WO2020138607A1

WO2020138607A1 - Method and device for providing question and answer using chatbot

Info

Publication number: WO2020138607A1
Application number: PCT/KR2019/007698
Authority: WO
Inventors: 김영재; 김두현; 박근영
Original assignee: 건국대학교 산학협력단
Priority date: 2018-12-27
Filing date: 2019-06-26
Publication date: 2020-07-02
Also published as: KR101982990B1

Abstract

Disclosed is a method and device for providing a question and an answer by using a chatbot. A method for providing a question and an answer by using a chatbot comprises the steps of: receiving a user input including a subject of a question via a user interface; determining whether the subject of the question is image data; if the subject of the question is image data, performing image data pre-processing on the image data; detecting a character from the image data on which the image data pre-processing has been performed; on the basis of the character extracted from the image data, selecting a response message corresponding to the image data from among candidate response messages stored in a database; and providing the selected response message to a user.

Description

Method and device for question and answer using chatbot

The following embodiments relate to a technique for answering a query using a chatbot.

A chatbot is an interactive messenger that AI analyzes the content of a question and provides an appropriate answer when you enter a question like chatting on a messenger. Leading companies at home and abroad have recently introduced chatbots to customer support services to reduce the number of personnel required to respond to customers, and to improve the quality of customer support services by rapidly responding 24 hours a day.

On the other hand, since the query object that the chatbot can recognize and respond to is limited to text data, a user who has difficulty in providing the query of the text data may have a great difficulty in using the chatbot function. In order to solve these problems, a technical means for developing a chatbot capable of providing a wider range of query and response functions by recognizing various types of data is required.

A query response method using a chatbot according to an embodiment may include receiving a user input including a query target through a user interface; Determining whether the query target is image data; If the query target is the image data, performing image data preprocessing on the image data; Detecting a character from the image data on which the image data pre-processing has been performed; Selecting a response message corresponding to the image data among candidate response messages stored in a database, based on a character extracted from the image data; And providing the selected response message to the user.

The preprocessing of the image data may include performing at least one of adjusting the resolution of the image data, changing the color of the image data to grayscale, and binarizing the image data.

The detecting of a character from the image data may include detecting a character included in the image data using optical character recognition.

A query response method using a chatbot according to an embodiment may include tokenizing the entire string when the number of character strings detected in the image data is equal to or less than a preset threshold; And listing the entire tokenized string as a token list.

A query response method using a chatbot according to an embodiment may include tokenizing the entire string when the number of character strings extracted from the image data is greater than a preset threshold; Randomly extracting a number of tokenized strings based on the threshold value; And listing the randomized tokenized strings as a token list.

The selecting of the response message may include performing morpheme analysis on the token list; And comparing the morpheme analysis result with candidate response messages stored in the database, and selecting a response message corresponding to the image data based on the comparison result.

The step of performing the morpheme analysis may include extracting words included in the string through morpheme analysis of the string included in the token list, and removing words that overlap with each other among the extracted words.

The selecting of the response message may include calculating a similarity between words included in candidate response messages stored in the database and words extracted from a string included in the token list; And selecting, among the candidate response messages, a candidate response message that satisfies the preset condition of the similarity as a response message corresponding to the image data.

When the similarity does not satisfy a preset condition, the method may further include providing a response message including a non-response message to the user.

A query response method using a chatbot according to an embodiment may include receiving a user input including a query target through a user interface; Determining whether the query target is image data; If the query target is the image data, detecting a character from the image data; Generating a token list by tokenizing the entire character string when the number of character strings detected in the image data is equal to or less than a preset threshold; Calculating similarity between words included in candidate response messages stored in a database and words extracted from a string included in the token list; Selecting a response message corresponding to the image data among the candidate response messages based on the similarity; And providing the selected response message to the user.

The query response method using a chatbot according to an embodiment may further include providing a response message including a non-response message to the user when the similarity does not satisfy a preset condition.

A query response device using a chatbot according to an embodiment includes a user interface for receiving a user input including a query target; A determination unit to determine whether the query target is image data; An image data pre-processing unit performing pre-processing of image data on the image data when the query target is the image data; A character detecting unit detecting a character from the image data on which the image data pre-processing has been performed; A processing unit for selecting a response message corresponding to the image data among candidate response messages stored in a database based on a character extracted from the image data; And an output unit providing the selected response message to a user.

The image data pre-processor may perform at least one of adjusting the resolution of the image data, changing the color of the image data to grayscale, and binarizing the image data.

The character detection unit may detect a character included in the image data using optical character recognition.

When the number of character strings detected in the image data is less than or equal to a preset threshold, the character detector may tokenize the entire character string and list the entire tokenized character string as a token list.

When the number of character strings extracted from the image data is greater than a preset threshold value, the character detection unit tokenizes the entire character string and randomly extracts the number of character strings based on the threshold value among the tokenized character strings, The randomized tokenized string may be listed as a token list.

The processing unit calculates the similarity between words included in candidate response messages stored in the database and words extracted from the string included in the token list, and the similarity among the candidate response messages satisfies a preset condition. The candidate response message to be selected may be selected as a response message corresponding to the image data.

When the similarity does not satisfy a preset condition, the processing unit may select a response message including a non-response message.

According to an embodiment, in the query response method using a chatbot, a query target is not limited to text data, and a response to a query target including image data can be provided, thereby enabling various types of query responses.

According to an embodiment, a query target containing image data is received, a character included in the image data is detected, and a candidate response message having the highest similarity among candidate response messages stored in the chatbot device database based on the detected character By selecting and providing as response data, it is possible to provide appropriate response data for a query target including image data.

1 is a view showing the overall configuration of a query response system using a chatbot according to an embodiment.

2 is a flow chart for explaining the operation of the query response method using a chatbot according to an embodiment.

3 and 4 are flowcharts for explaining an example of a query response method using a chatbot according to an embodiment.

5 is a diagram illustrating the configuration of a query response device using a chatbot according to an embodiment.

Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. However, various modifications may be made to the embodiments, and the scope of the patent application right is not limited or limited by these embodiments. It should be understood that all modifications, equivalents, or substitutes for the embodiments are included in the scope of rights.

The terms used in the examples are for illustrative purposes only and should not be construed as limiting. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, the terms "include" or "have" are intended to indicate the presence of features, numbers, steps, actions, components, parts or combinations thereof described in the specification, one or more other features. It should be understood that the existence or addition possibilities of fields or numbers, steps, operations, components, parts or combinations thereof are not excluded in advance.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by a person skilled in the art to which the embodiment belongs. Terms, such as those defined in a commonly used dictionary, should be interpreted as having meanings consistent with meanings in the context of related technologies, and should not be interpreted as ideal or excessively formal meanings unless explicitly defined in the present application. Does not.

In addition, in the description with reference to the accompanying drawings, the same reference numerals are assigned to the same components regardless of reference numerals, and redundant descriptions thereof will be omitted. In describing the embodiments, when it is determined that detailed descriptions of related well-known technologies may unnecessarily obscure the subject matter of the embodiments, detailed descriptions thereof will be omitted.

Referring to FIG. 1, when a query is received from a user, the query response device 100 using a chatbot is a device that provides a response to the query to the user using a chatbot. For example, the query response device 100 may receive an image type query as well as a text type query, and analyze each query to derive an optimal response suitable for the query and provide it to the user.

The query response device 100 includes an image data analysis unit 105 for analyzing image data, a morpheme analysis unit 110 for analyzing morphemes, and a response message providing unit 115 for selecting and providing a response message to a user, and user input It may include a user interface 160 that can receive.

In one embodiment, the image data analysis unit 105 includes an image data pre-processing unit 120 that performs image data pre-processing on the received image data, a character detection unit 125 that detects characters from the image data on which image data pre-processing has been performed, and It may include a tokenization performing unit 130 to tokenize the detected character string to list the tokenized character string as a token list.

The morpheme analysis unit 110 detects words by performing morpheme analysis on a string listed as a token list when the query target included in the user input is image data, and removes duplicate words from the processing unit 2 135 And when the query target included in the user input is text data, the processing unit 1 140 that classifies the text component analysis result through morphological analysis as a tag and adds tag information to the text data. . The response message providing unit 115 calculates the similarity between the morpheme analysis result received from the morpheme analysis unit 110 and the candidate response message, and selects a candidate response message that satisfies a preset similarity condition as a response message, 145 It may include a database 150 for storing the candidate response message and an output unit 155 capable of providing the response message to the user.

In the above embodiment, when the query target included in the user input received from the user is image data, the image data received from the user may be pre-processed by the image data preprocessing unit 120. The image data pre-processing unit 120 performs image data pre-processing so that the character detection unit 125 can detect characters included in the image data. The image data pre-processing unit 120 may perform binarization of the image data by adjusting the resolution of the image data and changing the color of the image data to grayscale. The character detection unit 125 may detect characters from image data on which image data preprocessing has been performed. The character detector 125 may detect a character from the pre-processed image data by utilizing an optical character recognition function that converts a character included in the image data into a machine-readable character. The character string detected by the character detector 125 may be tokenized by the tokenization performer 130, and the tokenized character string may be listed as a token list.

After the image data preprocessing is performed by the image data analysis unit 105, characters are detected, and tokenization is performed, the token list of the listed string is morphologically analyzed by the processing unit 2 135 included in the morpheme analysis unit 110 This can be done. The processor 2 135 may extract words included in the string through morpheme analysis of the string included in the token list. Similarity between candidate response messages stored in the database 150 in the processing unit 145 included in the response message providing unit 115 may be calculated as a result of the morpheme analysis. The processor 145 may calculate a similarity between words included in the morpheme analysis result and words included in the candidate response message, and select a candidate response message that satisfies a preset similarity condition as a response message. If there is no candidate response message that satisfies the similarity condition, the processing unit 145 may select a non-response message as the response message. The output unit 155 may provide a response message selected by the processing unit 145 to the user through a user interface.

In another embodiment, when the query target included in the user input received from the user is text data, the user through the morpheme analysis unit 110 and the response message providing unit 115 of the query response device 100 using a chatbot You can be provided with answers related to the query target. In the text data, the sentence component analysis through the morpheme analysis is performed in the processing unit 1 140 included in the morpheme analysis unit 110, and the tag information may be added to the text data by classifying the sentence component analysis results into tags. As a result of analyzing the sentence components through the morpheme analysis, the processing unit 150 included in the response message providing unit 115 may determine whether the sentence structure patterns match between candidate response messages stored in the database 150. When it is determined that a candidate response message having a pattern matching the morpheme analysis result exists, the processing unit 150 may select a candidate response message having a pattern matching the morpheme analysis result as a response message. When there are no or two or more candidate response messages that match the morpheme analysis result and the pattern, the processing unit 150 may select the non-response message as the response message. The response message selected by the processing unit 150 may be provided to the user through the output unit 155.

Hereinafter, a method of answering a query using a chatbot will be described in more detail with reference to the drawings.

2 is a flow chart for explaining the operation of the query response method using a chatbot according to an embodiment. A query response method using a chatbot may be performed by a query response device using a chatbot described herein.

Referring to FIG. 2, in step 210, the query response device may receive a user input including a query target through a user interface. In one embodiment, the chatbot may receive a query target including data in the form of an image, text, or the like as user input.

In step 220, the query response device may determine whether the query target is image data. When it is determined that the query target included in the user input is image data, in step 230, the query response device may perform image data preprocessing on the image data. In one embodiment, the query response device may perform resolution adjustment of image data received as user input, change color of image data to grayscale, binarization of image data, and the like.

In step 240, the query and response device may detect a character in the image data in which the image data pre-processing is performed through resolution adjustment, color change to grayscale, and binarization based on a preset threshold.

In one embodiment, the query response device may detect a character included in the image data using optical character recognition (OCR). Here, the optical character recognition may mean a function of converting characters included in image data into machine-readable characters. Characters detected in the image data through optical character recognition may be tokenized based on the blank character of the character string, and if the amount of the character string exceeds a certain threshold value, that is, a predetermined threshold value, the character value is assigned to a threshold value among the tokenized character strings. Strings can be randomly extracted as many as the base number. For example, if the number of character strings is equal to or less than a preset threshold, the entire string can be tokenized and the entire tokenized string can be listed as a token list. On the other hand, if the number of character strings is greater than a preset threshold, the entire string can be tokenized, and the tokenized string can be arbitrarily extracted from the tokenized string up to five times the threshold. Also, it is possible to list a randomized tokenized string as a token list.

In step 250, the query response device may select a response message corresponding to the image data among candidate response messages stored in the database, based on the characters extracted from the image data.

In one embodiment, the query response device may perform morpheme analysis on a token list composed of tokenized strings. Here, morpheme analysis may mean classifying the tokenized string into the smallest unit of words with meaning. Also, the morpheme analysis may include a process of extracting words included in the string through morphological analysis of the string included in the token list, and removing words that overlap with each other among the extracted words. The query response device may calculate the similarity between words extracted from the string through morphological analysis and words included in candidate response messages stored in the database. The query response device may select a candidate response message corresponding to the image data from a candidate response message that satisfies a preset condition from among the candidate response messages. At this time, if there is no candidate response message corresponding to the similarity level that satisfies the preset condition, the query response device may select a response message including a non-response message.

In step 260, the query response device may provide the selected response message to the user.

In another embodiment, in step 220, the query response device determines whether the query target is image data, and when it is determined that the query target is not image data, in step 270, text data included in the user input Characters included in can be detected.

3 and 4 are flowcharts for explaining an example of a method for answering a query according to an embodiment.

Referring to FIG. 3, in step 305, the query response device may receive a user input including a query target from a user through a user interface. According to an embodiment, the user input may include image data and text data.

In step 310, the query response device may determine whether the query target included in the user input is image data. If it is determined in step 310 that the query target included in the user input is text data, in step 350 the query response device classifies the sentence component analysis result through the morpheme analysis into tags and adds tag information to the text data. Can.

Alternatively, when it is determined that the query target included in the user input is the image data, the query response device may perform preprocessing of the image data on the image data received in operation 315. In the process of performing the image data pre-processing, the Q&A device can adjust the resolution of the image data to a resolution that facilitates character extraction, and changes the color of the image data to grayscale, thereby adapting the image data through adaptive thresholding. Binarization can be performed. By dividing the image data into zones and performing binarization of the image data through adaptive demarcation, which processes binarization for each zone, the image data preprocessor performs characterization in the image data rather than binarization based on the entire image data. Binarization of image data that is easy to extract can be performed.

In step 320, the query response device may detect a character in the image data. In one embodiment, the query response device may utilize an optical character recognition function to detect characters in image data.

In connection with the character detected in step 320, the query response device in step 325 may determine whether the number of character strings is greater than or less than a preset threshold. If it is determined that the number of detected strings is greater than a preset threshold, in step 330, the query response device may tokenize the entire string based on the blank character of the string, and N of the threshold among the tokenized strings The tokenized string can be randomly extracted as many times as the number of times (N is a natural number). If it is determined that the number of detected strings is less than or equal to a preset threshold, in step 335, the query response device may tokenize the entire string based on the blank character of the string.

In step 340, the query response device may list the strings tokenized in

steps

330 and 335 as a token list.

In step 345, the query response device may perform morpheme analysis on the string included in the token list. Through the morpheme analysis, for example, the sentence “Cheolsu has read a book” can be analyzed as “Cheolcheong/Ga/Book/Read/Read/It”. In one embodiment, the query response device may extract words from the classified string through performing morphological analysis, and remove duplicate words from the extracted words.

Based on the words extracted from the string as a result of performing step 345 and the text data analyzed as a result of performing step 350, the query answering device in step 355 may generate a morpheme analysis result.

Subsequently, referring to FIG. 4, based on the morpheme analysis result generated in step 355 of FIG. 3, the query answering device in step 405 may determine whether the query target is image data. When it is determined that the query target is image data, in step 410, the query response device may calculate the similarity between the morpheme analysis result and the candidate response message stored in the database.

In one embodiment, a method of calculating the similarity between words included in the morpheme analysis result and candidate response messages stored in the database may be as follows. The query response device may detect words included in each candidate response message stored in the database and list them as a list corresponding to each candidate response message. Among the words included in the morpheme analysis result and the words corresponding to the candidate response message included in the list, the number of words that match each other is divided by the number of words included in the list corresponding to the candidate response message and included in the morpheme analysis result. Similarity can be calculated between the words and candidate response messages stored in the database.

In step 415, the query response device may determine whether there is a candidate response message that satisfies a condition of a predetermined similarity among the similarities corresponding to the candidate response messages stored in the database. The condition of the predetermined similarity may be, for example, a case in which there is only one candidate response message corresponding to the maximum value among the calculated similarities. When there are two or more candidate response messages corresponding to the maximum value of the similarity, or 0, it may be determined that there is no candidate response message that satisfies the condition of the predetermined similarity.

If there is a candidate response message that satisfies a predetermined similarity condition, in step 420, the query response device may select the candidate response message as a response message. Conversely, when there is no candidate response message that satisfies a predetermined similarity condition, in step 425, the query response device may select a non-response message as a response message corresponding to the query target.

If the query target is not determined to be image data, in step 435, the query response device determines whether there is a candidate response message that matches the morpheme analysis result and the pattern based on the morpheme analysis result of the text data corresponding to the query target. I can judge.

If it is determined that there is a candidate response message that matches the morpheme analysis result and the sentence structure pattern, in step 440, the query response device may select a candidate response message that matches the morpheme analysis result and the sentence structure pattern as a response message. . On the other hand, if it is determined that there is no candidate response message that matches the morpheme analysis result and the sentence structure pattern, in step 425, the query response device may select a non-response message as the response message.

In step 430, the query response device may provide the selected response message to the user in response to the query target.

Referring to FIG. 5, the query response device 500 includes a user input interface 510, a determination unit 520, an image data pre-processing unit 530, a character detection unit 540, a processing unit 550, and an output unit 560. And a database 570.

The user input interface 510 may receive a user input including a query target from a user. The determination unit 520 may determine whether the query target included in the user input received by the user input interface 510 is image data. When it is determined that the query target is image data, the image data preprocessing unit 530 may perform image data preprocessing on the image data. Here, the image data pre-processing may include processes such as resizing the image data to a size suitable for extracting characters from the image data, changing the color of the image data to grayscale, and binarizing the image data based on a preset threshold. have.

The character detection unit 540 may detect characters from image data on which image data preprocessing has been performed. The character detection unit 540 may detect a character included in image data using optical character recognition. Characters detected from the image data through optical character recognition classify cases where the number of character strings is equal to or less than a preset threshold value and when the number of character strings is greater than a preset threshold value, and tokenize the strings in different ways and tokenize the strings into a token list Can be listed as

If the number of character strings detected in the image data is less than or equal to a preset threshold, the character detector 540 may tokenize the entire character string and list the entire tokenized character string as a token list. On the other hand, when the number of character strings detected in the image data is greater than a preset threshold, the character detector 540 tokenizes the entire character string, and randomly tokenizes the number of tokenized character strings based on the threshold value. Can be extracted. The extracted tokenized string can be listed as a token list. The number based on the threshold may be, for example, a number reaching 5 times the threshold.

The processor 550 may select a response message corresponding to the image data among candidate response messages stored in the database 570 based on the characters extracted from the image data.

In one embodiment, the processing unit 550 may perform morpheme analysis on the list of tokens listed by the character detection unit 540. Duplicate words may be removed from words extracted as a result of the morpheme analysis performed by the processing unit 550. The processing unit 550 may calculate the similarity of words extracted from the token list to words included in candidate response messages stored in the database 570. A candidate response message corresponding to the similarity having the highest value among the calculated similarities may be selected as the response message. If there is no single similarity with the highest value, the processing unit 550 may select a non-response message as the response message. The output unit 560 may provide the selected response message to the user.

The device described above may be implemented with hardware components, software components, and/or combinations of hardware components and software components. For example, the devices and components described in the embodiments include, for example, processors, controllers, arithmetic logic units (ALUs), digital signal processors (micro signal processors), microcomputers, field programmable gate arrays (FPGAs). , A programmable logic unit (PLU), microprocessor, or any other device capable of executing and responding to instructions, may be implemented using one or more general purpose computers or special purpose computers. The processing device may run an operating system (OS) and one or more software applications running on the operating system. Further, the processing device may access, store, manipulate, process, and generate data in response to the execution of the software. For convenience of understanding, a processing device may be described as one being used, but a person having ordinary skill in the art, the processing device may include a plurality of processing elements and/or a plurality of types of processing elements. It can be seen that may include. For example, the processing device may include a plurality of processors or a processor and a controller. In addition, other processing configurations, such as parallel processors, are possible.

The software may include a computer program, code, instruction, or a combination of one or more of these, and configure the processing device to operate as desired, or process independently or collectively You can command the device. Software and/or data may be interpreted by a processing device, or to provide instructions or data to a processing device, of any type of machine, component, physical device, virtual equipment, computer storage medium or device. , Or may be permanently or temporarily embodied in the signal wave being transmitted. The software may be distributed on networked computer systems and stored or executed in a distributed manner. Software and data may be stored in one or more computer-readable recording media.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, or the like alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiments or may be known and usable by those skilled in computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs, DVDs, and magnetic media such as floptical disks. Includes hardware devices specifically configured to store and execute program instructions such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of program instructions include high-level language code that can be executed by a computer using an interpreter, etc., as well as machine language codes produced by a compiler. The hardware device described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

As described above, although the embodiments have been described by a limited embodiment and drawings, those skilled in the art can make various modifications and variations from the above description. For example, the described techniques are performed in a different order than the described method, and/or the components of the described system, structure, device, circuit, etc. are combined or combined in a different form from the described method, or other components Alternatively, proper results can be achieved even if replaced or substituted by equivalents.

Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

Claims

In the question and answer method using the chatbot,

Receiving a user input including a query target through a user interface;

Determining whether the query target is image data;

If the query target is the image data, performing image data preprocessing on the image data;

Detecting a character from the image data on which the image data pre-processing has been performed;

Selecting a response message corresponding to the image data among candidate response messages stored in a database, based on a character extracted from the image data; And

Providing the selected response message to a user

Containing,

How to answer questions using chatbot.
According to claim 1,

Step of performing the image data pre-processing,

Adjusting the resolution of the image data, changing the color of the image data to grayscale, and performing at least one of binarization of the image data

Containing,

How to answer questions using chatbot.
According to claim 1,

The step of detecting a character in the image data,

Detecting a character included in the image data using optical character recognition

Containing,

How to answer questions using chatbot.
According to claim 1,

Tokenizing the entire character string when the number of character strings detected in the image data is equal to or less than a preset threshold; And

Listing the entire tokenized string into a token list

Further comprising,

How to answer questions using chatbot.
According to claim 1,

Tokenizing the entire string if the number of character strings extracted from the image data is greater than a preset threshold;

Randomly extracting a number of tokenized strings based on the threshold value; And

Listing the randomized tokenized strings as a token list

Further comprising,

How to answer questions using chatbot.
The method of claim 5,

The step of selecting the response message,

Performing morpheme analysis on the token list; And

Comparing the morpheme analysis result with candidate response messages stored in the database, and selecting a response message corresponding to the image data based on the comparison result

Containing,

How to answer questions using chatbot.
The method of claim 6,

The step of performing the morpheme analysis,

Extracting words contained in the string through morpheme analysis of the string included in the token list, and removing words that overlap with each other among the extracted words

Containing,

How to answer questions using chatbot.
The method of claim 7,

The step of selecting the response message,

Calculating similarity between words included in candidate response messages stored in the database and words extracted from a string included in the token list; And

Selecting, among the candidate response messages, a candidate response message that satisfies the preset condition of the similarity as a response message corresponding to the image data.

Containing,

How to answer questions using chatbot.
The method of claim 8,

If the similarity does not satisfy a preset condition, providing a response message including a non-response message to the user

Further comprising,

How to answer questions using chatbot.
In the question and answer method using the chatbot,

Receiving a user input including a query target through a user interface;

Determining whether the query target is image data;

If the query target is the image data, detecting a character from the image data;

Generating a token list by tokenizing the entire character string when the number of character strings detected in the image data is equal to or less than a preset threshold;

Calculating similarity between words included in candidate response messages stored in a database and words extracted from a string included in the token list;

Selecting a response message corresponding to the image data among the candidate response messages based on the similarity; And

Providing the selected response message to a user

Containing,

How to answer questions using chatbot.
The method of claim 10,

Tokenizing the entire string if the number of character strings extracted from the image data is greater than a preset threshold;

Randomly extracting a number of tokenized strings based on the threshold value; And

Listing the randomized tokenized strings as a token list

Further comprising,

How to answer questions using chatbot.
The method of claim 10,

The step of selecting the response message,

Calculating similarity between words included in candidate response messages stored in the database and words extracted from a string included in the token list; And

Selecting, among the candidate response messages, a candidate response message that satisfies the preset condition of the similarity as a response message corresponding to the image data.

Containing,

How to answer questions using chatbot.
The method of claim 8,

If the similarity does not satisfy a preset condition, providing a response message including a non-response message to the user

Further comprising,

How to answer questions using chatbot.
A computer-readable recording medium recording a program for executing the method of any one of claims 1 to 13 in a computer.
A user interface for receiving a user input including a query target;

A determination unit to determine whether the query target is image data;

An image data pre-processing unit performing pre-processing of image data on the image data when the query target is the image data;

A character detecting unit detecting a character from the image data on which the image data pre-processing has been performed;

A processing unit for selecting a response message corresponding to the image data among candidate response messages stored in a database based on a character extracted from the image data; And

An output unit that provides the selected response message to a user

Containing,

Q&A device using chatbot.
The method of claim 15,

The image data pre-processing unit,

Adjusting the resolution of the image data, changing the color of the image data to grayscale, and performing at least one of binarization of the image data,

Q&A device using chatbot.
The method of claim 15,

The character detection unit,

A character included in the image data is detected using optical character recognition,

Q&A device using chatbot.
The method of claim 15,

The character detection unit,

If the number of character strings detected in the image data is equal to or less than a preset threshold, tokenize the entire character string,

Listing the entire tokenized string into a token list,

Q&A device using chatbot.
The method of claim 15,

The character detection unit,

If the number of character strings extracted from the image data is greater than a preset threshold, tokenize the entire character string,

Randomly extract the number of tokenized strings based on the threshold value,

Listing the randomized tokenized string into a token list,

Q&A device using chatbot.
The method of claim 15,

The processing unit,

The similarity between words included in candidate response messages stored in the database and words extracted from the string included in the token list is calculated,

Among the candidate response messages, a candidate response message that satisfies the preset condition of similarity is selected as a response message corresponding to the image data.

Q&A device using chatbot.
The method of claim 20,

The processing unit,

When the similarity does not satisfy the preset condition, selecting a response message including a non-response message,

Q&A device using chatbot.