KR101677859B1 - Method for generating system response using knowledgy base and apparatus for performing the method - Google Patents

Method for generating system response using knowledgy base and apparatus for performing the method Download PDF

Info

Publication number
KR101677859B1
KR101677859B1 KR1020150126046A KR20150126046A KR101677859B1 KR 101677859 B1 KR101677859 B1 KR 101677859B1 KR 1020150126046 A KR1020150126046 A KR 1020150126046A KR 20150126046 A KR20150126046 A KR 20150126046A KR 101677859 B1 KR101677859 B1 KR 101677859B1
Authority
KR
South Korea
Prior art keywords
knowledge
input sentence
system response
name
extracted
Prior art date
Application number
KR1020150126046A
Other languages
Korean (ko)
Inventor
이근배
한상도
Original Assignee
포항공과대학교 산학협력단
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 포항공과대학교 산학협력단 filed Critical 포항공과대학교 산학협력단
Priority to KR1020150126046A priority Critical patent/KR101677859B1/en
Application granted granted Critical
Publication of KR101677859B1 publication Critical patent/KR101677859B1/en

Links

Images

Classifications

    • G06F17/30604—
    • G06F17/271—
    • G06F17/278—

Landscapes

  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Machine Translation (AREA)

Abstract

Disclosed are a method of generating a system response using a knowledge base and an apparatus for performing the method. The method of generating a system response suggested in the present invention comprises the following steps: extracting a name of an object from an input sentence upon receipt of the input sentence; extracting knowledge related to the name of an object from a knowledge base constructed in advance; and generating a system response for the input sentence by replacing a system response candidate extracted in advance so as to reflect the knowledge extracted from the knowledge base constructed in advance. Accordingly, various system responses may be generated with respect to speech of users, and user satisfactory for a conversation system may be enhanced.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a system response generation method using a knowledge base,

The present invention relates to an interactive system, and more particularly, to a method for generating a system response suitable for user utterance using a knowledge base and an apparatus for performing the system response.

BACKGROUND ART [0002] With the recent development of information processing technology, utilization of a dialogue system (Dialogue System) that provides interaction between a user terminal such as a smart phone, a tablet PC, a PDA (Personal Digital Assistant), and a smart home appliance has been popularized.

Currently, conversation systems are mainly used for purpose-oriented technologies that provide specific services to users and chat technologies that provide daily conversations. However, as the age of big data comes, conversation systems are becoming more popular in contents search, intelligent robots, Researches are being actively carried out for application to a wide range of technical fields such as PC, telematics, and home network.

The conversation system is required to have a technique for generating a system response suitable for user utterance.

Thus, conventionally, the intention of the user utterance is analyzed based on the corpus, the response example corresponding to the utterance intention of the user is extracted from the pre-constructed response sample database, and the extracted response example is compared with the context, We generated a system response suitable for user utterance by substituting the entity name for the purpose.

Specifically, for example, when a user utters a sentence of "I like Kim Yu-Na", a response example suitable for user utterance is extracted from a pre-constructed response sample database, and the object name of the response example is information By replacing it with 'Kim Yu-na', a system response such as "I like Kim Yu-na" or "Why do you like Kim Yu-na?"

The above-described conventional techniques generate a system response so that the user's uttered contents are included, thereby enhancing the realism of the conversation, but there is a limitation in that the system response that can be generated is simple and limited.

In addition, since the information included in the user utterance is referred to again through the system response, the conversation between the user and the conversation system proceeds monotonically, which lowers the satisfaction of users using the conversation system.

SUMMARY OF THE INVENTION An object of the present invention is to provide a system response generation method and apparatus capable of generating various system responses to a user utterance by using a knowledge base.

It is another object of the present invention to provide a system response generation method and apparatus capable of improving user satisfaction with an interactive system by generating various system responses to a user utterance using a knowledge base.

According to an aspect of the present invention, there is provided a method for generating a system response, the method comprising: extracting a named entity from an input sentence when an input sentence is received, Extracting knowledge associated with the entity name from a pre-established knowledge base, and generating a system response to the input sentence by replacing the system response candidate extracted in advance so as to reflect the knowledge extracted from the pre-constructed knowledge base .

Here, the step of extracting the object name from the input sentence includes analyzing the input sentence based on at least one of a part of speech tagger, a parser, a triple extractor, The part of speech information of each word constituting the sentence and the relationship information between the words can be extracted.

Here, the step of extracting the object name from the input sentence includes at least one of a name, a place name, an institution name, a name of object, and a time among the words constituting the input sentence based on the part of speech information and the relationship information of each word constituting the input sentence Can be extracted as the object name.

Here, the prebuilt knowledge base is a database structured by storing a tremendous amount of knowledge generated by an expert of various fields or a plurality of users in a triple form, and each knowledge is identified by the entity name Knowledge can be linked based on the type to which each knowledge belongs or the property each knowledge has.

The step of extracting the knowledge associated with the object name includes extracting knowledge having the same object name as the object name extracted from the input sentence in the pre-constructed knowledge base, extracting a plurality of knowledge And extracting knowledge associated with the input sentence from among the extracted knowledge.

Here, the step of selecting the knowledge associated with the object name of the input sentence includes a step of selecting the knowledge similarity between the object name extracted from the input sentence and the personal information of the user who inputs the input sentence, It is possible to determine the knowledge to be reflected in the system response among the plurality of knowledge extracted based on the calculated association score.

Here, the response template corresponding to the intention of the input sentence among the plurality of response templates stored in the previously constructed response candidate database may be selected as the system response candidate extracted in advance.

Here, generating a system response to an input sentence may generate a system response to the input sentence by replacing the response template with the knowledge constituting the input sentence and the knowledge selected from the prebuilt knowledge base .

According to another aspect of the present invention, there is provided an apparatus for generating a system response, the apparatus comprising: an object name extracting unit, which is implemented by an interactive system and extracts an object name from an input sentence when an input sentence is received; And a system response generator for generating a system response to the input sentence by replacing the system response candidate extracted in advance so as to reflect the knowledge extracted from the knowledge base constructed in advance and a knowledge extraction unit for extracting knowledge associated with the object name from the knowledge base .

According to the system response generating method and apparatus according to the embodiment of the present invention as described above, various system responses to user utterances can be generated by using the knowledge base.

Accordingly, user satisfaction with the conversation system can be improved, and the conversation system can be easily applied to a wide variety of technical fields.

1 is a flowchart illustrating a system response generation method using a knowledge base according to an embodiment of the present invention.
FIG. 2 is a diagram for explaining extraction of an object name in an input sentence according to an embodiment of the present invention.
FIG. 3 is a flowchart illustrating extraction of knowledge associated with an entity name extracted from an input sentence in a knowledge base according to an embodiment of the present invention.
4 is an exemplary diagram illustrating generating a system response to reflect knowledge extracted from a knowledge base according to an embodiment of the present invention.
5 is a block diagram illustrating an interactive system in which a system response generating device using a knowledge base according to an embodiment of the present invention is implemented.

While the invention is susceptible to various modifications and alternative forms, specific embodiments thereof are shown by way of example in the drawings and will herein be described in detail. It is to be understood, however, that the invention is not to be limited to the specific embodiments, but includes all modifications, equivalents, and alternatives falling within the spirit and scope of the invention. Like reference numerals are used for like elements in describing each drawing.

The terms first, second, A, B, etc. may be used to describe various elements, but the elements should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be referred to as a second component, and similarly, the second component may also be referred to as a first component. And / or < / RTI > includes any combination of a plurality of related listed items or any of a plurality of related listed items.

It is to be understood that when an element is referred to as being "connected" or "connected" to another element, it may be directly connected or connected to the other element, . On the other hand, when an element is referred to as being "directly connected" or "directly connected" to another element, it should be understood that there are no other elements in between.

The terminology used in this application is used only to describe a specific embodiment and is not intended to limit the invention. The singular expressions include plural expressions unless the context clearly dictates otherwise. In the present application, the terms "comprises" or "having" and the like are used to specify that there is a feature, a number, a step, an operation, an element, a component or a combination thereof described in the specification, But do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, or combinations thereof.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Terms such as those defined in commonly used dictionaries are to be interpreted as having a meaning consistent with the contextual meaning of the related art and are to be interpreted as either ideal or overly formal in the sense of the present application Do not.

Hereinafter, preferred embodiments according to the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 is a flowchart illustrating a method of generating a system response using a knowledge base according to an embodiment of the present invention. FIG. 2 is a diagram illustrating extraction of an object name from an input sentence according to an exemplary embodiment of the present invention.

FIG. 3 is a flowchart illustrating extraction of knowledge associated with an object name extracted from an input sentence in a knowledge base according to an embodiment of the present invention. FIG. 4 is a flowchart illustrating a method of extracting knowledge extracted from a knowledge base according to an embodiment of the present invention ≪ / RTI > FIG.

Referring to FIG. 1, a system response generating method using a knowledge base may be performed by a system response generating apparatus implemented in an interactive system.

Here, the conversation system includes a user terminal such as a smart phone, a tablet PC, a PDA (Personal Digital Assistant), a navigation device, a notebook computer, a smart home appliance, and a system robot, The present invention is not limited thereto and can be easily applied to various devices and technology fields requiring interaction with a user.

The conversation system must implement a technique that generates a system response corresponding to the user utterance. However, the conventional system response generating technology has a limitation in that the system response that can be generated for the user utterance is simple and limited, and the conversation between the user and the conversation system proceeds monotonously, and the satisfaction of the user using the conversation system is low.

In order to overcome the limitations of the prior art described above, the present invention proposes a system response generation method that extracts knowledge related to user utterance in a knowledge base and reflects it in a system response.

The system response generation method proposed in the present invention includes a step S100 of extracting an object name from an input sentence, a step S200 of extracting knowledge associated with an object name from the knowledge base, and a step S200 of reflecting the knowledge extracted from the knowledge base, To generate a system response (S300).

As the input sentence is received, a named entity can be extracted from the received input sentence (S100).

Herein, the input sentence means a sentence which is recognized by the user and converted into text, but is not limited thereto, and may include text information inputted by the user for the application installed in the user terminal or for use of the online service.

In order to extract an object name from an input sentence, first, an input sentence is analyzed by analyzing an input sentence based on at least one of a part of speech tagger, a parser, a triple extractor, It is possible to extract the relation information between the parts-of-speech information of each word and the word constituting the input sentence.

In this case, the part-of-speech information may mean information such as a noun, a pronoun, an investigation, an investigation, a verb, an adjective, an adjective, an adverb, and an admiration classified according to function, form or meaning of each word constituting the input sentence, Information may mean information such as a subject, a descriptor, an object, a bore, a modifier, and the like classified according to a role and a dependency relationship of each word constituting the input sentence in the input sentence, but is not limited thereto.

Named entities can be extracted from the words constituting the input sentence based on the extracted parts information and the relation information between the words constituting the input sentence by analyzing the input sentence.

Specifically, words having relation information of 'subject', 'object', and 'bore' can be extracted as the object name having the part of speech information of 'noun' among the words constituting the input sentence. At this time, the extracted entity name may mean a word representing a person, a place name, an institution name, a name of object, a time, etc., but is not limited thereto.

Specifically, for example, when the input sentence 10 of "I like Kim-yuna" is received as shown in FIG. 2, the words' I ',' like ',' Kim- yuna 'can extract the parts-of-speech information (11)' pronoun ',' verb ', and' noun 'for each. In addition, the role and dependency relation of the words constituting the input sentence 10 are analyzed in the input sentence 10, and the 'I' is the 'subject', the 'like' is the 'narrative', and the 'Kim-yuna' The relation information 13 such as 'object' can be extracted.

Thus, the word 'Kim-yuna' having the part of speech information 11 as a 'noun' and the relation information 13 'object' can be extracted as the object name 15.

In this example, the object name is extracted using the parts-of-speech information and the relationship information of the words constituting the input sentence. However, the present invention is not limited to this, and it is possible to perform labeling by using dictionary-based matching technique or corpus collected in advance. And then extract the object name from the input sentence using a probability-based language model such as word n-gram or the like.

Once the entity name is extracted from the input sentence, the knowledge associated with the entity name can be extracted from the pre-established knowledge base (S200).

Here, the knowledge base is a database structured and stored for each of a large amount of knowledge generated by experts in various fields or a plurality of users, and may represent a DBpedia, a freebase, and the like But is not limited thereto.

The knowledge base stores a vast amount of knowledge in the form of entity-type-entity-entity-entity-property-entity triples . In this case, an entity refers to unique information indicating each knowledge so as to identify a vast amount of knowledge stored in the knowledge base, a property means a property or property possessed by each knowledge, (type) can mean pre-set information to classify a vast amount of knowledge stored in a knowledge base by grouping similar knowledge of attributes.

For example, if knowledge in a knowledge base has a type of 'figure skater', such as' Kim Yu-na ', it will be linked to' Maeda Asada ', an intellectual with the same type as' Kim Yu- And can be stored in a triple-type structure. As such, the vast amount of knowledge stored in the knowledge base can be linked through types or attributes.

In order to extract the knowledge associated with the entity name extracted from the input sentence in the knowledge base, a step S210 of extracting knowledge having an entity name identical to the entity name of the input sentence as shown in FIG. 3 (S210) (S220) of extracting knowledge from the input sentence, and sorting knowledge associated with the input sentence among the extracted knowledge (S230).

Specifically, for example, when the input sentence 10 is analyzed and the object name 15 'Kim-yuna' is extracted as shown in FIGS. 2 and 4 (a) The knowledge having the same object name as the object name 'Kim-yuna' can be extracted (S210).

The knowledge base extracts the knowledge associated with the 'figure skater' which is the type of the object name 'Kim-yuna', or extracts the knowledge associated with the attributes' Kim-yuna ',' coach ',' birthday ',' &Quot; (S220). ≪ / RTI >

Among the plurality of knowledge extracted from the knowledge base, knowledge having a relatively high correlation with the input sentence can be selected (S230). Specifically, it is possible to calculate the semantic similarity between each of the knowledge extracted from the input sentence and each of a plurality of knowledge extracted from the knowledge base, or calculate the similarity degree from the knowledge base based on the number or frequency of use of each of the plurality of knowledge, The association scores for each of the extracted knowledge items are calculated and the knowledge to be reflected in the system response among the plurality of knowledge items can be determined using the calculated association scores.

In particular, in calculating the association score for each of a plurality of knowledge extracted from the knowledge base to determine knowledge to be reflected in the system response, a system response suitable for each user can be generated by reflecting the user's personal information and interest information.

When the knowledge associated with the input sentence is extracted from the knowledge base, a system response to the input sentence can be generated by replacing the system response candidate extracted in advance so as to reflect the knowledge (S300).

Here, the response candidate corresponding to the intention of the input sentence among the plurality of response templates stored in the pre-established response candidate database may be selected as the system response candidate.

If the input sentence 10 of "I like Kim-yuna" is received from the user as shown in FIG. 4 (a), the intention of the user to utter the input sentence 10 becomes the question, greeting, Or the like can be extracted. For this, probabilistic language models such as word n-grams can be used after labeling through corpus collected beforehand, but not limited thereto.

If the intention of the input sentence 10 " I like Kim-yuna " is extracted by self introduction, among the plurality of response templates stored in the pre-established response candidate database, as shown in FIG. 4C, Template (30) "Do you [verb] [knowledge], too?" Can be selected. At this time, the response template 30 is composed of the contents of the input sentence 10 and a slot that can reflect the knowledge 20 extracted from the knowledge base.

Therefore, the [verb] slot of the response template 30 is replaced with 'like' corresponding to the verb among the words constituting the input sentence 10 as shown in FIG. 4 (a) Slot, the system response (40) "Do you like Asada Mao, too?" For the input sentence 10 is generated by replacing the slot with the knowledge (20) 'Asada Mao' extracted from the knowledge base as shown in FIG. can do.

As described above, the present invention can provide various system responses as compared with the conventional system response generation technology by generating the system response by reflecting both the words constituting the input sentence and the knowledge extracted from the knowledge base, User satisfaction is expected to be improved.

5 is a block diagram illustrating an interactive system in which a system response generating device using a knowledge base according to an embodiment of the present invention is implemented.

Referring to FIG. 5, the system response generating apparatus 100 may be implemented in the interactive system 200.

Here, the conversation system 200 includes a user terminal such as a smart phone, a tablet PC, a PDA (Personal Digital Assistant), a navigation device, a notebook computer, a smart home appliance, and a system robot, The terminal may be mounted on a server connected to a wired / wireless network, but the present invention is not limited thereto and can be easily applied to various devices and technical fields requiring interaction with a user.

The dialog system 200 includes a user utterance understanding unit 210, a dialogue management unit 220, a natural language generation unit 230, and a response candidate database 240. In particular, in the present invention, a system response generating apparatus 100 for extracting knowledge related to user utterance in a knowledge base and reflecting it in a system response may be implemented in the interactive system 200.

First, the system response generation apparatus 100 proposed by the present invention may include an entity name extraction unit 110, a knowledge extraction unit 120, a system response generation unit 130, and a knowledge base 140.

The entity name extraction unit 110 may extract a named entity from the received input sentence as the input sentence 10 is received. Herein, the input sentence 10 means a sentence which is converted into text by recognizing a voice uttered by the user, but is not limited to this, and may include text information inputted by the user for application installed in the user terminal or for use of the online service have.

More specifically, the object name extracting unit 110 analyzes an input sentence based on at least one of a part of speech tagger, a parser, a triple extractor, and a speech classifier, The relationship information between the parts of speech information constituting each word and the words constituting the input sentence are extracted and based on this, words expressing the name, place name, institution name, object name, time, named entity.

The knowledge extraction unit 120 may extract knowledge related to the entity name from the knowledge base 140 constructed in advance as the entity names included in the input sentence 10 are extracted from the entity name extraction unit 110. [

Here, the knowledge base 140 refers to a database structured and stored with a large amount of knowledge generated by experts in various fields or by a plurality of users.

The knowledge base 140 may store a vast amount of knowledge in the form of entity-type-entity-entity-entity-property-entity triple . In this case, an entity refers to unique information indicating each knowledge so as to identify a vast amount of knowledge stored in the knowledge base, a property means a property or property possessed by each knowledge, (type) can mean pre-set information to classify a vast amount of knowledge stored in a knowledge base by grouping similar knowledge of attributes. Thus, a vast amount of knowledge stored in a knowledge base can be linked through types or attributes.

That is, the knowledge extraction unit 120 extracts knowledge having the same entity name as the entity name of the input sentence in order to extract the knowledge associated with the entity name extracted from the input sentence, extracts a plurality of knowledge associated with the extracted knowledge, The knowledge associated with the input sentence can be selected from the plurality of knowledge.

In particular, the knowledge extraction unit 120 extracts a plurality of knowledge extracted from an input sentence and a plurality of knowledge extracted from a knowledge base, in order to select knowledge that is relatively related to the input sentence among a plurality of knowledge extracted from the knowledge base , Or calculates the association score for each of a plurality of knowledge extracted from the knowledge base based on the number or frequency of use of each of the plurality of knowledge items in the generation of the system response, The knowledge to be reflected in the system response can be determined from a plurality of knowledge.

In addition, in calculating the association score for each of a plurality of knowledge extracted from the knowledge base to determine knowledge to be reflected in the system response, a system response suitable for each user can be generated by reflecting the user's personal information and interest information.

The system response generating unit 130 may generate a system response to the input sentence by replacing the extracted system response candidate so that the knowledge extracted from the knowledge base 140 is reflected.

To this end, the user utterance understanding unit 210 can grasp the intention from the input sentence 10 input by the user. In detail, a semantic frame for the input sentence 10 is extracted based on at least one of a part of speech tagger, a parser, a triple extractor, and a speech classifier, The user can extract the intention of uttering the input sentence 10 using at least one of a question, a greeting, and a self introduction.

The dialogue management unit 220 can select a system response candidate corresponding to the intention of the input sentence 10 extracted from the user utterance understanding unit 210 from the response candidate database 240 previously constructed. At this time, the pre-established response candidate DB 240 stores various response templates according to the intention of user utterance.

The system response generation unit 130 receives the response template corresponding to the intention of the input sentence 10 in cooperation with the dialogue management unit 220 and receives the response sentence corresponding to the words constituting the input sentence 10 and the words selected from the knowledge base 140 The system response for the input sentence 10 can be generated by replacing the response template with knowledge.

The natural language generation unit 230 may generate the system response generated by the system response generation unit 130 as a natural language and output the generated natural language as a system response 40 to the user.

The system configuration of the dialog system 200 according to the embodiment of the present invention is shown in the system response generator 100, the user utterance understanding unit 210, the dialogue management unit 220, the natural language generation unit 23, 240, and the configuration of the system response generating apparatus 100 is described as the entity name extracting unit 110, the knowledge extracting unit 120, the system response generating unit 130, and the knowledge base 140, At least two of the constituent parts may be combined to form one constituent part or one constituent part may be divided into a plurality of constituent parts to perform a function and the case of the integrated and separate embodiments of each constituent part is not deviated from the essence of the present invention Are included in the scope of the present invention.

In addition, the operation of the system response generating apparatus 100 and the interactive system 200 according to the embodiment of the present invention can be implemented by a computer-readable program or code on a computer-readable recording medium. A computer-readable recording medium includes all kinds of recording apparatuses in which data that can be read by a computer system is stored. In addition, the program or code may be stored and executed in a distributed manner distributed over networked computer systems.

It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit or scope of the present invention as defined by the following claims It can be understood that

10: Input sentence 11: Part of speech information
13: Relationship information 15: Object name
20: Extracted knowledge 30: Response template
40: System response
100: System response generator 110: Object name extractor
120: knowledge extracting unit 130: system response generating unit
140: knowledge base 200: conversation system
210: user utterance understanding part 220: conversation management part
230: natural language generation unit 240: response candidate DB

Claims (16)

A method performed by an apparatus for generating a system response of an interactive system,
Extracting a named entity from the received input sentence as the input sentence is received;
Extracting knowledge associated with the entity name from a pre-established knowledge base; And
Generating a system response for the input sentence by replacing a system response candidate extracted in advance so as to reflect the knowledge extracted from the pre-built knowledge base,
Wherein the knowledge base stores each knowledge identified by the entity name in a triple type structure by linking each knowledge based on a type to which each knowledge belongs or an attribute of each knowledge,
Wherein the knowledge associated with the entity name is knowledge related to the input statement among a plurality of knowledge linked to knowledge having an entity name equal to the extracted entity name in the knowledge base based on the type or the attribute. Way. The method according to claim 1,
Wherein the step of extracting the entity name from the input sentence comprises:
Analyzing the input sentence based on at least one of a part of speech tagger, a parser, a triple extractor, and a speech classifier to extract part of speech information of each word constituting the input sentence, And extracting relationship information between the first and second information. The method of claim 2,
Wherein the step of extracting the entity name from the input sentence comprises:
Extracting a word representing at least one of a name, a place name, an institution name, an object name, and a time among the words constituting the input sentence based on the part of speech information of each word constituting the input sentence and the relation information between the words, Lt; RTI ID = 0.0 > 1, < / RTI > delete delete The method according to claim 1,
The knowledge associated with the input sentence,
Calculating an association score for each of the plurality of knowledge based on the semantic similarity degree with the entity name extracted from the input sentence and the personal information and interest information of the user who has input the input sentence, ≪ / RTI > is determined among a plurality of knowledge. The method of claim 6,
The system response candidate, which is extracted in advance,
Wherein a response template corresponding to the intention of the input sentence is selected from a plurality of response templates stored in a pre-established response candidate database. The method of claim 7,
Wherein generating a system response for the input statement comprises:
Generating a system response for the input sentence by replacing the selected response template with words comprising the input sentence and knowledge selected from the pre-built knowledge base. An apparatus for generating a system response, which is implemented by an interactive system,
An object name extracting unit for extracting a named entity from the received input sentence when an input sentence is received;
A knowledge extraction unit for extracting knowledge associated with the entity name from a pre-established knowledge base; And
And a system response generator for generating a system response to the input sentence by replacing the system response candidate extracted in advance so as to reflect the knowledge extracted from the knowledge base constructed in advance,
Wherein the knowledge base stores each knowledge identified by the entity name in a triple-type structure by linking each knowledge based on a type to which each knowledge belongs or an attribute of each knowledge,
Wherein the knowledge associated with the entity name is knowledge related to the input sentence among a plurality of knowledge connected to knowledge having the same entity name as the entity name extracted from the input sentence in the knowledge base based on the type or the attribute. System response generator. The method of claim 9,
The object-
Analyzing the input sentence based on at least one of a part of speech tagger, a parser, a triple extractor, and a speech classifier to extract part of speech information of each word constituting the input sentence, And extracts relationship information between the plurality of apparatuses. The method of claim 10,
The object-
Extracting a word representing at least one of a name, a place name, an institution name, an object name, and a time among the words constituting the input sentence based on the part of speech information of each word constituting the input sentence and the relation information between the words, The system response generation device comprising: delete delete The method of claim 9,
The knowledge associated with the input sentence,
Calculating an association score for each of the plurality of knowledge based on the semantic similarity degree with the entity name extracted from the input sentence and the personal information and interest information of the user who has input the input sentence, Wherein the system response is determined from a plurality of knowledge. 15. The method of claim 14,
The system response candidate, which is extracted in advance,
Wherein a response template corresponding to the intention of the input sentence is selected from among a plurality of response templates stored in a previously constructed response candidate database. 16. The method of claim 15,
Wherein the system response generating unit comprises:
Wherein the system response generator generates a system response for the input sentence by replacing the selected response template with words comprising the input sentence and knowledge selected from the pre-built knowledge base.
KR1020150126046A 2015-09-07 2015-09-07 Method for generating system response using knowledgy base and apparatus for performing the method KR101677859B1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
KR1020150126046A KR101677859B1 (en) 2015-09-07 2015-09-07 Method for generating system response using knowledgy base and apparatus for performing the method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
KR1020150126046A KR101677859B1 (en) 2015-09-07 2015-09-07 Method for generating system response using knowledgy base and apparatus for performing the method

Publications (1)

Publication Number Publication Date
KR101677859B1 true KR101677859B1 (en) 2016-11-18

Family

ID=57537687

Family Applications (1)

Application Number Title Priority Date Filing Date
KR1020150126046A KR101677859B1 (en) 2015-09-07 2015-09-07 Method for generating system response using knowledgy base and apparatus for performing the method

Country Status (1)

Country Link
KR (1) KR101677859B1 (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101872863B1 (en) * 2016-11-24 2018-07-02 주식회사 와이즈넛 Business support method using a chatting bot
WO2019147039A1 (en) * 2018-01-26 2019-08-01 주식회사 머니브레인 Method for determining optimal conversation pattern for goal achievement at particular time point during conversation session associated with conversation understanding ai service system, method for determining goal achievement prediction probability, and computer-readable recording medium
KR20190131270A (en) * 2018-05-16 2019-11-26 (주)아이브릭스 The syntax grammar rules automatic generation method of understanding user query intention
KR20200122957A (en) * 2019-04-19 2020-10-28 한국과학기술원 Suggestion of evidence sentence for utterance in debate situation
CN112035609A (en) * 2020-08-20 2020-12-04 出门问问信息科技有限公司 Intelligent dialogue method and device and computer readable storage medium
KR20240074484A (en) 2022-11-21 2024-05-28 국립창원대학교 산학협력단 Dialogue system using external knowledge and Dialogue processing method thereof

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20090126043A (en) * 2008-06-03 2009-12-08 한국과학기술원 Automatic construction method and automatic construction system of relations between ontology classes of technical terms using general domain corpus clustering
KR20140062656A (en) * 2012-11-14 2014-05-26 한국전자통신연구원 Spoken dialog management system based on dual dialog management using hierarchical dialog task library

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20090126043A (en) * 2008-06-03 2009-12-08 한국과학기술원 Automatic construction method and automatic construction system of relations between ontology classes of technical terms using general domain corpus clustering
KR20140062656A (en) * 2012-11-14 2014-05-26 한국전자통신연구원 Spoken dialog management system based on dual dialog management using hierarchical dialog task library

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
심효섭 외 2명, ‘Semantic parsing 기반 지식 베이스 질의응답 시스템의 어휘-의미 패턴 질의 템플릿을 통한 보완’, 제26회 한글 및 한국어 정보처리 학술대회 논문집, 2014 *

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101872863B1 (en) * 2016-11-24 2018-07-02 주식회사 와이즈넛 Business support method using a chatting bot
WO2019147039A1 (en) * 2018-01-26 2019-08-01 주식회사 머니브레인 Method for determining optimal conversation pattern for goal achievement at particular time point during conversation session associated with conversation understanding ai service system, method for determining goal achievement prediction probability, and computer-readable recording medium
KR20190131270A (en) * 2018-05-16 2019-11-26 (주)아이브릭스 The syntax grammar rules automatic generation method of understanding user query intention
KR102099670B1 (en) 2018-05-16 2020-04-10 (주)아이브릭스 The syntax grammar rules automatic generation method of understanding user query intention
KR20200122957A (en) * 2019-04-19 2020-10-28 한국과학기술원 Suggestion of evidence sentence for utterance in debate situation
KR102254300B1 (en) * 2019-04-19 2021-05-21 한국과학기술원 Suggestion of evidence sentence for utterance in debate situation
CN112035609A (en) * 2020-08-20 2020-12-04 出门问问信息科技有限公司 Intelligent dialogue method and device and computer readable storage medium
CN112035609B (en) * 2020-08-20 2024-04-05 出门问问创新科技有限公司 Intelligent dialogue method, intelligent dialogue device and computer-readable storage medium
KR20240074484A (en) 2022-11-21 2024-05-28 국립창원대학교 산학협력단 Dialogue system using external knowledge and Dialogue processing method thereof

Similar Documents

Publication Publication Date Title
US11915692B2 (en) Facilitating end-to-end communications with automated assistants in multiple languages
KR101677859B1 (en) Method for generating system response using knowledgy base and apparatus for performing the method
US10936664B2 (en) Dialogue system and computer program therefor
KR101634086B1 (en) Method and computer system of analyzing communication situation based on emotion information
KR20210158344A (en) Machine learning system for digital assistants
KR101309042B1 (en) Apparatus for multi domain sound communication and method for multi domain sound communication using the same
KR101751113B1 (en) Method for dialog management based on multi-user using memory capacity and apparatus for performing the method
EP1901283A2 (en) Automatic generation of statistical laguage models for interactive voice response applacation
CN108538294B (en) Voice interaction method and device
KR20070102267A (en) Dialog management system, and method of managing dialog using example-based dialog modeling technique
KR101627428B1 (en) Method for establishing syntactic analysis model using deep learning and apparatus for perforing the method
JP2015219583A (en) Topic determination device, utterance device, method, and program
KR101410601B1 (en) Spoken dialogue system using humor utterance and method thereof
JP5073024B2 (en) Spoken dialogue device
US20220165257A1 (en) Neural sentence generator for virtual assistants
US8401855B2 (en) System and method for generating data for complex statistical modeling for use in dialog systems
Dyriv et al. The user's psychological state identification based on Big Data analysis for person's electronic diary
Hardy et al. The Amitiés system: Data-driven techniques for automated dialogue
Braunger et al. A comparative analysis of crowdsourced natural language corpora for spoken dialog systems
Henderson et al. Data-driven methods for spoken language understanding
JP5860439B2 (en) Language model creation device and method, program and recording medium
CN113096649B (en) Voice prediction method, device, electronic equipment and storage medium
Pan et al. Type-II dialogue systems for information access from unstructured knowledge sources
Parameswari et al. Voice Based Search Engine for Visually Impaired People.
Hady et al. Unsupervised Active Learning of CRF Model for Cross Lingual Information Extraction.

Legal Events

Date Code Title Description
E701 Decision to grant or registration of patent right
GRNT Written decision to grant