KR102643045B1 - Chatbot service providing system for analyzing by time and method thereof - Google Patents

Abstract

A chatbot service provision system and method that can be analyzed at each point in time are provided. The above method is a method of providing a chatbot service that can be analyzed at each point in time performed by a chatbot management server, comprising: receiving conversation data including a query from a user terminal through a chatbot service; Analyzing conversation data including the query to generate analysis data and classifying a conversation model; Selecting a chatbot terminal corresponding to the classified conversation model; and generating conversation data containing an answer corresponding to the query by the chatbot terminal searching and extracting data stored in learning data. may include.

Description

Chatbot service provision system and method that can be analyzed by time point {CHATBOT SERVICE PROVIDING SYSTEM FOR ANALYZING BY TIME AND METHOD THEREOF}

The present invention relates to a chatbot service provision system and method that can be analyzed by time point, and in particular, by learning data by time point in consideration of ambiguous data and giving the concept of time, information can be provided to the user by time point. This relates to a chatbot service provision system and method that can be analyzed.

In general, CHATBOT is a technology that uses natural language processing to provide services between computers and people as if they were talking to each other, as if talking to a person without using an interface such as the web or an application. It is a technology that allows you to obtain information.

For example, it allows machines to handle services such as telling the weather, movie tickets, and restaurant reservations as if they were talking to a human.

In addition, chatbots have become a service at the level of providing predetermined answers according to pre-entered algorithms, but as natural language analysis and processing technology develops along with big data processing technology, it is providing optimal answers that take into account various variables.

In other words, chatbot services have made rapid progress with big data analysis, machine learning, and natural language processing technology to provide conversation services by inferring the correct answer and predicting the next question based on analysis of the user's conversation content. Furthermore, recently, simple It is expanding beyond conversation services to include processing various operations such as shopping and payments.

The matters described as background technology above are only for the purpose of improving understanding of the background of the present invention, and should not be taken as an acknowledgment that they correspond to prior art already known to those skilled in the art.

Republic of Korea Patent Publication No. 10-2018-0098455

The problem to be solved by the present invention is to provide a chatbot service provision system and method that can be analyzed at each point in time.

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

A method of providing a chatbot service that can be analyzed by time point according to an embodiment of the present invention to solve the above-described problem is a method of providing a chatbot service that can be analyzed by time point performed by a chatbot management server. Receiving conversation data including a query; Analyzing conversation data including the query to generate analysis data and classifying a conversation model; Selecting a chatbot terminal corresponding to the classified conversation model; and generating conversation data containing an answer corresponding to the query by the chatbot terminal searching and extracting data stored in learning data. may include.

In one embodiment of the present invention, the step of classifying the conversation model may include classifying the conversation model by extracting and analyzing morphemes, verb phrases, and content words included in conversation data including the query.

In one embodiment of the present invention, the step of classifying the conversation model includes automatically performing typo correction corresponding to the extracted morpheme, the verb phrase, and the content word, and through the typo inspection. When a typo is searched, it is determined whether the searched typo is an actual typo in response to the standard data, and if it is an actual typo, the typo can be corrected in response to the standard data.

In one embodiment of the present invention, the step of classifying the conversation model includes automatically performing typo correction corresponding to the extracted morpheme, the verb phrase, and the content word, and through the typo inspection. When a typo is searched, it may include determining whether the searched typo is an actual typo in response to the reference data, and determining whether or not the typo is misanalyzed through typo analysis when it is not an actual typo.

In one embodiment of the present invention, if the typo is a misanalysis through the typo analysis, the morpheme, the verb phrase, and the content word can be re-extracted to generate the analysis data.

In one embodiment of the present invention, generating the learning data through repeated learning based on big data; may include.

In one embodiment of the present invention, generating the learning data includes collecting the big data; Converting the collected big data into classifiable data; extracting and removing ambiguous data from the classified big data; and iteratively learning time-specific data using the big data from which the ambiguous data has been removed.

In one embodiment of the present invention, the step of converting the big data into classifiable data includes converting the big data into machine generated data, semi-generated data, and user-processed data. Generated Data, Enterprise Specialized Data, Enterprise Normalized Data, and Enterprise Specialized Vocabulary Data.

In one embodiment of the present invention, the step of repeatedly learning the time-specific data includes data by type including Realtime, Daily, Monthly, Quarterly, and Yearly. can be learned repeatedly.

In one embodiment of the present invention, before converting the big data into classifiable data, preprocessing the big data; It may further include.

In one embodiment of the present invention, the step of preprocessing the big data involves tokenizing the words included in the big data using a space removal filter and a special character removal filter so that the words included in the big data are divided into minimum units of meaningful words. Steps to perform; After the tokenization operation is completed, performing a refining operation to highlight the meaning of the remaining words by removing noise data for words with low frequency of occurrence or words that are repeated a large number of times included in the big data; and normalizing the big data after the refining operation is completed; may include.

In one embodiment of the present invention, updating the learning data in real time in response to conversation data including the answer; It may further include.

In one embodiment of the present invention, inspecting the learning data; It may further include.

In addition, a method of providing a chatbot service that can be analyzed at each point in time according to another embodiment of the present invention to solve the above-described problem involves a chatbot terminal receiving conversation data containing user information and queries from a user terminal through a chatbot service. step; generating analysis data by the chatbot terminal analyzing conversation data including the query based on the user information; And a step of the chatbot terminal searching learning data using the analysis data based on the user information and generating conversation data including an answer; may include.

In one embodiment of the present invention, the chatbot management server generates the learning data by iteratively learning based on big data; may include.

In one embodiment of the present invention, the step of generating the learning data includes: the chatbot management server collecting big data; The chatbot management server preprocesses the big data and converts it into classifiable data; The chatbot management server divides the big data into Machine Generated Data, Semi Generated Data, Human Generated Data, Enterprise Specialized Data, and Enterprise General Data ( Classifying by task, individual, and company into Enterprise Normalized Data and Enterprise Specialized Vocabulary Data; And the chatbot management server generates the learning data by repeatedly learning the classified data by type, including Realtime, Daily, Monthly, Quarterly, and Yearly. step; may include.

In addition, a chatbot service providing system that can be analyzed at each point in time according to another embodiment of the present invention to solve the above-described problem is a chatbot that provides real-time chatbot services to user terminals through chatbot terminals selected by situation, field, and individual. Management server; It includes, the chatbot management server analyzes conversation data containing a query received from the user terminal, generates analysis data, classifies the conversation model, and then creates the conversation model from the learning data generated by considering ambiguous data. Data corresponding to can be extracted to generate conversation data including answers and transmitted to the user terminal through the chatbot terminal.

In one embodiment of the present invention, the chatbot management server preprocesses the big data collected in consideration of the ambiguous data and converts it into classifiable data, and converts the preprocessed data into machine generated data, semi-automatic data. Processed data (Semi Generated Data), User processed data (Human Generated Data), Enterprise Specialized Data, Enterprise Normalized Data, and Enterprise Specialized Vocabulary Data are used for each task and individual. , the learning data can be generated by classifying by company and repeatedly learning the classified data in real time, daily, monthly, quarterly, and yearly.

In addition, a chatbot service providing system capable of analysis at each point in time according to another embodiment of the present invention to solve the above-described problem includes a chatbot management server that provides a real-time chatbot service to a user terminal; It includes, and the chatbot management server analyzes conversation data containing user information and queries received from the user terminal, generates analysis data, classifies the conversation model, and then generates analysis data from the learning data generated by considering ambiguous data. Data corresponding to the conversation model and the user information is extracted to generate conversation data including an answer and transmitted to the user terminal, and the chatbot management server extracts morphemes, verb phrases, and verb phrases included in the conversation data including the query. The analysis data can be generated by extracting content words and extracting and analyzing them by situation, field, and individual.

In one embodiment of the present invention, the chatbot management server preprocesses the big data collected in consideration of the ambiguous data and converts it into classifiable data, and converts the preprocessed data into machine generated data, semi-automatic data. Processed data (Semi Generated Data), User processed data (Human Generated Data), Enterprise Specialized Data, Enterprise Normalized Data, and Enterprise Specialized Vocabulary Data are used for each task and individual. , the learning data can be generated by classifying by company and repeatedly learning the classified data in real time, daily, monthly, quarterly, and yearly.

In addition, the program according to one embodiment of the present invention is combined with a computer, which is hardware, and is stored in a computer-readable recording medium so as to perform the method of providing a chatbot service that can be analyzed for each point in time.

Other specific details of the invention are included in the detailed description and drawings.

According to the present invention, by using a chatbot service to learn data at each point in time, taking ambiguous data into account, and giving a concept of time, it is possible to provide information to the user at each point in time while minimizing the cost incurred during the consultation process.

According to the present invention, by clearly understanding the query in consideration of ambiguous data and providing information analyzed by viewpoint to the user in a conversational manner, errors and search times for ambiguous data are minimized. Convenience and reliability can be improved while respecting user diversity.

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

1 is a diagram illustrating a chatbot service provision system capable of analysis at each point in time according to an embodiment of the present invention.
FIG. 2 is a detailed diagram for explaining the chatbot management server shown in FIG. 1.
Figure 3 is a diagram illustrating a method of providing a chatbot service that can be analyzed by time point, which is an embodiment of the present invention.
FIG. 4 is a diagram for explaining the step of generating conversation data shown in FIG. 3.
FIG. 5 is a diagram for explaining the steps of generating learning data shown in FIG. 3.

The advantages and features of the present invention and methods for achieving them will become clear by referring to the embodiments described in detail below along with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below and may be implemented in various different forms. The present embodiments are merely provided to ensure that the disclosure of the present invention is complete and to provide a general understanding of the technical field to which the present invention pertains. It is provided to fully inform the skilled person of the scope of the present invention, and the present invention is only defined by the scope of the claims.

The terminology used herein is for describing embodiments and is not intended to limit the invention. As used herein, singular forms also include plural forms, unless specifically stated otherwise in the context. As used in the specification, “comprises” and/or “comprising” does not exclude the presence or addition of one or more other elements in addition to the mentioned elements. Like reference numerals refer to like elements throughout the specification, and “and/or” includes each and every combination of one or more of the referenced elements. Although “first”, “second”, etc. are used to describe various components, these components are of course not limited by these terms. These terms are merely used to distinguish one component from another. Therefore, it goes without saying that the first component mentioned below may also be a second component within the technical spirit of the present invention.

Unless otherwise defined, all terms (including technical and scientific terms) used in this specification may be used with meanings commonly understood by those skilled in the art to which the present invention pertains. Additionally, terms defined in commonly used dictionaries are not interpreted ideally or excessively unless clearly specifically defined.

Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings.

FIG. 1 is a diagram illustrating a chatbot service provision system capable of analysis at each point in time according to an embodiment of the present invention, and FIG. 2 is a detailed diagram illustrating the chatbot management server shown in FIG. 1.

As shown in Figures 1 and 2, the chatbot service providing system 1 that can be analyzed by time point, which is an embodiment of the present invention, includes a user terminal 10, a chatbot terminal 20, and a chatbot management server 30. You can.

In this embodiment, the first to nth user terminals 10-1, 10-2, ...10-n (hereinafter referred to as the user terminal 10) and the first to nth chatbot terminals 20-1, 20-2, ...20-n) (hereinafter referred to as chatbot terminal 20) may be disclosed in plural numbers, but is not limited thereto. For example, the chatbot terminal 20 may be implemented in a form in which a plurality of components are integrated with each other in an actual material environment.

Here, the user terminal 10, chatbot terminal 20, and chatbot management server 30 can transmit and receive data in real-time synchronization using a wireless communication network. The wireless communication network may support various long-distance communication methods, such as Wireless LAN (WLAN), DLNA (Digital Living Network Alliance), Wibro (Wireless Broadband: Wibro), and Wimax (World Interoperability for Microwave Access: Wimax). ), GSM (Global System for Mobile communication), CDMA (Code Division Multi Access), CDMA2000 (Code Division Multi Access 2000), EV-DO (Enhanced Voice-Data Optimized or Enhanced Voice-Data Only), WCDMA (Wideband CDMA) , HSDPA (High Speed Downlink Packet Access), HSUPA (High Speed Uplink Packet Access), IEEE 802.16, Long Term Evolution (LTE), LTEA (Long Term Evolution-Advanced), broadband wireless mobile communication service (Wireless Mobile) Broadband Service: Various communication methods such as WMBS (WMBS), BLE (Bluetooth Low Energy), Zigbee, RF (Radio Frequency), LoRa (Long Range), etc. may be applied, but are not limited to these and include various widely known wireless or mobile communications. method may be applied.

The user terminal 10 can execute the chatbot service provided by the chatbot management server 30 through communication with the chatbot terminal 20.

Specifically, the user terminal 10 can transmit and receive conversation data with the chatbot terminal 20 in real time through the chatbot service.

For example, when the user terminal 10 transmits conversation data including the query content to the chatbot terminal 20, it receives conversation data including the answer content corresponding to the query content, or conversation data including the query content. When received from the chatbot terminal 20, conversation data including answer content corresponding to the inquiry content can be transmitted to the chatbot terminal 20. In other words, the user terminal 10 can operate with the chatbot terminal 20 and users who can ask/answer questions by task, individual, or company, or users belonging to the company.

To this end, the user terminal 10 can generate conversation data including questions/answers using a separate input tool. For example, the user terminal 10 asks/asks questions using at least one of a microphone, touch screen, mouse, microphone, keyboard, or camera. Conversation data including answers can be generated, but is not limited to this. At this time, the conversation data may consist of text or voice, but is not limited to this and may also consist of videos or images.

The user terminal 10 can receive conversation data visually and audibly separately or simultaneously. For example, it may include a display that can output symbols, letters, numbers, etc. on the screen depending on the operating state, a lamp that outputs color changes or blinks, or a speaker that outputs audio.

The user terminal 10 may advance membership registration to the chatbot management server 30 using user information in order to perform the chatbot service with the chatbot terminal 20, but is not limited to this. Here, user information may include name, gender, age, contact information, and occupation, but is not limited to this, and information about the user's current situation may also be entered.

Depending on the embodiment, the user terminal 10 may generate a feedback signal for the chatbot service after or while using the chatbot service and transmit it to the chatbot terminal 20. In contrast, the user terminal 10 may transmit the chatbot service before using it. At this time, the feedback signal may include information such as satisfaction, accuracy, and reliability of response data for each task, individual, and company regarding the chatbot service, but is not limited to this.

Depending on the embodiment, the user terminal 10 may receive a feedback control signal corresponding to the feedback signal from the chatbot terminal 20 before, during, or after using the chatbot service. At this time, the feedback control signal may include improved or updated information on the chatbot service corresponding to the feedback signal. Here, the feedback control signal may include event information. Event information may include advertisements or information about discounts or events for chatbot services, but is not limited to this.

Depending on the embodiment, the user terminal 10 may transmit and receive conversation data with the chatbot management server 30 in real time through the chatbot service.

As such, the user terminal 10 may include various portable electronic communication devices that support communication with the chatbot terminal 20 and the chatbot management server 30. For example, the user terminal 10 is a separate smart device, such as a smart phone, a personal digital assistant (PDA), a tablet, a wearable device, a smartwatch, It may include various portable terminals such as glass-type terminals (including Smart Glass, HMD (Head Mounted Display), etc.) and various IoT (Internet of Things) terminals, but in contrast, non-portable desktop computers and It may include electronic communication devices such as workstation computers.

In addition, the user terminal 10 can operate using an application program (application program or application) in the present disclosure, and such application program can be downloaded from an external server or chatbot management server 30 through wireless communication. You can.

The chatbot terminal 20 is a terminal that provides a chatbot service with the user terminal 10. When receiving the conversation data included from the user terminal 10, the chatbot terminal 20 sends the conversation data containing the answer corresponding to the inquiry to the user terminal 10. ) can be transmitted.

In this embodiment, the chatbot terminal 20 can be classified by situation, field, and individual and provide the corresponding chatbot service to the user terminal 10.

For example, when the conversation data with the user terminal 10 is situation-specific data, the first chatbot terminal 20-1 capable of searching by situation performs a chat service, and the conversation data with the user terminal 10 is field-specific. In the case of data, the second chatbot terminal (20-2) capable of searching by field provides a chat service, and if the conversation data with the user terminal 10 is personal data, the nth chatbot terminal (20-n) capable of searching by individual ) can provide a chat service, but is not limited to this, and each chatbot terminal 20 can provide a chatbot service for each situation, field, and individual.

Such a chatbot terminal 20 may include various portable electronic communication devices that support communication with the user terminal 10 and the chatbot management server 30. For example, as separate smart devices, smart phones, PDAs (Personal Digital Assistants), tablets, wearable devices (e.g., smartwatches, glass-type terminals) It may include, but is not limited to, various terminals such as (including Smart Glass, HMD (Head Mounted Display), etc.) and various IoT (Internet of Things) terminals.

Referring to FIG. 2, the chatbot management server 30 may include a data transmission/reception unit 32, a database unit 34, a monitoring unit 36, and a management control unit 38.

The data transmitting and receiving unit 32 can transmit and receive conversation data including questions/answers with the user terminal 10 or the chatbot terminal 20.

Additionally, the data transmitting and receiving unit 32 can receive user information from the user terminal 10 or the chatbot terminal 20.

Depending on the embodiment, the data transmitting and receiving unit 32 receives a feedback signal from the user terminal 10 or the chatbot terminal 20, and sends a feedback control signal corresponding to the feedback signal to the user terminal 10 or the chatbot terminal 20. It can be sent to .

The database unit 34 can store data transmitted and received from the user terminal 10 or the chatbot terminal 20 through a wireless communication network.

The database unit 34 can store data supporting various functions of the chatbot management server 30. The database unit 34 can store a number of application programs (application programs or applications) running on the chatbot management server 30, data for the operation of the chatbot management server 30, and commands. At least some of these applications may be downloaded from an external server through wireless communication.

The monitoring unit 36 monitors the operating state of the user terminal 10 due to user manipulation, the operating state of the chatbot terminal 20, the operating state of the chatbot management server 30, and the user terminal 10 or chatbot terminal 20. Data transmitted and received between and the chatbot management server 30 can be monitored through the screen. In other words, by checking the usage status of the user terminal 10 or the chatbot terminal 20 in real time, it is possible to conveniently use the user and provide the user with greater confidence.

The management control unit 38 executes a chatbot service with the user terminal 10 or the chatbot terminal 20, analyzes the conversation data received from the user terminal 10, generates analysis data, classifies the conversation model, and classifies the conversation model. Data corresponding to the conversation model can be extracted from the learning data generated by considering the enemy data, generate conversation data, and transmit it to the user terminal 10 or chatbot terminal 20. At this time, the management control unit 38 has started to generate analysis data using all conversation data received from the user terminal 10 or the chatbot terminal 20, but is not limited to this and can also be generated using user information. .

Specifically, the management control unit 38 can classify the conversation model by extracting morphemes, verb phrases, and content words included in the conversation data, extracting and analyzing them by situation, field, and individual, respectively, and generating analysis data.

More specifically, the management control unit 38 performs typo checking on the extracted morphemes, verb phrases, and content words in response to standards, and when typos occur, the typos are corrected in response to the standard data to generate analysis data. . Here, the standard data may include, but is not limited to, spelling and standard language determined by the National Institute of the Korean Language.

In contrast, the management control unit 38 performs typo checking on the extracted morphemes, verb phrases, and content words in response to the standard data, and if no typos occur, determines whether or not there is a misanalysis through typo analysis, and determines whether there is a misanalysis. Analysis data can be generated by re-extracting morphemes, verb phrases, and content words. However, if no typos occur, the management control unit 38 determines whether or not there are misanalyses through typo analysis, and if there are no misanalyses, it can generate analysis data using the extracted morphemes, verb phrases, and content words. Accordingly, by clarifying typos and generating analysis data by extracting and analyzing morphemes, verb phrases, and content words, the intention of the query contained in the conversation data can be clearly identified and an accurate answer with a high degree of understanding can be delivered to the user.

Depending on the embodiment, the management control unit 38 may analyze conversation data by situation, field, and individual based on user information to generate analysis data and classify the conversation model.

At this time, the management control unit 38 may analyze the conversation data and preprocess the conversation data before generating analysis data.

Specifically, the management control unit 38 may preprocess the conversation data by sequentially performing tokenization, cleaning, and normalization so that the conversation data is accurately recognized. Here, if the conversation data is voice, image, or video, it can be converted into text.

Specifically, the management control unit 38 can perform a tokenization operation to filter conversation data consisting of text. That is, the management control unit 38 can perform tokenization using a space removal filter and a special character removal filter so that words included in the text are divided into minimum units of meaningful words.

The management control unit 38 may perform a refinement operation to remove noise data included in conversation data for which tokenization has been completed. That is, the management control unit 38 may perform a refining operation to remove noise data about words that appear less frequently or words that are repeated a lot so that the meaning of the remaining words is highlighted.

Depending on the embodiment, the management control unit 38 may quantify the noise data and determine whether to delete it in response to the user information of the user terminal 10.

The management control unit 38 can normalize text data for which the refinement process has been completed.

Additionally, the management control unit 38 can generate conversation data by searching data stored in the learning data and extracting data corresponding to the conversation model.

In addition, the management control unit 38 can generate learning data by repeatedly learning data using deep learning techniques or machine learning techniques based on big data. Here, big data may include various types of data such as text, voice, video, and image.

Specifically, the management control unit 38 can classify the collected data by task, individual, and company. That is, the management control unit 38 can preprocess the collected data and convert it into data that can be classified.

For example, referring to Table 1, the management control unit 38 divides the pre-processed data into machine generated data, semi-automatically processed data, human generated data, and company-specific data. Data can be classified by type, including Enterprise Specialized Data, Enterprise Normalized Data, and Enterprise Specialized Vocabulary Data, by job, individual, and company.

category explanation machine processing data
(Machine Generated Data) Real-time data automatically generated by machines Semi-automatically processed data
(Semi Generated Data) Arithmetic calculation formula and artificially set target data User Processed Data
(Human Generated Data) Data that is artificially created, judged, and defined Company-specific data
(Enterprise Specialized Data) Data limited to specific companies enterprise universal data
(Enterprise Normalized Data) Universal data regardless of company classification Company-specific vocabulary data
(Enterprise Specialized Vocabulary Data) Vocabulary data used only by specific companies

In other words, the management control unit 38 converts the collected data into machine-processed data such as real-time data automatically generated by the machine, such as serial numbers of parts included in the product serial number and automatic evaluation values for each shipment. Can be classified.

Among the collected data, the management control unit 38 includes additional arithmetic calculation formulas and target data artificially set by the user, such as matters related to KPI (Key Performance Indicato), employees achieving monthly sales goals, and departments achieving monthly production goals. , data can be classified into semi-automatically processed data such as this month's expense claim preliminary settlement results, deposits compared to company expenses, and inventory status.

Among the collected data, the management control unit 38 can artificially classify the data into user-processed data such as data created, judged, and defined by users, such as production plans, daily business reports, meeting attendee lists, and operating profits. .

The management control unit 38 may classify the collected data into data limited to a specific company, for example, company-specific data such as automobile production and automobile consumption.

The management control unit 38 may classify the collected data into general data that does not distinguish between companies, for example, business trip expenses, overtime meals, etc.

The management control unit 38 may classify the collected data into general data used only by a specific company, for example, corporate data such as the sweetener trehalose.

Additionally, the management control unit 38 can extract and remove ambiguous data from data classified based on user information.

Specifically, the management control unit 38 may remove ambiguous data by considering homonyms, polysemy, metaphors, vocabulary, and sentences based on user information.

For example, the management control unit 38 asks, “How much is our company’s operating profit this year?” based on user information. In the query 'this year' and 'operating profit', ambiguous data can be extracted. In other words, this year's standard follows the previous year's standard until March of the following year in terms of accounting standards, and this year based on user information refers to operating profit from January 1 to the present, so ambiguous data is provided based on user information. can be deleted. Accordingly, by removing ambiguous data, the user's trust can be improved by more clearly generating and delivering an answer corresponding to the user's question intention.

In addition, the management control unit 38 can divide the data from which ambiguous data has been removed by time and repeatedly learn the time-specific data to generate learning data for each task, individual, and company.

For example, as shown in Table 2, learn by task, individual, and company by repeatedly learning data by type, including Realtime, Daily, Monthly, Quarterly, and Yearly. Data can be generated.

category explanation Realtime Data learned in real time Daily Data learned every day Monthly Data learned every month Quarterly Data learned quarterly Yearly Data learned by year

In other words, the management control unit 38 can generate learning data by repeatedly learning in real time data that needs to be learned in real time among the collected data, for example, data such as whether or not a product produced in one hour has a defect.

The management control unit 38 can generate learning data by repeatedly learning data that needs to be learned every day among the collected data, such as daily newspaper articles mentioning the company, newspaper articles related to competitors, and economic news.

The management control unit 38 can generate learning data by repeatedly learning data that needs to be learned every month among the collected data, such as monthly reporting details or monthly KPI measurement for each task, individual, or company.

The management control unit 38 may generate learning data by repeatedly learning data learned quarterly among the collected data, for example, data such as company regulations that are temporarily revised each quarter or quarterly sales settlement data.

The management control unit 38 may generate learning data by repeatedly learning data learned annually, such as annual financial statements, annual settlement reports, and annual research results reports, among the collected data.

Additionally, the management control unit 38 can update the learning data in real time by considering conversation data including answers.

Depending on the embodiment, the management control unit 38 may further improve answer similarity by generating learning data through a review process using an external search engine, experts, etc.

Additionally, the management control unit 38 can generate a feedback control signal that can improve the chatbot service in response to the feedback signal received from the user terminal 10.

Depending on the embodiment, the management control unit 38 may generate a feedback control signal containing event information to increase the use of the chatbot service of the user terminal 10 or the chatbot terminal 20.

Such a chatbot management server 30 may be implemented by hardware circuits (e.g., CMOS-based logic circuits), firmware, software, or a combination thereof. For example, it can be implemented using transistors, logic gates, and electronic circuits in the form of various electrical structures.

The operation of the chatbot service providing system that can be analyzed at each point in time according to an embodiment of the present invention having this structure is as follows. Figure 3 is a diagram illustrating a method of providing a chatbot service that can be analyzed by viewpoint, which is an embodiment of the present invention, Figure 4 is a diagram illustrating the step of generating conversation data shown in Figure 3, and Figure 5 is a diagram illustrating Figure 3 This is a diagram to explain the steps for generating learning data shown in .

First, the chatbot management server 30 may execute the chatbot service with the user terminal 10 requesting the chatbot service through the chatbot terminal 20, but may not be limited to this.

Here, the chatbot management server 30 can receive user information input from the user terminal 10. At this time, the user information may include name, gender, age, contact information, and occupation, but is not limited to this, and information about the user's current situation may also be entered.

Additionally, in order to run the chatbot service, the user terminal 10 may advance membership registration by entering user information, but is not limited to this.

Next, the chatbot management server 30 can receive conversation data including questions/answers from the user terminal 10 (S10).

Next, the chatbot management server 30 can analyze the conversation data received from the user terminal 10, generate analysis data, and classify the conversation model (S12).

Specifically, referring to FIG. 4, the chatbot management server 30 may preprocess conversation data received from the user terminal 10 (S100).

Next, the chatbot management server 30 can extract and analyze morphemes, verb phrases, and content words included in the conversation data from the preprocessed conversation data (S110). At this time, the chatbot management server 30 can classify the conversation model by analyzing conversation data by situation, field, and individual based on user information and generating analysis data.

Next, the chatbot management server 30 can automatically check for typos in the extracted morphemes, verb phrases, and content words (S120).

Specifically, when a typo is found through typo inspection (S130), the chatbot management server 30 can determine whether the searched typo is an actual typo in response to the standard data.

If the searched typo is an actual typo (S140), the chatbot management server 30 can generate analysis data by correcting the typo in response to the standard data (S150, S160). Here, the standard data may include, but is not limited to, spelling and standard language determined by the National Institute of the Korean Language.

On the other hand, if the searched typo is not an actual typo (S140), the chatbot management server 30 can determine whether the typo is analyzed through typo analysis.

If the typo is a misanalysis through typo analysis (S170), the chatbot management server 30 can generate analysis data by re-extracting the morpheme, verb phrase, and content word.

Meanwhile, if the typo is not a misanalysis through typo analysis (S170), the chatbot management server 30 can generate analysis data using the extracted morphemes, verb phrases, and content words.

Additionally, if the searched typo is not an actual typo (S130), the chatbot management server 30 can generate analysis data using the extracted morphemes, verb phrases, and content words.

Next, the chatbot management server 30 can select the chatbot terminal 20 corresponding to the conversation model classified based on user information (S14).

In other words, the chatbot terminal 20 can be classified by situation, field, and individual and provide the corresponding chatbot service to the user terminal 10.

Next, the chatbot terminal 20 can generate conversation data by searching data stored in the learning data and extracting data corresponding to the conversation model (S16, S18).

Here, the learning data can be generated by the chatbot management server 30 repeatedly learning data using deep learning techniques or machine learning techniques based on big data.

For example, as shown in FIG. 5, the chatbot management server 30 can collect big data (S200).

Specifically, the chatbot management server 30 can collect various data by job, individual, and company based on big data. Here, big data may include various forms of data such as text, voice, video, and image.

Next, the chatbot management server 30 can preprocess the collected data and convert it into data that can be classified (S210).

For example, the chatbot management server 30 may preprocess conversation data by sequentially performing tokenization, cleaning, and normalization so that the conversation data is accurately recognized. Here, if the conversation data is voice, image, or video, it can be converted into text.

For example, the chatbot management server 30 can perform tokenization using a space removal filter and a special character removal filter so that words included in the text are divided into the minimum unit of meaningful words.

In addition, after the tokenization process is completed, the chatbot management server 30 performs a refinement process to highlight the meaning of the remaining words by removing noise data about words that appear less frequently or are repeated a lot in the conversation data. You can. At this time, the chatbot management server 30 can quantify the noise data and determine whether to delete it in response to the user information of the user terminal 10.

Additionally, the chatbot management server 30 can normalize text data for which the refinement process has been completed.

Next, the chatbot management server 30 can classify the collected data by task, individual, and company (S220).

Specifically, the chatbot management server 30 divides the collected data into Machine Generated Data, Semi Generated Data, Human Generated Data, and Enterprise Specialized Data. , Data can be classified by type, including Enterprise Normalized Data and Enterprise Specialized Vocabulary Data, by task, individual, and company.

Next, the chatbot management server 30 can extract and remove ambiguous data from data classified based on user information (S230).

Specifically, the chatbot management server 30 can remove ambiguous data by considering homonyms, polysemy, metaphors, vocabulary, and sentences based on user information.

For example, the chatbot management server 30 asks “How much is our company’s operating profit this year?” based on user information. In the query 'this year' and 'operating profit', ambiguous data can be extracted. In other words, this year's standard follows the previous year's standard until March of the following year in terms of accounting standards, and this year based on user information refers to operating profit from January 1 to the present, so ambiguous data is provided based on user information. can be deleted. Accordingly, by removing ambiguous data, the user's trust can be improved by more clearly generating and delivering an answer corresponding to the user's question intention.

Next, the chatbot management server 30 can classify the data from which ambiguous data has been removed by time and repeatedly learn the data by time (S240).

Specifically, the chatbot management server 30 repeatedly learns data by type, including Realtime, Daily, Monthly, Quarterly, and Yearly, for each task, individual, and company. Learning data can be generated separately.

Next, the chatbot management server 30 can generate learning data through an inspection process (S250, S260).

If the generated learning data does not pass the inspection process (S250), the chatbot management server 30 performs the data classification step (S220), the ambiguous data extraction and removal step (S240), or the data learning step by time point (S240). ), you can go back to at least one step and repeat the step to generate learning data. Accordingly, user reliability can be further improved while increasing the accuracy of answers.

According to the embodiment, the user terminal 10 determines the similarity of the answer data received from the chatbot management server 30 and generates a feedback signal, and the chatbot management server 30 generates a feedback signal received from the user terminal 10. In response, control signals that can improve chatbot services can be generated.

Depending on the embodiment, the user terminal 10 may receive a control signal containing improved or updated information on the chatbot service in response to the feedback signal before, during, or after using the chatbot service.

Next, the chatbot management server 30 can update the learning data in real time using conversation data including answers.

Lastly, the chatbot management server 30 can terminate the chatbot service when the user terminal 10 requests termination of the chatbot service.

The steps of the method or algorithm described in connection with embodiments of the present invention may be implemented directly in hardware, implemented as a software module executed by hardware, or a combination thereof. Software modules can be RAM (Random Access Memory), ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable ROM), Flash Memory, hard disk, removable disk, CD-ROM, or It may reside on any type of computer-readable recording medium well known in the art to which the invention pertains.

Above, embodiments of the present invention have been described with reference to the attached drawings, but those skilled in the art will understand that the present invention can be implemented in other specific forms without changing its technical idea or essential features. You will be able to understand it. Therefore, the embodiments described above should be understood in all respects as illustrative and not restrictive.

1: Chatbot service provision system that can be analyzed by point in time
10: user terminal
20: Chatbot terminal
30: Chatbot management server

Claims (21)

In a method of providing a chatbot service that can be analyzed by time point performed by a chatbot management server,
Generating learning data through repeated learning based on big data;
Receiving conversation data including a query from a user terminal through a chatbot service;
Classifying a conversation model by analyzing conversation data including the query based on the learning data to generate analysis data;
Selecting a chatbot terminal corresponding to the classified conversation model; and
generating conversation data containing an answer corresponding to the query by the chatbot terminal searching and extracting data stored in the learning data; Including,
The step of generating the learning data is,
collecting the big data;
generating preprocessed data by preprocessing the big data and converting it into classifiable data;
The preprocessed data is divided into Machine Generated Data, Semi Generated Data, Human Generated Data, Enterprise Specialized Data, Enterprise Normalized Data, and Classifying by type including Enterprise Specialized Vocabulary Data to generate classification data classified by task, individual, and company;
extracting and removing ambiguous data from the classified big data; and
The big data from which the above ambiguous data has been removed is divided by time on the basis of Realtime, Daily, Monthly, Quarterly and Yearly and repeatedly studied to provide data for each task, individual and company. Including a step of generating the learning data of the star,
The step of generating the classification data is,
The machine-processed data including real-time data automatically generated by a machine from the pre-processed data, the semi-automatically processed data including arithmetical calculation formulas and artificially set target value data, and the data artificially created, judged, and defined. The user-processed data including the user-processed data, the company-specific data including data limited to a specific company, the company universal data including universal data regardless of company classification, and the company containing vocabulary data used only for a specific company. Classifying by job, individual, and company based on specialized vocabulary data; and
A method of providing a chatbot service that can be analyzed at each point in time, including the step of inspecting the data classified by task, individual, and company for duplication and similarity. According to paragraph 1,
The step of classifying the conversation model is,
A method of providing a chatbot service that can be analyzed by viewpoint, classifying the conversation model by extracting and analyzing morphemes, verb phrases, and content words included in conversation data including the query. According to paragraph 2,
The step of classifying the conversation model is,
A step of automatically performing typo correction corresponding to the extracted morpheme, the verb phrase, and the content word,
When a typo is found through the typo inspection, it is determined whether the searched typo is an actual typo in response to standard data, and when it is an actual typo, the typo is corrected in response to the standard data. A method of providing a chatbot service that can be analyzed at each point in time. According to paragraph 2,
The step of classifying the conversation model is,
A step of automatically performing typo correction corresponding to the extracted morpheme, the verb phrase, and the content word;
When a typo is found through the typo inspection, determining whether the searched typo is an actual typo in response to the standard data, and determining whether or not the typo is analyzed through typo analysis when it is not an actual typo; analyzing by time point, including; Possible chatbot service provision methods. According to paragraph 4,
A method of providing a chatbot service that can be analyzed at each point in time, in which, if the typo is a misanalysis through the typo analysis, the morpheme, the verb phrase, and the content word are re-extracted to generate the analysis data. delete delete delete delete delete According to paragraph 1,
The step of preprocessing the big data is,
performing a tokenization operation using a space removal filter and a special character removal filter so that words included in the big data are divided into minimum units of meaningful words;
After the tokenization operation is completed, performing a refining operation to highlight the meaning of the remaining words by removing noise data for words with low frequency of occurrence or words that are repeated a large number of times included in the big data; and
Normalizing the big data after the refining operation is completed; A method of providing a chatbot service that can be analyzed at each point in time, including. According to paragraph 1,
Updating the learning data in real time in response to conversation data including the answer; A method of providing a chatbot service that can be analyzed at each point in time, further comprising: delete A step where the chatbot management server generates learning data by repeatedly learning based on big data;
A chatbot terminal receiving conversation data including user information and inquiries from a user terminal through a chatbot service;
generating analysis data by the chatbot terminal analyzing conversation data including the query based on the user information; and
The chatbot terminal searches the learning data using the analysis data based on the user information and generates conversation data including an answer; Including,
The step of generating the learning data is,
The chatbot management server collecting big data;
The chatbot management server preprocesses the big data and converts it into classifiable data to generate preprocessed data;
The chatbot management server divides the preprocessed data into Machine Generated Data, Semi Generated Data, Human Generated Data, Enterprise Specialized Data, and Enterprise General Data ( Creating classified data classified by job, individual, and company by classifying them by type, including Enterprise Normalized Data) and Enterprise Specialized Vocabulary Data;
Extracting and removing ambiguous data from the classified big data by the chatbot management server; and
The chatbot management server repeatedly learns the big data from which the ambiguous data has been removed by dividing it by time based on Realtime, Daily, Monthly, Quarterly, and Yearly. Including; generating the learning data for each individual, individual, and company,
The step of generating the classification data is,
The machine-processed data including real-time data automatically generated by a machine from the pre-processed data, the semi-automatically processed data including arithmetical calculation formulas and artificially set target value data, and the data artificially created, judged, and defined. The user-processed data including the user-processed data, the company-specific data including data limited to a specific company, the company universal data including universal data regardless of company classification, and the company containing vocabulary data used only for a specific company. Classifying by job, individual, and company based on specialized vocabulary data; and
A method of providing a chatbot service that can be analyzed at each point in time, including the step of inspecting the data classified by task, individual, and company for duplication and similarity. delete delete A chatbot management server that provides real-time chatbot services to user terminals through chatbot terminals selected by situation, field, and individual; Including,
The chatbot management server analyzes conversation data containing a query received from the user terminal, generates analysis data, classifies the conversation model, and then collects data corresponding to the conversation model from the learning data generated by considering ambiguous data. Extracts and generates conversation data containing answers and transmits it to the user terminal through the chatbot terminal,
The chatbot management server preprocesses the big data collected in consideration of the ambiguous data and converts it into classifiable data, and converts the preprocessed data into machine generated data, semi-generated data, Classified data categorized by job, individual and company into Human Generated Data, Enterprise Specialized Data, Enterprise Normalized Data and Enterprise Specialized Vocabulary Data. Generates classified data, extracts and improves the ambiguous data from the big data, and analyzes the big data from which the ambiguous data has been removed in real time, daily, monthly, and quarterly And the above-mentioned learning data for each task, individual, and company is generated through repetitive learning by dividing it by time on an annual basis,
The classification data is,
The machine-processed data including real-time data automatically generated by a machine using the pre-processed data, the semi-automatically processed data including arithmetical calculation formulas and artificially set target value data, and the data artificially created, judged, and defined. The user-processed data including the user-processed data, the company-specific data including data limited to a specific company, the company universal data including universal data regardless of company classification, and the company containing vocabulary data used only for a specific company. A chatbot service provision system that can be analyzed at each point in time, created by classifying data by job, individual, and company based on specialized vocabulary data, and then inspecting the data classified by job, individual, and company for duplication and similarity. delete Chatbot management server that provides real-time chatbot services to user terminals; Including,
The chatbot management server analyzes conversation data containing user information and queries received from the user terminal, generates analysis data, classifies the conversation model, and then creates the conversation model and Extract data corresponding to the user information, generate conversation data containing answers, and transmit it to the user terminal,
The chatbot management server is,
The analysis data is generated by extracting and analyzing morphemes, verb phrases, and content words included in the conversation data containing the query, respectively, by situation, field, and individual, and
The chatbot management server is,
Considering the above ambiguous data, the collected big data is pre-processed and converted into classifiable data, and the pre-processed data is converted into machine generated data, semi-generated data, and user-processed data. Generated Data, Enterprise Specialized Data, Enterprise Normalized Data, and Enterprise Specialized Vocabulary Data are classified by job, individual, and company to generate classified data. Data Extract and improve the ambiguous data from the big data, and analyze the big data from which the ambiguous data has been removed in real time, daily, monthly, quarterly, and yearly formats. As a standard, the above-mentioned learning data for each task, individual, and company is generated through repeated learning by dividing by time period,
The classification data is,
The machine-processed data including real-time data automatically generated by a machine from the pre-processed data, the semi-automatically processed data including arithmetical calculation formulas and artificially set target value data, and the data artificially created, judged, and defined. The user-processed data including the user-processed data, the company-specific data including data limited to a specific company, the company universal data including universal data regardless of company classification, and the company containing vocabulary data used only for a specific company. A chatbot service provision system that can be analyzed at each point in time, created by classifying data by job, individual, and company based on specialized vocabulary data, and then inspecting the data classified by job, individual, and company for duplication and similarity. delete A computer program stored in a computer-readable recording medium, combined with a hardware computer, to perform the method of any one of claims 1 to 5, 11 to 12, and 14.

Images