KR20200032873A - Parsing system for chat bot service based on book information and control method thereof - Google Patents

Abstract

The present invention relates to a parsing system for a chatbot service based on book information, which comprises: a book information input unit collecting or receiving e-books, extracting at least one file necessary for analysis or search of each e-book, and transmitting the extracted file to a control unit; a control unit parsing the file received from the book information input unit, searching for words, sentences, paragraphs, and chapters included in each e-book based on the basic information, table of contents information, and detailed information of each e-book, and automatically generating a plurality of response sentences which can be communicated through a chatbot on a specific subject in a book in a designated specific field; and a database unit storing the plurality of response sentences generated by the control unit.

Description

PARSING SYSTEM FOR CHAT BOT SERVICE BASED ON BOOK INFORMATION AND CONTROL METHOD THEREOF}

The present invention relates to a parsing system for a book information-based chatbot service and a control method thereof, more specifically, by processing information of an e-book so that an answer database for a book chatbot service can be automatically generated. The present invention relates to a parsing system for an information-based chatbot service and a control method thereof.

Recently, 'chatbot' technology combining artificial intelligence and messenger has attracted attention.

The chat bot (Chat Bot) is software that performs a text or voice-based conversation with a user as a conversation partner, and is used for customer service or information collection purposes, and is used in combination with artificial intelligence assistant (Artificial Assitance). .

For example, the chatbot includes Google Assistant, Apple Siri, and Samsung Bixby. However, currently developed chatbots mostly perform functions related to daily life (eg, weather, home appliance control, shopping, etc.), and do not perform conversation functions for specific fields of expertise, as well as the points that users ask There are many problems with errors, such as not being able to understand (grasp) accurately.

Therefore, in order to solve such a problem, that is, the chatbot accurately recognizes the gist or meaning of a conversation exchanged with various users, and in order to continue an appropriate conversation with users, the texts that can be selected as conversation contents are learned in advance. In preparation, a function of providing text of the most appropriate content among the texts prepared by learning in advance as an answer should be implemented.

Currently, most chatbots create a database by manually writing in advance a myriad of context-sensitive responses that can occur during a conversation. Therefore, in order to communicate with a chatbot about a specific field (or special field), the developer (or manager) who will create the database must have sufficient knowledge about the specific field (or special field) before the situation related to the specific field Star responses can be prepared in advance.

However, as mentioned above, creating a database with a contextual answer manually created by a developer requires a lot of manpower, time, and effort, and furthermore, it requires a specialized manpower for this. There is a problem that can not smoothly proceed with the conversation.

If this problem is applied to the book (book) field, it will be described as follows.

For example, in order to search for a book of the content desired by the current user, select a category (group) of the book to be searched by the user in the search box of the homepage (for example, an internet bookstore, electronic library, portal site, etc.), and then part of the title, And it is common to search by directly entering the user as the keyword as the author.

However, in order to search for a book through a conversation with a chatbot, rather than a keyword search in which the user thinks the keyword entered in the search box as described above, the chatbot understands the question in the form of a sentence input by the user, and the book (book) In addition to the basic information (eg book name, publisher, author / translator, subject language, book introduction, price, category), the entire contents of the book (book) (eg table of contents information, detailed information) should be included in the database.

At this time, in order for the chatbot to understand the question in the form of a sentence input by the user, a developer (or a developer who has read the book) who has sufficient knowledge about the field of the book directly (ie, manually) uses the user for the book. You need to create a database by filling out the content to ask (ie situation) and answers to this situation (or question).

However, in the situation where hundreds or thousands of books are being published every day, developers (or administrators who have read the books) can directly (manually) respond multiple times to conversations with users that may occur regarding the book. It is virtually impossible to write, and as a result, the update of the chatbot for a specific specialized field (for example, a book field) is delayed, which causes a problem that a conversation with a user cannot be smoothly performed.

Background of the invention is disclosed in Republic of Korea Patent Registration No. 10-1851785 (2018.04.18. Registration, chatbot training set generation device and method).

According to an aspect of the present invention, the present invention was created to solve the above problems, and processes information of an e-book to automatically generate an answer database for a book chatbot service, based on book information. Its purpose is to provide a parsing system and control method for its chatbot service.

A parsing system for a book information-based chatbot service according to an aspect of the present invention includes: a book information input unit that collects or receives e-books, extracts at least one file necessary for analysis or search of each e-book, and delivers it to a control unit; By parsing the file received from the book information input unit, search for words, sentences, paragraphs, and chapters included in each e-book based on the basic information, table of contents information, and detailed information of each e-book, The control unit that automatically generates a plurality of response sentences that can be communicated through a chatbot on a specific subject in a book in a specific field; And a database unit storing a plurality of response sentences generated by the control unit.

In the present invention, the control unit, the book information parsing unit for parsing (parsing) the file received from the book information input unit, to analyze the basic information, table of contents information, and detailed information of each e-book; And search for words, sentences, paragraphs, and chapters based on the basic information, table of contents information, and detailed information of each parsed e-book, and a specific field designated by an administrator in advance or designated by a user during a chatbot conversation. Characterized in that it comprises; a response generating unit for automatically generating a plurality of response sentences having a variety of sentence types that can be communicated through a chatbot on a specific topic of.

In the present invention, the control unit further comprises a keyword generating unit for designating or generating keyword information for generating the plurality of response sentences, and the keyword information includes a conversation field, a conversation topic, a conversation purpose, and a conversation. As information reflecting the propensity, it is characterized in that it is input in the form of a plurality of words or sentences from a system administrator or a chatbot user.

In the present invention, the control unit further includes a response sentence processing unit that adjusts the priority of a response sentence desired by a chatbot user among response sentences stored in the database unit for each response sentence. The desired response sentence is a response sentence having a high hit rate or a response sentence having a high number of times adopted by a chatbot user.

In the present invention, the control unit performs a database unit update that adds a rating for a keyword of each answer sentence stored in the database unit in the form of metadata as additional information to help the chatbot select a response sentence, and the keyword rating It is characterized in that, in the order of the object, subject, verb, and bore, the keyword rating is designated as the higher priority with the frequently mentioned words first.

In the present invention, when the priority information of the answer sentences stored in the database unit is adjusted, the control unit updates the database unit, and also learns the types of the answer sentences in which the priority information is adjusted, and then generates the answer sentences. It is characterized by generating an answer sentence reflecting the learning result.

In the present invention, the control unit is characterized in that implemented in the form of a computer or server containing a CPU (CPU).

In a method of controlling a parsing system for a book information-based chatbot service according to another aspect of the present invention, a control unit of a parsing system for a book information-based chatbot service designates or generates keyword information for generating a plurality of response sentences step; Analyzing, by the control unit, information on e-books based on the keyword information, searching and designating a specific e-book and a specific area of the e-book to generate the response sentence; Searching for a sentence including at least one of the keyword information among text information of the specific e-book and a specific specific area of the e-book; And generating, by the control unit, the plurality of sentences searched for, and the plurality of answer sentences by combining the sentences in a predetermined manner, and then correcting the dialogues and storing them in a database unit.

In the present invention, the keyword information is information reflecting a conversation field, a conversation subject, a conversation purpose, and a conversation tendency, and is characterized in that it is input in a plurality of word forms or sentence form through the system and an administrator or a chatbot user. Is done.

In the present invention, the information of the e-books is characterized by including basic information, table of contents information, and detailed information.

In the present invention, a sentence including at least one or more keyword information may be searched for at least one or more sentences, and the searched sentence may include a short sentence form or a longer sentence form. It is characterized by.

In the present invention, the control unit extracting a key keyword from each of a plurality of response sentences stored in the database unit; Setting, by the control unit, a keyword rating for each answer sentence, and also classifying the keywords into individual phrases and semantic phrases; And performing an update to add additional information generated by generating the keyword rating and classification information for each answer sentence in the form of metadata in the database unit.

In the present invention, the additional information is characterized in that the chatbot is information that can be reflected in selecting a response suitable for a situation among a plurality of response sentences stored in the database unit.

In the present invention, the control unit accumulating the adoption information for each sentence of the response selected by the chatbot; Setting, by the control unit, a priority of answer adoption based on the accumulated adoption information; And the control unit updates the database unit by reflecting the priority of adopting the answer, and after learning the answer type having the highest answer adoption priority, generates a response sentence in a form that reflects the learned result when generating a subsequent response sentence. It characterized in that it further comprises a step of.

In the present invention, the adoption information is characterized in that the chatbot is information that the user or system administrator selects satisfaction or dissatisfaction during a conversation with the user.

According to an aspect of the present invention, the present invention enables to automatically generate an answer database for a book chatbot service by processing information of an e-book.

In addition, according to another aspect of the present invention, the present invention makes it possible to lower the error rate of answering a user's question.

In addition, according to another aspect of the present invention, the present invention enables users to communicate with users as accurately as possible about newly published books in various fields.

In addition, according to another aspect of the present invention, the present invention allows the user to recommend a book that the user may want to read based on the content of the conversation with the user.

1 is an exemplary view showing a schematic configuration of a parsing system for a book information-based chatbot service according to an embodiment of the present invention.
2 is a flowchart illustrating a control method of a parsing system for a book information-based chatbot service according to a first embodiment of the present invention.
FIG. 3 is an exemplary view of FIG. 2 in which a plurality of sentences including keyword information searched by the control unit are combined into interactive answer sentences and shown in a table form.
4 is a flowchart illustrating a control method of a parsing system for a book information-based chatbot service according to a second embodiment of the present invention.
5 is a flowchart illustrating a control method of a parsing system for a book information-based chatbot service according to a third embodiment of the present invention.
FIG. 6 is an exemplary view of FIG. 4 showing updated answer sentences by additionally adding rating and classification information for all keywords extracted from each answer sentence stored in the database unit, and a response priority of each sentence.

Hereinafter, an embodiment of a parsing system for a book information-based chatbot service according to the present invention and a control method thereof will be described with reference to the accompanying drawings.

In this process, the thickness of the lines or the size of components shown in the drawings may be exaggerated for clarity and convenience. In addition, terms to be described later are terms defined in consideration of functions in the present invention, which may vary according to a user's or operator's intention or practice. Therefore, the definition of these terms should be made based on the contents throughout the present specification.

1 is an exemplary view showing a schematic configuration of a parsing system for a book information-based chatbot service according to an embodiment of the present invention.

As shown in FIG. 1, the parsing system 100 for a book information-based chatbot service according to this embodiment includes a book information input unit 110, a keyword generation unit 120, a control unit 130, and response sentence processing. It includes a section 140, and a database section 150.

The book information input unit 110 collects or receives electronic books published by a plurality of publishers 11 and 12.

Here, the e-book refers to a digital book that can be used as a book by recording information such as text or images on an electronic medium, and is digitally transmitted and received through a wired or wireless communication network to read, view, and listen to the content on a computer or mobile terminal. Books.

There are various types of e-books, but there are many e-books having an epub (EPUB) file structure (e.g., e-book.epub), and e-books in the EPUB format include a list of all files included therein. While it contains, it serves as a navigation file (eg content.opf) and the table of contents information of the book (this table of contents is linked to the detailed information through a link), while determining the order of the book (eg chapter, section, section, etc.) The file (eg toc.ncx) is included by default.

Therefore, the book information input unit 110 extracts a file (for example, toc.ncx, content.opf, etc.) necessary for analysis or search of the collected or received e-books and transmits it to the control unit 130.

Accordingly, the control unit 130 parsing the file (eg, toc.ncx, content.opf, etc.) received from the book information input unit 110, basic information, table of contents information, and details of each e-book Based on the information, search for words, sentences, paragraphs, and chapters included in the e-book, and talk about a specific subject (eg, a subject in a field designated by an administrator or chatbot user) in a book in a specific field Various answers (eg short answer type, short type answer, long type answer, etc.) are automatically prepared (generated) and learned in advance. At this time, the learning is continuously performed as long as the system operates.

For example, the book information parsing unit 131 of the control unit 130 parses a file (eg, toc.ncx, content.opf, etc.) received from the book information input unit 110, so that the basics of each electronic book are parsed. Analyze information, table of contents information, and detailed information.

In addition, the answer generator 132 searches for words, sentences, paragraphs, and chapters based on the basic information, table of contents information, and detailed information of each parsed e-book, and the administrator has previously designated or is in a chatbot conversation. A variety of answers (for example, short-answer answers, short-answer answers, long-answer answers, etc.) are automatically prepared (generated) in which a user can talk on a specific topic in a specific field designated by the user (see FIG. 3).

At this time, the plurality of answers automatically prepared (generated) may be prepared (generated) in advance, or may be prepared (generated) in real time during a chatbot conversation.

Meanwhile, the keyword generation unit 120 designates or generates keyword information for the control unit 130 to prepare (create) a plurality of answers.

The keyword information is information reflecting a conversation field, a conversation topic, a conversation purpose, and a conversation tendency, and may be input in a plurality of word forms or in a sentence form (see keywords in FIG. 3).

Accordingly, the keyword generator 120 directly receives keyword information from the developer (or system administrator) 21 or the chatbot conversation user 22, or the developer (or system administrator) 21 or chatbot user 22. Keyword information can be automatically extracted from the sentence input by).

At this time, the information (ie, keyword) extracted as the keyword information is information indicating at least one of a conversation field, a conversation topic, a conversation purpose, and a conversation propensity.

Therefore, the control unit 130 searches for books related to the keyword information based on keyword information generated (or designated or extracted) in the keyword generation unit 120, and includes words, sentences, and paragraphs among the searched books. Search for, and chapters, and automatically prepare (create) a variety of answers (for example, short answer, short answer, long answer, etc.) to talk about a specific subject in the specific field ( 3).

Meanwhile, keyword information extracted or generated by the keyword generator 120 is not only used by the controller 130, but is generated by the controller 130 by using the keyword information in the answer sentence processing unit 140 as well. Process (or update) multiple replied answers (or reply sentences).

For example, among the plurality of answers (or answer sentences) stored in the database unit 150, the answer sentence processing unit 140 selects the answer desired by the chatbot user (that is, the answer with the highest hit rate) by the chatbot and chats. It is processed (or updated) so that it can be output as the answer.

In other words, among the plurality of answers (or answer sentences) generated by the control unit 130 and stored in the database unit 150, the answer sentence processing unit 140 is a response desired by the chatbot user (that is, a response with a high hit rate) Or, it performs the function of adjusting the priority (or adoption level) of the answer with a high number of times the user adopts.

In addition, the answer sentence processing unit 140 may provide additional information (eg, keyword rating (ie, priority) & classification (ie, population, meaning phrase)) to help the chatbot select an answer. In the form of metadata, updates can be performed.

For reference, in the order of the keyword, the object, the subject, the verb, and the bore, the ranking of the keyword may be designated as a higher priority with the frequently mentioned words first.

On the other hand, when the answers (or answer sentences) are processed (or updated in the database) in the answer sentence processing unit 140, the control unit 130 learns the adjusted sentence priority information.

Accordingly, the control unit 140 may generate an answer reflecting the learned result when generating an answer (or answer sentence) after the learning.

At this time, for convenience of explanation, in the present embodiment, the keyword generation unit 120 and the answer sentence processing unit 140 are described as separate components, but in practice, the control unit 130 may include the keyword generation unit 120 and Note that the answer sentence processing unit 140 may also include functions. In addition, the control unit 130 may be implemented in the form of a computer or server including a CPU (CPU).

Hereinafter, an operation of the parsing system for the book information-based chatbot service will be described with reference to FIGS. 2 to 6.

2 is a flowchart illustrating a control method of a parsing system for a book information-based chatbot service according to a first embodiment of the present invention.

Referring to FIG. 2, the control unit 130 receives keyword information directly from the developer (or system administrator) 21 or the chatbot user 22, or the developer (or system administrator) 21 or chatbot user ( In a manner in which keyword information is automatically extracted from a sentence input by 22), keyword information for preparing (creating) a plurality of answers may be specified or generated (S101).

Here, the keyword information is information reflecting a conversation field, a conversation topic, a conversation purpose, and a conversation tendency, and may be input in a plurality of word forms or in a sentence form (see keywords in FIG. 3).

When keyword information for preparing (creating) a plurality of answers is designated or generated as described above, the control unit 130 displays information of e-books (eg, basic information, table of contents information, detailed information) based on the keyword information. By analyzing, a specific e-book for preparing (creating) the answer and a specific area of the e-book are searched and designated (S102).

In addition, the controller 130 searches for a sentence including at least one or more of the keyword information among text information of the specific e-book and a specific area (eg, word, sentence, paragraph, and chapter) of the e-book. (S103).

At this time, one sentence including at least one or more of the keyword information may be searched, or more sentences may be searched. In addition, the searched sentence may be a short sentence form (eg, a short sentence form) or a longer sentence form (eg, a long sentence form) (see FIG. 3).

Accordingly, the controller 130 automatically generates a plurality of answers (or answer sentences) by combining the searched plurality of sentences and the sentences in a predetermined manner (S104).

In this case, although not specifically shown in the drawings, the generated answers (or answer sentences) may be corrected to an interactive answer sentence (or dialog).

FIG. 3 is an exemplary view of FIG. 2 in which a plurality of sentences including keyword information searched by the control unit are combined into interactive answer sentences and shown in a table form.

As shown in FIG. 3, assuming that four sentences (eg, 1,2,3,4 sentences) are searched, sentences 1,2,3,4 are short sentences, and 5,6 times The sentence is a long sentence sentence created by combining the short sentence sentences. At this time, a long sentence type sentence generated by combining the plurality of sentences may be generated more than illustrated.

In addition, the controller 130 stores the generated plurality of answers (or answer sentences) in the database unit 150 (S105).

Accordingly, the chatbot (not shown) should select the most appropriate answer according to the conversation situation with the user among a plurality of answers (or response sentences) stored in the database unit 150 to conduct a conversation. However, as the current state, the chatbot cannot know what is the most appropriate answer for each situation (that is, according to the intention of the user's question) among the plurality of answers (or answer sentences) stored in the database unit 150, resulting in As a result, the chatbot (not shown) has the same result as selecting any one of the answers (or answer sentences) stored in the database unit 150 in a random manner.

Therefore, the control unit 130, in order to enable the chatbot (not shown) to select the most appropriate answer for each situation (ie, according to the intention of the user's question), an answer (or answer) stored in the database unit 150 Sentence).

4 is a flowchart for explaining a control method of a parsing system for a book information-based chatbot service according to a second embodiment of the present invention, and more specifically, the chatbot (not shown) is context-specific (that is, the user's In order to be able to select the most appropriate answer (depending on the intention of the question), additional information about the answers (or answer sentences) stored in the database unit 150 (that is, the chatbot can reflect in selecting the answer) It is a flow chart for explaining a method of updating the database unit 150 by adding additional information).

Referring to FIG. 4, the controller 130 receives the generated plurality of answers (or answer sentences) (S201), and additionally extracts key keywords from the plurality of answers (or answer sentences) (S202). ).

For example, since the plurality of answers (or answer sentences) is a sentence containing at least one keyword information, the searched sentence includes more words (ie, words that can be used as keywords) in addition to the keyword information to form a sentence. have. Therefore, the control unit 130 further extracts at least one key keyword included in the answer (or answer sentence). At this time, the at least one key keyword includes keyword information previously included.

When a key keyword is additionally extracted from a plurality of answers (or answer sentences) generated by the control unit 130 as described above, a rating (that is, priority) of keywords for each answer (or answer sentence) is set, and It is classified into a population and a meaning phrase (S203) (see Fig. 6).

FIG. 6 is an exemplary diagram illustrating updated answer sentences by additionally adding rating and classification information for all keywords extracted from each answer sentence stored in the database unit, and a response priority of each sentence, in FIG. 4.

However, in FIG. 6, the class of the keyword (that is, priority) and the indications for classifying the individual phrase and the semantic phrase are not specifically illustrated, but this may be added in the form of metadata that can be referenced by a chatbot (not shown). to be.

Setting the rating (i.e., priority) to keywords extracted from each answer (or answer sentence) stored in the database unit 150 as described above (i.e., keywords including additional extracted core keywords), and classifying them into individual phrases and semantic phrases When is completed, the database unit 150 is updated by applying the keyword rating and classification for each answer sentence (eg, rating information of the keyword and classification information of the individual phrase and semantic phrase in the form of metadata) (S204). .

However, as described above, additional information (that is, additional information that the chatbot can reflect in selecting an answer) is added to the database unit 150 as, for example, metadata of keyword rating information and classification information of individual phrases and semantic phrases in the form of metadata. Even if it is updated by adding information), it is not possible to guarantee that the adoption rate (that is, the answer in the form of a sentence mainly desired by the user) is high for a response selected by the chatbot (not shown) with reference to the additional information. That is, the adoption rate of the answer may vary according to the conversation tendency of the user.

Therefore, in order to reflect the adoption rate (ie, the answer in the form of a sentence mainly desired by the user), it is necessary to process the response (or answer sentence) stored in the database unit 150 by reflecting the adoption priority.

5 is a flowchart for explaining a control method of a parsing system for a book information-based chatbot service according to a third embodiment of the present invention, and more specifically, the chatbot (not shown) is a sentence form desired by a user (eg: In order to be able to select an answer of a short sentence type, a middle sentence type, a long sentence type, etc.), the database part 150 is updated by reflecting the priority of the answers (or answer sentences) stored in the database part 150 It is a flow chart for explaining the method.

Referring to FIG. 5, the controller 130 accumulates adoption information for each answer (or answer sentence) selected by a chatbot (not shown) (S301).

For example, the adoption information refers to information in which a chatbot (not shown) selects satisfaction or dissatisfaction by a user (or developer or administrator) during a conversation with the user.

When the adoption information for each answer (or response sentence) is accumulated as described above, the control unit 130 sets a priority (that is, the priority for adopting the answer) based on the accumulated adoption information (S302).

When the priority (that is, the priority for adopting the answer) for each answer (or response sentence) is set as described above, the control unit 130 reflects the priority (that is, the priority for adopting the answer), and then the database unit 150 Update, and learn the answer type (eg, short answer type, short sentence type, long sentence type, etc.) with a high priority (that is, the priority for adopting the answer) to reflect the result of learning when generating a subsequent answer (or answer sentence). It is possible to generate an answer in the form (S303) (see FIG. 6).

For example, referring to FIG. 6, among the plurality of answers (or answer sentences) generated by the control unit 130, response priority information for each answer sentence is added in the form of metadata.

As a result, the chatbot (not shown) can select a response and proceed with a conversation by referring to the priority information for adopting the answer, thereby enabling a more accurate and complete conversation with the users.

As described above, the present embodiment is a method of processing the information of the e-book and automatically generating an answer database for the book chatbot service, and then adding additional information in order to reduce the error rate of answering the user's question. By updating the above database, it is possible not only to communicate with users as accurately as possible about newly published books in various fields, but also to make it easy for users to select books that they want to read based on the contents of the conversation. There is.

The present invention has been described above with reference to the embodiment shown in the drawings, but this is only exemplary, and those skilled in the art to which the art pertains may have various modifications and other equivalent embodiments. You will understand the point. Therefore, the technical protection scope of the present invention should be defined by the following claims.

11, 12: Publisher 21: Developer
22: Chatbot user 110: Book information input unit
120: keyword generation unit 130: control unit
131: book information parsing unit 132: answer generating unit
140: answer sentence processing unit 150: database unit

Claims (15)

A book information input unit that collects or receives e-books, extracts at least one file necessary for analysis or search of each e-book, and transmits it to the control unit;
By parsing the file received from the book information input unit, searching for words, sentences, paragraphs, and chapters included in each e-book based on the basic information, table of contents information, and detailed information of each e-book, The control unit that automatically generates a plurality of response sentences that can be communicated through a chatbot on a specific subject in a book in a specific field; And
Parsing system for a book information-based chatbot service comprising a; database unit for storing a plurality of response sentences generated by the control unit.
According to claim 1, The control unit,
A book information parsing unit parsing a file received from the book information input unit and analyzing basic information, table of contents information, and detailed information of each e-book; And
Based on the basic information, table of contents information, and detailed information of each parsed e-book, words, sentences, paragraphs, and chapters are searched for a specific field designated by an administrator in advance or designated by a user during a chatbot conversation. A parsing system for a book information-based chatbot service, comprising: an answer generation unit that automatically generates a plurality of response sentences with various sentence types that can communicate through a chatbot on a specific subject.
According to claim 1, The control unit,
It further includes a keyword generator for designating or generating keyword information for generating the plurality of response sentences,
The keyword information,
Information that reflects the field of conversation, the topic of conversation, the purpose of conversation, and the nature of conversation
A parsing system for a book information-based chatbot service, characterized in that it is input in the form of multiple words or sentences from a system administrator or a chatbot user.
According to claim 1, The control unit,
The response sentence processing unit further adjusts the priority of the response sentence desired by the chatbot user among response sentences stored in the database unit for each response sentence.
The response sentence most desired by the chatbot user is,
A parsing system for book information-based chatbot services, characterized in that the response sentences with a high hit rate or the response sentences with a high number of times adopted by the chatbot user.
The method of claim 4, wherein the control unit,
To help the chatbot choose the answer sentence,
Updates the database section that adds the ratings for keywords of each answer sentence stored in the database section as metadata as additional information,
The keyword rating,
A parsing system for a book information-based chatbot service, characterized in that a keyword is ranked as a higher priority in order of object, subject, verb, and bore, frequently mentioned words.
The method of claim 4, wherein the control unit,
When priority information of response sentences stored in the database unit is adjusted, the database unit is updated, and
After learning the types of response sentences in which the priority information is adjusted,
A parsing system for a book information-based chatbot service characterized by generating an answer sentence reflecting the learned result when generating a response sentence.
According to claim 1, The control unit,
A parsing system for a book information-based chatbot service, characterized in that it is implemented in the form of a computer or server containing a CPU.
The controller of the parsing system for the book information-based chatbot service designates or generates keyword information for generating a plurality of response sentences;
Analyzing, by the control unit, information on e-books based on the keyword information, searching and designating a specific e-book and a specific area of the e-book to generate the response sentence;
Searching for a sentence including at least one of the keyword information among text information of the specific e-book and a specific specific area of the e-book; And
The control unit generates a plurality of response sentences by combining the retrieved plurality of sentences and the sentences in a predetermined manner, and then corrects the dialogues and stores them in a database unit. Control method of parsing system for service.
The method of claim 8, wherein the keyword information,
Information that reflects the field of conversation, the topic of conversation, the purpose of conversation, and the nature of conversation
Control method of the parsing system for book information-based chatbot service, characterized in that it is input in a plurality of word forms or sentence form through the system and the administrator or chatbot user.
The method of claim 8, wherein the information of the e-books,
A method of controlling a parsing system for a book information-based chatbot service comprising basic information, table of contents information, and detailed information.
The method of claim 8, wherein the sentence including at least one or more of the keyword information,
At least one or more sentences can be searched, and
The searched sentence is a short sentence type or includes a longer sentence type. A method of controlling a parsing system for a book information-based chatbot service.
The method of claim 8,
Extracting key keywords from the plurality of answer sentences stored in the database unit by the control unit;
Setting, by the control unit, a keyword rating for each answer sentence, and classifying the keyword into a subject phrase and a meaning phrase; And
Parsing system for a book information-based chatbot service further comprising the step of performing an update to add additional information generated in the form of metadata for each response sentence keyword rating and classification information in the form of a database. Control method.
The method of claim 12, wherein the additional information,
A control method of a parsing system for a book information-based chatbot service, characterized in that the chatbot is information that can be reflected in selecting a response suitable for a situation among a plurality of response sentences stored in the database unit.
The method of claim 8,
Accumulating adoption information for each sentence of the response selected by the chatbot by the control unit;
Setting, by the control unit, a priority of answer adoption based on the accumulated adoption information; And
The control unit updates the database unit by reflecting the priority of adopting the answer, and after learning the answer form having the highest answer adoption priority, generates a response sentence in a form that reflects the learned result when generating a subsequent answer sentence. Step; Control method of the parsing system for a book information based chatbot service further comprising a.
15. The method of claim 14, The adoption information,
A control method of a parsing system for a book information-based chatbot service characterized in that the chatbot is information that a user or a system administrator has selected satisfaction or dissatisfaction during a conversation with the user.

Images