation providing data 370. [ The above-described keywords are merely examples, and appropriate keywords relating to work performed by the call center can be included in the database.

The customer consultation intention 330 data may include data relating to the intention to be consulted by the customer, and data on the category in which the customer requests the consultation. The customer consultation intent 330 may be analyzed based at least in part on the keyword recognized in the customer consultation data 310. [ The control module 205 may recognize the keyword in the customer consultation data 310 and classify the customer consultation data 310 according to the customer consultation intent 330 based at least in part on the recognized keyword. For example, "I was hospitalized at the hospital yesterday, and my left arm bones were broken, but I saw that my mom had a premium when she was bones in the past. , The intention to recognize a keyword such as "bumped", "hospitalized", "bone", "premium", and the intention to request consultation of the customer based at least in part on the keyword can be classified and predicted as " have. Further, in one aspect of the present invention, the control module 205 can grasp the context in the customer's consultation contents through the time-series recognition of the keyword described above, and classify and predict the intention of the customer's request for consultation as " have. The above-mentioned customer counseling intention is merely an example, and all kinds of tasks that are processed by the call center counseling may be included in the customer's counseling intent.

The feedback 350 data may include data for an agent to perform consultation at the agent terminal 130 by analyzing the intention of the client to request the consultation. For example, the feedback may include a consultation code for providing the counselor terminal, information for providing the counseling interface to the counseling terminal, information for providing the counseling client's intention data to the counseling terminal, And at least one of information for providing information. The above information is merely an example, and may include appropriate information for processing a consultation matter identified in the consultation contents of the customer.

The consultation providing data 370 may include a consultation history performed by the consultant terminal 130. The details of the consultation may include details of the use of the consultant terminal 130 in order to perform the consultation in response to the consultation request of the customer. For example, in response to a consultation request related to a customer's "insurance claim ", a consultant may inquire the customer about the customer's inquiry, If an interface is used, a history of use of such an interface and the like can be included in the consultation providing data. The above-mentioned consultation providing data is merely an example, and may include details of all operations that can be performed in the agent terminal in response to a consultation request of the client.

The customer identification information 390 may include information that can distinguish the customer from other customers. For example, personal information such as the name, date of birth, sex, etc. of each customer may be included to distinguish the A customer and the B customer. Such customer identification information may be encrypted and stored in accordance with relevant laws and regulations. It may also contain appropriate information to distinguish a particular customer from other customers.

4 schematically illustrates a server operation according to an embodiment of the present invention.

4, the server 110 extracts the keyword 430 by analyzing the customer consultation data 420 in the on-going consultation at the call center 410. As shown in FIG. The server 110 predicts and classifies the customer consultation intent 440 based at least in part on the extracted keywords and provides the corresponding feedback 450 based at least in part on the classified customer consult intent. The agent terminal 130 receives the feedback information from the server and can display (460) it on the agent terminal.

In an aspect of the present invention, the server 110 or the agent terminal 130 may record the voice of the customer when consultation is in progress at the call center 410. [ At this time, when the recorded voice or the recorded voice is converted into a character string, the character string may be the customer consultation data 420. [ For example, in a call made between an agent and a customer at a call center, "I was hospitalized at the hospital yesterday, and my left biceps were broken, but my mom had a premium when she broke a bone in the past. Quot ;, and converts it into a character string and stores it in the customer consultation data 420. [0120] FIG. The above-mentioned customer consultation data is merely an example, and all the consultation contents in non-face-to-face communication can be included in the customer consultation data.

In one aspect of the present invention, the server 110 may analyze the customer consultation data and extract the keywords 430. In this case, the keyword may be stored in the database 207 in advance. Also, the keyword 430 may be extracted by analyzing the customer consultation data previously generated by the server 110 together with the consultation providing data. For example, "I was hospitalized at the hospital yesterday, and my left arm bones were broken, but I saw that my mom had a premium when she was bones in the past. ) Can be analyzed to extract key keywords such as "injured", "hospital", "hospitalized", "bone", "broken", "premium" The extraction of the above-mentioned keywords is only an example, and appropriate keywords can be extracted from various customer consultation data.

In one aspect of the present invention, the server 110 may predict and classify the customer consulting intent 440 based, at least in part, on the recognized keywords. The customer consultation intent 440 may include types of services that the customer can request through the call center, with the intention of requesting the consultation when the customer requests the consultation. For example, when keywords such as "injured", "hospital", "hospitalized", "bone", "broken", "insurance premium", etc. are extracted by analyzing the customer consultation data 420, The control module 205 of the server 110 can know that the intention of the customer to request consultation is "insurance claim ". In this case, the server 110 may classify the customer consultation intention as "insurance claim" for the consultation case. At this time, based on the time series analysis of the plurality of keywords, the customer's counseling intention to be consulted by the customer can be classified. For example, if the keyword "hurt" is recognized and the next recognized keyword is "premium," then the customer's intention to request a consultation is likely to be "claim." However, if keywords such as "hospitalization", "insurance agent", and "examination" are recognized after the keyword "hurt" is recognized, the customer's intention to request the consultation may become " The above-described prediction of the server 110 may be based on a statistical correlation between a keyword and a customer consulting intention. The user or server 110 can analyze the customer consultation data and the consultation history provided by the agent, and store the correlation between the keyword and the customer consultation intention in the database based on the statistics. For example, a keyword extracted from the customer consultation data in a specific customer consultation intention may be a keyword having a high probability of matching the corresponding customer consultation intention. The server 110 can predict and classify the customer consulting intention of the consultation case with the customer consulting intention having the highest probability of being corresponded through keyword recognition and analysis. In addition, the server 110 may classify the customer consultation intent based at least in part on the number of recognitions of keywords corresponding to a particular customer consultation intention. For example, in the case where the keyword related to "insurance claim" is recognized three times and the keyword related to "premium payment" is recognized twice, the customer's intention to request the consultation is more likely to be & The intention of the customer consultation may be classified as "claim". The above-described classification is merely an example, and all kinds of services that an agent can provide can be classified as a customer consulting intention.

In one aspect of the invention, the server 110 may provide a corresponding feedback 450 operation based at least in part on the classified customer consultation intent. The feedback operation includes an operation of providing a counseling code to an agent terminal, an operation of providing a counseling interface to an agent terminal, an operation of providing client counseling intent data to an agent terminal, and an operation of providing an insurance- And may include at least one operation. For example, in the case where the intention of the customer's consultation request is determined as "insurance claim ", the server 110 provides the consultant code for executing the consultation interface related to" Insurance claim ", or to indicate to the agent terminal that the intention of the client's consultation request of the consultation request is" claim claim ", or to provide the agent terminal with insurance such as claim details of the customer, insurance payment details, You can provide code to display related information. The above-described feedback operation is merely exemplary, and appropriate actions may be included in the present invention to assist the agent in performing actions through the agent terminal 130 based, at least in part, on the identified customer's intent to request a consultation.

In one aspect of the present invention, the agent terminal 130 receives the feedback received from the server, and can output the consultation information, the customer consultation intention, the customer consultation interface, the evaluation information input interface for the feedback, . For example, by the feedback provided from the server 110 in the above-described example, the agent terminal 130 can display an interface related to "insurance claim ". These interfaces may include customer identification information, customer subscription product information, customer claims history, a list of documents required for claims, and the amount of insurance claims expected to be paid. The operation of the above-described agent terminal 130 is merely an example, and appropriate information corresponding to the information received from the server 110 may also be provided in the agent terminal 130. [

In one aspect of the present invention, the server 110 may receive evaluation information 470 on feedback from the agent terminal 130. The evaluation information on this feedback may include information such as the accuracy of the customer consultation intention predicted by the server 110 and the evaluation of the appropriateness of the feedback provided in response to the customer consultation intention. For example, in the case where the keywords 430 such as "injured", "hospitalized", "bone", "broken", "insurance premium", "received" are recognized in the consultation data 420, May determine that the customer's intention to request a consultation is "insurance claim" based at least in part on the recognized keyword. At this time, the server 110 may transmit the determined consultation request intention to the agent terminal 130. At this time, the agent can input the "insurance claim" received through the agent terminal 130 to evaluate whether the intention of the customer's counseling is appropriate or inappropriate. If the agent talks to the customer and the customer's intent is "claim," then positive evaluation information can be entered for the feedback provided by the server. However, if the customer's counseling intent was actually a "premium payment complaint," negative feedback information could be entered for the feedback provided by the server. The provision of the above-described evaluation information is merely an example, and the agent terminal 130 can provide appropriate evaluation information in response to information provided by the server 110. [

In one aspect of the present invention, the server 110 can update the correlation of the feedback action with the keyword and the customer consultation intention or the customer consultation intention based on the evaluation information on the feedback received from the agent terminal 130 have. For example, the server 110 recognizes keywords such as "injured", "hospitalized", "bones", "broken", "insurance premium" and the like, Claims Claims "and an interface related to claim insurance are provided by the agent terminal 130, the agent terminal 130 determines that the contents determined by the server 110 are inappropriate in view of the contents of consultation with the customer , And can provide negative evaluation information. The agent terminal 130 can receive the evaluation information from the agent through the user input unit. When such negative evaluation information is received, it is possible to reduce the correlation between the keywords "injured", "hospitalized", "bone", "broken", "premium" Correlation can be updated. In addition, the server 110 judges that the intention of the customer's request for consultation is "insurance claim" to provide feedback such as displaying the customer identification information, the customer's subscription product information, the customer's insurance claim history, The agent terminal 130 can provide the negative evaluation information if the information "customer's family relationship information" is unnecessary or improper. When such negative evaluation information is received, the correlation between the customer consultation intention and the feedback operation can be updated by reducing the correlation between the customer consultation intention called "insurance claim" and the feedback operation "providing the customer's family relationship information & . The operation of updating the correlation between the keyword and the customer consultation intention or the customer consultation intention and the feedback operation is performed by the server 110 itself by correcting the correlation or recording the degree of inconsistency and the number of the correlation, It is possible to easily grasp the case where the prediction is inaccurate. The above-described update operation is merely an example, and appropriate actions for updating the correlation between the keyword and the customer consultation intention or the customer consultation intention and the feedback operation in response to the evaluation information of the agent are included in the present invention.

Figure 5A illustrates an example of a probabilistic mapping of keywords, customer consultation intentions, and feedback in accordance with one embodiment of the present invention.

In one aspect of the present invention, the server 110 receives the customer consultation data 310 from the agent terminal 130, recognizes the keyword 320 in the customer consultation data 310, Based on the time-series analysis, the customer may classify the customer consultation intent to be consulted and provide corresponding feedback based at least in part on the classified customer consultation intent. In this case, the server 110 can predict the customer's counseling intention as the state changes at least partially based on the keyword recognized during the consultation in the initial state.

For example, if the state immediately after the start of consultation in the call center is S1, the state may change based on the keyword recognized in the ongoing consultation. At this time, the server 110 can determine that the ongoing consultation state is S3 when the ongoing consultation is recognized in the S1 state and the keyword B is recognized. When an ongoing consultation is in the S1 state, and the keyword B is recognized, S2, S3, and S4 may be present in the next possible state of consultation. Based on the statistical analysis of the customer consultation data, the server 110 determines that the probability that the next state of the ongoing consultation is S2 is 20%, the probability of S3 is 70%, and the probability of S4 when the ongoing consultation is recognized in the S1 state and the keyword B is recognized It can be determined in advance that it is 10%. If the initial state of the consultation is S1 and the keyword B is recognized, the server 110 may determine S3, which has the highest probability of becoming the next state of the ongoing consultation, to 70%, as the next state. In this way, if a specific keyword is recognized in the current state of consultation, the server can determine the consultation state having the highest probability of becoming the next state in the recognized keyword as the next consultation state. For example, when the consultation is started (S1) and the keyword "injured" is recognized, the status of "premium charging (S3)" is higher than the status related to "payment complaint S2" , The server 110 can determine that the ongoing consultation is in a state related to "premium charge S3 ". The above-described probability for each state is only an example, and according to one aspect of the present invention, an appropriate probability that the server can predict the customer's counseling intent can be used.

Also, according to an embodiment of the present invention, the server 110 may update the correlation between the keyword and the customer consultation intention or the customer consultation intention and the feedback operation based on the evaluation information on the feedback received from the agent terminal 130 update. For example, in the example described above, the server 110 may recognize the keyword B in the initial state S1 (S1) and determine the S3 state and provide feedback related thereto. In this regard, the server 110 can receive negative evaluation information from the agent terminal 130. In this case, when the keyword B is recognized in the S1 state, the server 110 adjusts the probability that S3 becomes the next state from 70% to 50%, and increases the probability that S2 becomes the next state from 20% to 40% have. The above-described updating method is merely an example, and an appropriate method for increasing the accuracy of the prediction of the server 110 can be used.

FIG. 5B illustrates an example of a correlation between a keyword, a customer consultation intention, and feedback according to an embodiment of the present invention.

In one aspect of the present invention, the server 110 receives the customer consultation data 310 from the agent terminal 130, recognizes the keyword 320 in the customer consultation data 310, Based on the time-series analysis, the customer may classify the customer consultation intent to be consulted and provide corresponding feedback based at least in part on the classified customer consultation intent.

In one aspect of the present invention, the server 110 may recognize the keywords 320 from the customer consultation data, and may predict and classify the intent of the customer to consult based on the time series analysis of the recognized plurality of keywords. For example, the server 110 can statistically determine S1 and S2 states after the S1 in the initial consultation state (S1, 510). The server 110 may statistically analyze a plurality of customer consultation data to store correlations of respective states. In this case, if the keyword B is recognized in the ongoing consultation, the ongoing consultation can be in the S3 state (530). The state of the ongoing counseling may be in the S3 state in the S1 state based on the probability, and such probability may follow the manner described with reference to FIG. 5A described above. At this time, the server 110 may provide feedback that can be provided in S3 state. When the state of the server 110 is from S1 to S3, the customer consulting intention in the consultation case is highly predicted by the customer consulting intention related to the S3 state. Accordingly, the server 110 recognizes the keyword and predicts the intention of the customer consulting in consideration of this. If the ongoing consultation is in the S3 state and the keyword E is recognized, the next state of ongoing consultation may be S6. The next state of the ongoing consultation in this way can be the state with the highest probability that the next state can be in the keyword recognized in the current state. If the server 110 recognizes the keyword I in the S6 state, the server 110 can predict and classify the customer consulting intention in the consultation case to P3 according to an aspect of the present invention. In addition, the server 110 may provide feedback that may be provided in the S6 state in accordance with an aspect of the present invention. In addition, the server 110 may provide feedback that may be provided in the customer consultation intention P3 in accordance with an aspect of the present invention.

In addition, the server 110 can predict the intention of customer consultation according to the progress of the state change. In the S1 state, the customer counseling intention can be predicted to be one of P1, P2, P3, and P4. If the B keyword is recognized in the S1 state and becomes S3 state, the customer consulting intention can be predicted to be one of P3 and P4. If the keyword E is recognized in S3 state and becomes S6, the customer consulting intention can be predicted as P3. In the manner described above, the server 110 may determine the customer consultation intention based at least in part on the time-sensitive keyword in the current state of ongoing consultation.

FIG. 6 illustrates an example of a correlation between a keyword, a customer consultation intention, and feedback according to another embodiment of the present invention.

In one aspect of the present invention, the server 110 receives customer consultation data 310 from the agent terminal 130, recognizes the keyword 320 in the customer consultation data 310, To classify the customer consultation intent 330 of that consultation case and provide corresponding feedback based at least in part on the classified customer consultation intent.

In one aspect of the present invention, the server 110 recognizes the keyword 320 from the customer consultation data 310 and predicts and classifies the customer consultation intent 330 of the consultation case based at least in part on the recognized keyword . For example, when the server 110 recognizes the keywords 320 of A, B, C, D, E, and F in the customer consultation data 310, Customer counseling intent can be categorized. The classifying operation may be an operation of classifying the customer consulting intention based at least in part on the number of recognition of keywords corresponding to a specific customer consultation intention. Also, when a specific keyword is recognized for a specific customer counseling intention, a method of excluding a specific customer counseling intention in classifying the customer counseling intent may also be used. A may be associated with a customer consulting intention 330 called P1, P2, P4, B related to a customer consulting intention P2, P3, and C related to a customer consulting intention 330, P2, P5 D can be associated with a customer consulting intention 330, P4, P5, E can be associated with a customer consulting intention 330, P2, P3, and F can be associated with a customer consulting intention P5 have. In this case, keywords A, B, C, and E were recognized for the keyword related to the customer consultation intention P1, keywords A, B, C, and E were recognized for the keyword related to P2, keywords B and E were recognized for the keyword related to P3, , D were recognized, and keywords associated with P5 were recognized as C, D, and F, respectively. In this case, the order of the customer consulting intention in which the keyword 320 is widely recognized is P2 (4 times), P5 (3 times), P4 (2 times), P3 (2 times), P1 (1 time). Accordingly, the server 110 determines that the customer's counseling intention is highly likely to be P2 (610) or P5 (620) having a high recognition frequency of the keyword corresponding to the specific customer consultation intention based at least in part on the recognized keyword , And P2 (610) and P5 (620), respectively. As a method of classifying the customer counseling intention, for example, it is also possible to exclude the keyword C from the P1 customer consultation intention and not to classify it as the P1 customer counseling intention when the keyword C is recognized. The specific customer consulting intention corresponding to the highest probability can be predicted or classified through the count of the number of occurrences of the keywords corresponding to the specific customer consulting intention. The operation of classifying the customer's counseling intention based at least in part on the above-described recognized keywords is merely an example, and appropriate analysis methods such as multiple regression analysis, association analysis, etc., as well as counts of the number of iterations are included in the present invention .

In addition to counting the number of occurrences of keywords corresponding to a specific customer counseling intention, specific keywords may be weighted with respect to a specific customer counseling intention to predict or classify a specific customer counseling intent. In the above example, if the keyword F is recognized and the probability that the customer's counseling intention is P5 is higher than the predetermined level, the server 110 gives a weight to the keyword F for the customer counseling intention P5, As shown in FIG. These weights can be set in advance by statistical methods through analysis of customer consultation data. It can also be updated through the analysis of the server 110 and the evaluation information of the agent terminal. The weighting and counseling intention classification described above is merely an example, and an appropriate method that can match the keyword and client counseling intention by analyzing the collected customer consultation data can be used most appropriately.

In one aspect of the present invention, the server 110 may provide corresponding feedback 350 based at least in part on the classified customer consultation intent. For example, the server 110 may provide corresponding feedback based at least in part on the classified customer consultation intent in the case of classifying the customer consulting intent as P2 (610), P5 (620). Feedbacks F1, F2, and F3 can be corresponded to the customer consulting intention P2. Feedback, F2, and F3 may be corresponded to the customer consulting intention P5. In the above example, F1 is the most likely to be the customer consulting intention, so F1, F2, and F3 corresponding to P2 can be provided as feedback. For example, if P2 is called "insurance claim", F1 can be "insurance claim", F2 is "subscription product information", F3 is "insurance claim list" In addition, the customer consultation intent of "claiming" may be the feedback itself. In addition, F2 and F3 feedback corresponding to P5, which is likely to be a customer consultation intention, can also be provisionally provided. The feedback corresponding to the above-mentioned customer consultation intention is only an example, and various feedback operations corresponding to the customer consultation intention can be performed in the server 110. [

Further, in an aspect of the present invention, the server 110 can update the correlation of the keyword with the customer consultation intention or the customer consultation intention and the feedback operation based on the evaluation information on the feedback received from the agent terminal 130 . "Correlation" may be a probability that a particular customer consultation intention corresponds to a particular keyword. In the above-described example, if the server 110 determines that the customer's counseling intention is P2 based on the keywords A, B, C, and E, and the negative evaluation information is received from the agent terminal 130, The correlation between the keyword and the customer's counseling intention is updated when the keyword F is more likely to correspond to the P5 than the intention P2, or the probability that the customer corresponds to the customer counseling intention P2 You can update the relationship. Further, in the above-described example, the server 110 classifies the customer consulting intention as P2 and provides the F1, F2, and F3 feedback based at least in part on the customer consulting intention, When the information is received, the feedback F1 is excluded from the customer consultation intention P2, the probability of providing the feedback F1 is modified to be low, the feedback F4 is added when the customer consulting intentions P2 and P5 are recognized together, or the feedback F4 is provided We can update the correlation between customer intentions and feedback by increasing the probability of doing so. In addition, an evaluation of the correlation between the keyword and the customer consultation intention or the customer consultation intention and feedback can be calculated and stored. The user or the like can modify the correlation based on the evaluation of the calculated correlation. The update of the correlation described above is merely an example, and the correlation between the keyword and the customer consultation intention or the customer consultation intention and the feedback operation can be updated by various evaluation methods.

7 shows an example of a method of extracting keywords from consultation providing data according to another embodiment of the present invention.

In one aspect of the present invention, the server 110 groups (710, 750) the customer consultation data into counseling provision data units including consultation details that have been provided from the counselor, and generates keywords (717, 719, 757 , 759), and store the correlation between the extracted keyword and the consultation providing data. Such a correlation may mean a probability that specific consultation providing data correspond to extracted keywords. The grouping operation described above may include, in a specific consultation case, matching the consultation providing data in the consultation case with the customer consultation data in the specific consultation case. The consultation providing data may include at least one of the details of the consultation interface used by the agent at the agent terminal according to the contents of the conversation with the client, the consultation history performed at the agent terminal, and the insurance related information of the corresponding client.

For example, if consultation related to "insurance claim" is performed by an agent, there may be a record of the interface executed by the agent at the agent terminal 130 with respect to the consultation performed by the agent. The server 110 can receive counseling provision data including this record from the agent terminal 130. [ The customer consultation data can be grouped (710, 750) based on the consultation providing data. For example, the customer consultation data in the consultation case in which the interface related to the "insurance claim" has been executed in the agent terminal 130 can be grouped on the basis of the consultation providing data called "insurance benefit provision ". Referring to the drawing, there may be a group of customer consultation data 713, 715, 716 based on the A consultation providing data 711. In addition, there may be a group of customer consultation data 753, 755, 756 based on the B consultation providing data. Keywords can be extracted from the customer consultation data based at least in part on the number of occurrences of a specific word in the customer consultation data of each group in the plurality of grouped customer consultation data. For example, in the customer consultation data 713, 715, and 717 based on the A consultation provision data, the word U appears 85 times (717), V appears 70 times (719), W appears 65 times In one case, U with the highest number of iterations can have a high correlation with A counseling provided data. Therefore, U can be a keyword having a high correlation with A counseling provision data. In addition, V and W with high repetition times can also be keywords as meaningful words. The server 110 may store the correlation between the keyword U and the consultation providing data. More specifically, if the number of occurrences of the word " injured "is very high in a plurality of customer consultation data grouped into a group called" insurance claim ", a keyword having a high correlation with " , And the server 110 can store the correlation between the keyword "hit" and the consultation providing data "insurance claim ". In addition, there may be a group of customer consultation data 753, 755, 756 based on the B consultation providing data 751. [ In this case, when the word X appears 54 times (757), Y appears 40 times (759), and Z appears 34 times, X having the highest number of repetitive occurrences can have a high correlation with the B consultation providing data. Therefore, X can be a keyword having a high correlation with the A consultation data. At this time, if the relation between the B consultation providing data and the A consultation providing data is low, X may be a keyword having a low correlation with the A consultation providing data. More specifically, if the number of occurrences of the word "monthly payment" is extremely high in a plurality of customer consultation data grouped into a group "insurance new admission ", the" monthly payment & , And the server 110 may store the correlation between the keyword "monthly payment" and the consultation providing data "insurance new registration ". However, in the case where the correlation between the "new insurance" and the "insurance claim" is low, the keyword "monthly payment" becomes a keyword having a low correlation with the "insurance claim" Claims "can be saved. The above-described keyword extraction method is merely an example, and an appropriate method for extracting the correlation between the keyword and the consultation providing data other than the count of the number of times of repeated occurrence of words can be included in the present invention.

Correlation of the stored keywords with the consultation provision data is such that the server 110 categorizes the customer consultation intent in the customer consultation data and provides the corresponding feedback based at least in part on the classified customer consultation intent Can be used to increase.

FIG. 8 illustrates a method for predicting a customer consultation intention and generating feedback information according to an embodiment of the present invention.

Additional steps other than those shown in FIG. 8 may also be included in one aspect of the present invention, and some of the steps shown in FIG. 8 may be omitted. In addition, at least some of the steps shown in FIG. 8 may be performed by the user terminal 120 and / or the agent terminal 130. [

In one aspect of the present invention, the server 110 may receive 810 the on-going customer consultation data via the receiving module 201. At this time, the on-going customer consultation data may include at least one of a conversation between the agent and the customer performed at the call center, or a string between conversations of the customer's voice to a string.

In one aspect of the present invention, the server 110 may recognize the keyword in the received customer consultation data (820). Since the customer consultation data is atypical data, the server 110 can extract the keyword through the analysis of the big data. At this time, the server 110 can store the keyword in the database in advance, and can use the keyword extracted from the customer consultation data according to another aspect of the present invention. In the server 110, not only the words matching the keyword in the customer consultation data, but also various big data analysis techniques for finding meaning in the unstructured data can be used.

In one aspect of the present invention, the server 110 may classify the customer counseling intent 830 based at least in part on the recognized keyword. The server 110 may classify the customer counseling intention based at least in part on the number of recognitions of the keywords corresponding to the specific customer counseling intention in the customer consultation data as described above. In addition, the server 110 may classify the customer counseling intent to be consulted by the customer based on a time series analysis of a plurality of keywords. For example, if the A keyword 510 and the C keyword 530 are recognized in order, the customer's consulting intention of the consultation request is likely to be P3 and P4. Accordingly, the server 110 can predict and classify the customer counseling intention to P3 and P4. In addition, as described above, the customer's counseling intent can be classified on condition that a specific keyword is not recognized for a specific counseling intent. The above description is only an example, and an appropriate method for classifying the customer consultation intention by recognizing and combining a plurality of keywords may be included in the present invention.

In one aspect of the present invention, the server 110 may provide 840 corresponding feedback based at least in part on the classified customer consultation intent. As described above, the feedback provided by the server 110 includes providing a counseling code to the agent terminal, providing a counseling interface to the counseling terminal, providing client counseling intent data to the counseling terminal, And / or < / RTI > In addition, the server 110 may provide a plurality of feedbacks corresponding to customer consultation intentions.

In one aspect of the invention, the server 110 may receive evaluation information on feedback from the agent terminal 130 (850). The agent terminal 130 can transmit affirmative or negative evaluation information to the server about the client counseling intention classified by the server and the feedback provided by the server. At this time, the agent terminal 130 may receive an input through which the agent evaluates the feedback provided by the server through the user input unit.

In one aspect of the present invention, the server 110 may update the correlation between the keyword and the customer consultation intention or the customer consultation intention and the feedback operation based on the evaluation information received from the agent terminal 130 (860 ). As described above, when negative feedback is given to the customer's counseling intention, the server 110 corrects the correlation or the probability of correspondence between the keyword and the customer's counseling intention to a low level, or stores the negative feedback count on the keyword and the customer counseling intent The user can easily grasp the correlation between the keyword and the customer consulting intention. Also, when negative feedback is received on the feedback as described above, the server 110 may reduce the correlation between the customer's counseling intention and the feedback, store the number of times the customer's counseling intention and negative feedback information about the feedback are received, And the like, can easily grasp the correlation between the keyword and the customer's counseling intention.

FIG. 9 shows a method of extracting keywords from consultation providing data according to another embodiment of the present invention.

9 may also be included in one aspect of the present invention, and the steps of some of the steps shown in FIG. 9 may be omitted.

In an aspect of the present invention, the server 110 may convert the voice of the customer into a character string and store the converted voice as the customer consultation data (910) in the telephone conversation between the agent and the customer in the call center. Also, as described above, the server 110 may store the customer's voice as the customer consultation data. Such customer consultation data can be used as various types of unstructured data analysis methods which can find a specific meaning in unstructured data as unstructured data.

In one aspect of the present invention, the server 110 may group the customer consultation data into consultation providing data units including consultation details that have been provided from the agent (920). The grouping operation may include an operation of matching specific counseling provision data with customer consultation data on which the specific counseling provision data is generated. For example, the customer consultation data of consultation cases in which the function related to "claiming" has been executed can be grouped on the basis of "insurance claims" The consultation providing data may include at least one of information of a consultation interface used by an agent at an agent terminal according to contents of a conversation with a client, consultation details performed at an agent terminal, and insurance related information of the customer. In addition, the grouping operation described above may include an operation of matching specific counseling provision data with customer consultation data in which the specific counseling provision data is generated. For example, in the first customer consultation, when the consultation providing data "insurance claim" occurs, the customer consultation data collected in the first customer consultation can be matched with the consultation providing data "insurance claim".

In one aspect of the present invention, the server 110 may extract 930 keywords from each group of grouped customer consultation data. As described above, the server 110 can extract keywords from the customer consultation data based at least in part on the number of occurrences of a specific word in each group, from a plurality of customer consultation data grouped on the basis of specific consultation providing data .

In an aspect of the present invention, the server 110 may store the correlation between the extracted keyword and the consultation providing data in the database 207 (940). As described above, in the case of a word having a large number of repetitions in the first consultation providing data group, it can be a first keyword having a high correlation with the first consultation providing data. Therefore, in this case, the server 110 may store the correlation between the first keyword and the first consultation provision data. Further, in the case of a word having a large number of repetitions in the second consultation providing data group, the second keyword can be a second keyword having a high correlation with the second consultation providing data, and the correlation between the second consultation providing data and the first consultation providing data A word having a large number of repetitions in the second consultation providing data group may be a second keyword having a low correlation with the first consultation providing data. Therefore, in this case, the server can store a high correlation between the second keyword and the second consultation providing data, and a low correlation between the second keyword and the first consultation providing data. The correlation between the stored keyword and the consultation providing data can be utilized as the probability that the consultation providing data specific to a specific keyword corresponds.

The stored keyword may be a keyword for the server 110 to recognize the keyword in the customer consultation data according to an aspect of the present invention. The correlation between the stored keyword and the counseling provision data is such that the server 110 recognizes the keyword in the customer consultation data and classifies the customer counseling intention based at least in part on the recognized keyword according to an aspect of the present invention, Lt; RTI ID = 0.0 > at least < / RTI > Correlation between stored keywords or keywords and counseling data can also be used as training material for counselor education.

In one or more exemplary implementations, the functions presented herein may be implemented in hardware, software, firmware, or a combination thereof. When implemented in software, the functions may be stored on or transmitted via one or more instructions or code on a computer readable medium. The storage medium may be any general purpose computer or any available medium that can be accessed by a special purpose computer. By way of example, and not limitation, such computer-readable media can comprise any form of computer readable medium, such as RAM, ROM, EEPROM, CD-ROM or other optical disk storage media, magnetic disk storage media or other magnetic storage devices, But not limited to, a general purpose computer, a special purpose computer, a general purpose processor, or any other medium that can be accessed by a particular processor. The discs and discs used here include compact discs (CDs), laser discs, optical discs, DVDs, floppy discs, and Blu-ray discs where disc plays the data magnetically, As shown in FIG. The combinations may also be included within the scope of computer readable media.

Those skilled in the art will appreciate that the various illustrative elements, components, logical blocks, modules, and algorithm steps described above may be implemented as electronic hardware, computer software, or combinations thereof. In order to clarify the interchangeability of hardware and software, various illustrative components, blocks, modules and steps have been described in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. Skilled artisans may implement these functions in varying ways for each particular application, but such implementation decisions are not necessarily outside the scope of the invention.

The various illustrative logical blocks and modules described in connection with the present disclosure may be implemented or performed with a general purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, Or transistor logic, discrete hardware components, or any combination designed to implement the functions described herein. A general purpose processor may be a microprocessor; In an alternative embodiment, such a processor may be an existing processor, controller, microcontroller, or state machine. A processor may be implemented as a combination of computing devices, such as, for example, a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a DSP core, or a combination of such configurations.

For a hardware implementation, the various illustrative logic, logic blocks, and modules of processing units described in connection with the aspects disclosed herein may be implemented with one or more application specific integrated circuits (ASICs), digital signal processors (DSPs) Such as digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), discrete gate or transistor logic, discrete hardware components, general purpose processors, Controllers, microprocessors, other electronic units designed to perform the functions described herein, or a combination thereof. The general purpose processor may be a microprocessor, but, in the alternative, it may be any conventional processor, controller, microcontroller, or state machine. The processor may also be implemented in a combination of computing devices (e.g., a DSP and a microprocessor, a plurality of microprocessors, a combination of one or more microprocessors in conjunction with a DSP core, or any other suitable configuration). Additionally, at least one processor may include one or more modules capable of implementing one or more of the steps and / or operations described herein.

In addition, various aspects or features described herein may be implemented as a method, apparatus, or article of manufacture using standard programming and / or engineering techniques. Moreover, steps and / or operations of a method or algorithm described in connection with the aspects disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. Additionally, in some aspects, steps or acts of a method or algorithm may be present as a machine-readable medium, or as a combination of at least one or any combination of codes or instructions on a computer-readable medium, It can be integrated into computer program stuff. The term article of manufacture as used herein is intended to encompass a computer program accessible from any suitable computer-readable device or medium.

The description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features presented herein.

Claims (15)

A computer-readable medium comprising:
The computer-readable medium stores instructions that, when executed by a computer, cause the computer to perform the following operations:
Converting the consultation voice between the customer terminal and the agent terminal into a character string;
Recognizing a plurality of keywords from the converted character string;
Generating time series connection information of the plurality of keywords based on the recognition time of each of the keywords;
And generating a plurality of pieces of status information corresponding to the first keyword when the first keyword is recognized for the first time and a recognition sequence between the second keyword and the first keyword when the second keyword is recognized after the first keyword, Determining a correlation between the number of recognition of at least one of the second keyword and the first keyword and the consultation intention, and determining a state in which at least a part of the state information of the plurality of state information is removed based on the determined correlation Determining the customer's counseling intent corresponding to the time-series connection information; And
Generating feedback information for the customer on the basis of the determined counseling intention, the determined counseling intent includes at least one category of an insurance claim, an insurance money payment, a subscription insurance product inquiry, a customer complaint, and an insurance product subscription inquiry, Wherein the feedback information on the customer includes information for outputting a user interface (UI) for an agent terminal for assisting an agent's consulting work on a category corresponding to the determined counseling intention;
/ RTI >
Computer-readable medium. The method according to claim 1,
The time-series connection information includes:
Wherein the information includes information on the types of the plurality of keywords, the recognition order, and the number of recognition times.
Computer-readable medium. delete The method according to claim 1,
Providing the feedback information to the agent terminal;
Receiving evaluation information on the feedback information from the agent terminal; And
Updating the correlation between the time series connection information and the consultation intention or the correlation between the consultation intention and the feedback information based on the received evaluation information;
≪ / RTI >
Computer-readable medium. The method according to claim 1,
Displaying in real time a consulting interface including the feedback information in a state where the customer terminal and the agent terminal are in a call connection state to the agent terminal;
≪ / RTI >
Computer-readable medium. 1. A server for providing a consulting interface predicted from customer ' s consultation data,
The server comprises one or more processors and a memory configured to store instructions executable by the one or more processors,
The processor comprising:
A consultation voice between the customer terminal and the agent terminal is converted into a character string, a plurality of keywords are recognized from the converted character string, time series connection information of the plurality of keywords is generated based on the recognition time of each keyword, Determine the customer's counseling intent corresponding to the thermal connection information, and generate feedback information for the customer based on the determined counseling intent,
The method of claim 1, wherein the determining of the customer's intention comprises: generating a plurality of status information corresponding to the first keyword if the first keyword is first recognized; and if the second keyword is recognized after the first keyword, Determining a correlation between at least one of a keyword and the first keyword, a recognition rate of at least one of the second keyword and the first keyword, and a consulting intention, and determining the correlation between the plurality of status information Is removed in a manner such that at least some of the state information is removed, and
Wherein the determined counseling intent includes at least one of an insurance claim, an insurance money payment, an insurance product inquiry, a customer complaint, and an insurance product subscription inquiry, and the feedback information on the customer includes at least one of category (UI) for an agent terminal for assisting an agent's consulting work,
A server to provide a consultation interface. delete delete delete delete delete delete delete delete delete

Images