KR101964514B1 - An ambiguous expression analysis device and method considering user status - Google Patents

Abstract

An apparatus and method for analyzing mediocre expression considering a user state are provided. An apparatus for analyzing a spontaneous expression considering a user state according to an embodiment of the present invention includes an external information acquisition module for acquiring status information of a user and chat information of a user as external information; And a median expression processing module for generating a median expression analysis result using the external information and generating response information to be provided to the user using the median expression analysis result.

Description

[0001] The present invention relates to an apparatus and method for analyzing ambiguous expressions,

The present invention relates to an apparatus and a method for analyzing a medial expression in consideration of a user state, and more particularly, to an apparatus and a method for acquiring a state of a user and using the same to analyze a medial expression.

In general, chatbots are services that interact with the user through rules or AI (AI) on the interactive interface, that is, conversation robots. These chatbots are using artificial intelligence, which is represented by AlphaHolder, so that conversation with users and ability to provide necessary information are gradually increasing.

On the other hand, common languages that users use to communicate with chatbots have various homonyms, plural words, analogy, vocabulary, and sentence structures. This linguistic feature is a cause of ambiguity, and therefore there is a problem that the reliability of the chatbot is reduced by the user acquiring the intention and other responses from the chatbot.

For example, there is a problem that the sentence "many words" can be construed as two meanings: that there are a large number of horses due to homonyms and that many words are spoken.

In addition, the sentence 'He likes flowers more than me' means 'He likes flowers more than he likes flowers' and 'He likes flowers more than he likes me' The sentence "Beck is a bear" can be interpreted as two meanings, and the sentence of "Beck is a bear" means that the bear 's name is "Bob" and that "Bob is acting like a bear" have.

KR 2014-0079598 A

According to an aspect of the present invention, there is provided an apparatus and method for analyzing a medullary expression in which a state of a user is acquired and a user's intention about a meditative expression can be grasped I want to.

According to an aspect of the present invention, there is provided an apparatus for analyzing medullary expressions in consideration of a user state. The median expression analyzing apparatus considering the user state includes an external information acquiring module for acquiring status information of a user and chat information of a user as external information; And a median expression processing module for generating a median expression analysis result using the external information and generating response information to be provided to the user using the median expression analysis result.

Wherein the external information acquisition module comprises: a chat information acquisition module for acquiring chat information input by the user; And a user state determination module for obtaining at least one state of the user's body information and surrounding environment information to determine the state of the user.

Wherein the user state determination module comprises: a measurement unit that obtains at least one of information on the user's body and surrounding environment; A measurement information analyzer for analyzing the external information acquired by the measurement unit and generating at least one analysis result of the user's state information; And a user state setting unit for setting a state of the current user using at least one of the information analysis results.

The median expression processing module may include a medial expression analysis module for generating an analysis result of the median expression included in the chat information; And a response information generation module for generating response information to be provided to the user using the result of the analysis of the heuristic expression.

The ambiguous expression analysis module may include a user state storage unit for storing the state of the user among the external information acquired by the external information acquisition module; A chat information analyzer for analyzing chat information of the user among the external information acquired by the external information acquisition module; And the user's state and the chat information to determine the meaning of the ambiguous expression and to select the judgment result.

The heuristic analysis module may further include an accuracy learning unit for further obtaining the response information and the user's chat information with respect to the response information on the result of the cynical expression determination and accumulating the accuracy of the cynical expression determination result; As shown in FIG.

Wherein the external information acquisition module further acquires the interaction information of the user and the interaction information includes at least one of activity information performed by the user using the communication network with another user and information previously stored in the internal storage device by the user It can be one.

And a keyword processing module for extracting a keyword from the interaction information and classifying the extracted keyword in accordance with a predetermined criterion, wherein the keyword and the chat information are further used to identify the user And generate the response information to be provided.

Wherein the keyword processing module comprises: a keyword extracting unit for analyzing the interaction information and extracting a keyword including the meaning of the interaction information; And a keyword classifier for classifying the keywords extracted by the keyword extractor according to a predetermined rule.

According to an aspect of the present invention, there is provided a method for analyzing medial expression considering a state of a user. The method includes the steps of: acquiring status information of a user and chat information of a user as external information; Analyzing the external information according to the type of external information; Determining and selecting a meaning of the chronological expression of the chat information using the external information analysis result; And generating response information to be provided to the user using the determined meaning.

The external information may be at least one of status information of the user's body information and surrounding environment information, and chat information input by the user.

The step of acquiring the external information may further comprise setting a state of the current user using at least one state information of the user's body information and the surrounding environment information.

The step of generating the response information may generate response information corresponding to the user's intention using the meaning of the user state and the abbreviation.

The step of generating the response information may further include accumulating the accuracy of the semantic determination result of the ambiguous expression using the additional chat information of the user after generating the response information and providing the response information to the user can do.

And accumulating the accuracy when the additional chat information of the user is positive and providing response information using another substitute expression when the additional chat information is negative or does not exist.

The external information may further include interaction information of the user, the interaction information may include at least one of activity information that the user performs with the other user using the communication network, and information that the user has previously stored in the internal storage device have.

The step of analyzing the external information may further include extracting a keyword from the interaction information when the external information is the interaction information, and classifying the extracted keyword according to a predetermined criterion.

The step of generating the response information may further generate the response information to be provided to the user by using the keyword and the chat information.

The apparatus and method for analyzing medial expression in consideration of a user state according to an embodiment of the present invention are capable of performing a meditative expression analysis with high accuracy using a user's physical condition and environment information.

In addition, the apparatus and method for analyzing medial expression in consideration of a user state according to an embodiment of the present invention can acquire a user's response to a meditative expression and increase the accuracy of the medullary expression analysis result.

FIG. 1 is a view illustrating a chatbot system including a median expression analyzing apparatus in consideration of a user state according to an embodiment of the present invention.
FIG. 2 is a block diagram illustrating a median expression analyzing apparatus in consideration of a user state according to an embodiment of the present invention.
FIG. 3 is a block diagram illustrating in detail the external information acquisition module of the medial expression analysis apparatus in consideration of the user state according to the embodiment of the present invention.
FIG. 4 is a block diagram illustrating a user state determination module of a median expression analysis apparatus in consideration of a user state according to an embodiment of the present invention.
FIG. 5 is a block diagram illustrating a median expression analysis module among medial expression analysis apparatuses in consideration of a user state according to an embodiment of the present invention.
FIG. 6 is a flowchart illustrating a median expression analysis method in consideration of a user state using a meditative expression analyzing apparatus considering a user state according to an embodiment of the present invention.
Fig. 7 is a flowchart showing in more detail Fig.
FIG. 8 is a flowchart illustrating a method of increasing the accuracy of the median expression in the median expression analysis method considering the user state according to the embodiment of the present invention.
FIG. 9 is a block diagram including a keyword processing unit as an additional configuration for generating response information among the apparatus for analyzing a medial expression in consideration of a user state according to an embodiment of the present invention.
FIG. 10 is a block diagram showing a keyword processor according to an embodiment of the present invention in more detail.
11 is a more detailed flowchart for analyzing chat information of a median expression analysis method in consideration of a user state according to an embodiment of the present invention.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings, which will be readily apparent to those skilled in the art to which the present invention pertains. The present invention may be embodied in many different forms and is not limited to the embodiments described herein. In order to clearly illustrate the present invention, parts not related to the description are omitted, and the same or similar components are denoted by the same reference numerals throughout the specification.

FIG. 1 is a view illustrating a chatbot system including a median expression analyzing apparatus in consideration of a user state according to an embodiment of the present invention. Referring to FIG. 1, a chatbot system 100 including a median expression analysis apparatus considering a user state according to an embodiment of the present invention receives a user's interaction from the interaction providing set 110, And is configured to acquire user information or provide information via communication with the user terminal 150. [ Hereinafter, the median expression analyzing apparatus 130 considering the user state included in the chatbot 130 will be described in the same manner as the chatbot 130 for convenience of explanation.

FIG. 2 is a block diagram illustrating a medial expression analyzing apparatus in consideration of a user state according to an embodiment of the present invention. FIG. 3 is a block diagram of a medial expression analyzing apparatus considering a user state according to an embodiment of the present invention. FIG. 4 is a block diagram illustrating a user state determination module of the medial expression analysis apparatus in consideration of the user state according to an embodiment of the present invention in more detail. FIG. FIG. 8 is a block diagram showing the medial expression analysis module of the medial expression analysis apparatus considering the state in more detail. Hereinafter, a median expression analyzing apparatus considering a user state according to an embodiment of the present invention will be described in detail with reference to FIG. 2 to FIG. 5. For convenience of explanation, a medial expression analyzing apparatus, Device.

Referring to FIG. 2, the meditative expression analyzing apparatus 130 considering the user state according to an embodiment of the present invention acquires the state of the user and the chat information of the user as external information, And generates response information provided to the user using the result of the analysis of the median expression. At this time, the apparatus 130 for analyzing the hull expression includes an external information acquisition module 210 and an idiomatic expression processing module 230.

The external information acquisition module 210 acquires the status information of the user and the chat information of the user as the external information. The external information acquisition module 210 may acquire the individual information by using the chat information acquisition module 310 and the user status And a determination module 330. [

The chat information acquisition module 310 acquires the user's chat information. The chat information acquisition module 310 may acquire chat information input by a user through an input device directly connected to the user terminal 150 or the analysis unit 130. [ At this time, the chat information input by the user may be information that can be converted into a digital signal such as text, voice, video, sound, and image.

The user status determination module 330 obtains the status information of the user and determines the status of the user. The user state determination module 330 may obtain at least one state of the user's body information and surrounding environment information, and may determine the state of the user using the acquired state. In this case, the status information of the user acquired by the user status determination module 330 is preferably the status information when the user generates chat information. To this end, the user status determination module 330 includes a measurement unit 410, An information analysis unit 430 and a user state setting unit 450. [

The measuring unit 410 acquires the user's body information and surrounding environment information. The measuring unit 410 acquires the user's body information and surrounding environment information using various sensors, and the measuring unit 410 may be included in the user terminal 150 at this time.

The user's body information acquired by the measuring unit 410 may be heart rate, blood pressure, oxygen saturation, sweat, eye movement, pupil size, and the like, and the ambient environment information may include ambient noise, noise, location, time, And so on. To this end, the measuring unit 410 may include at least one sensor.

The measurement information analysis unit 430 analyzes the information and generates analysis results. The measurement information analyzing unit 430 analyzes at least one of the user's body information and the surrounding environment information obtained by the measuring unit 410 to generate at least one user's status information.

The user state setting unit 450 obtains the user state information and sets the state of the current user. The user state setting unit 450 obtains the user state information generated by the measurement information analyzing unit 430 and analyzes what state the current user is by analyzing the user state information. For example, Lt; / RTI >

Meanwhile, the median expression processing module 230 of the median expression analyzing apparatus 130 considering the user state according to an embodiment of the present invention uses the external information acquired from the external information acquiring module 210, Generates the analysis result, and generates the response information provided to the user using the result of the analysis of the median expression. For this, the median expression processing module 230 includes a median expression analysis module 231 and a response information generation module 233.

The clumsy expression analysis module 231 generates an analysis result of the chronological expression contained in the chat information. The median expression analysis module 231 receives the set state of the user and the chat information acquired by the external information acquisition module 210 to acquire and analyze the medial expression. To this end, the heuristic expression analysis module 231 includes a user state storage unit 510, a chat information analysis unit 530, and an idiomatic expression selection unit 550.

The user state storage unit 510 receives and stores the user state acquired by the user state determination module 330. The user state storage unit 510 may generate and store an independent storage space for each user state using the user state set by the user state determination module 330. [

The chat information analyzing unit 530 receives and analyzes the chat information acquired by the chat information acquiring module 310. The chat information analyzing unit 530 analyzes the chat information to determine whether there is a negative representation. For example, the chat information analyzing unit 530 may use a method of analyzing using morphemes such as n grams.

The chronological expression selecting unit 550 receives the chronological expression acquired by the chat information analyzing unit 530 and selects the meaning of the chronological expression using the state of the user stored in the user state storing unit 510. The chronological expression selecting unit 550 recognizes the meaning of the chronological expression included in the chat information by combining the chat information and the state at the moment when the user inputs the chat information.

For example, if the chat information is 'night', the chronological expression selecting unit 550 checks the state of the moment when the user inputs the chat information, and if the time is before the sunset, the chronological expression selecting unit 550 It is possible to select the expression corresponding to the user's intention by judging the 'night' of the chat information as a plant rather than a time.

The response information generation module 233 receives the meanings of the meanings selected by the mean expression analysis module 231 and generates response information. The response information generation module 233 generates response information corresponding to the user's chat information using the selected annotation and transmits the generated response information to the user.

The chronological expression analysis module 231 of the chronological expression analyzer 130 considering the user state according to an embodiment of the present invention may further include an accuracy learning unit 570.

The accuracy learning unit 570 further obtains the user's chat information for the response information generated by the response information generation module 233 to determine whether the previously generated response information is correct information. In this case, if the user does not input the additional chat information or inputs the positive chat information, the accuracy learning unit 570 determines that the meaning of the heuristic expression for the state of the user selected by the temporal expression selection unit 550 is correct , And the accuracy of the combination can be increased.

In other words, for example, if the user state B is combined with the A chat information including the chronological expression to derive C of the middle representation and provided to the user, if the user sends positive feedback, the accuracy learning unit 570 calculates the A information And the accuracy of the C derivation formula for the combination of B states can be increased.

Meanwhile, FIG. 6 to FIG. 8 show a method for analyzing the Chinese characters in consideration of the user state using the Chinese character expression analyzing apparatus considering the user state according to an embodiment of the present invention. FIG. 6 is a flow chart showing a method for analyzing medial expression in consideration of a user state using a meditative expression analyzing apparatus considering a user state according to an embodiment of the present invention, FIG. 7 is a flowchart showing in more detail FIG. 6, FIG. 4 is a flowchart illustrating a method of increasing the accuracy of a median expression in a median expression analysis method considering a user state according to an embodiment of the present invention. Hereinafter, a method for analyzing medial expression in consideration of a user state according to an embodiment of the present invention will be described in more detail with reference to FIG. 6 to FIG.

The method 600 for analyzing a spontaneous expression according to an exemplary embodiment of the present invention includes a step S610 of obtaining external information, a step S620 of analyzing external information, A step S630 of determining and selecting the meaning of the chronological expression of the chat information, and a step S640 of generating and providing response information to be provided to the user using the determined meaning.

First, the meditative expression analyzing apparatus obtains status information of a user and chat information of a user as external information (step S610). The median expression analyzing apparatus obtains the status information of the user and the chat information of the user as the outline information. At this time, the state of the user acquired by the apparatus for analyzing the Chinese characters is preferably the user's physical condition.

The chat information may be information input by a user through a user terminal or an input device directly connected to the apparatus for analyzing a narrative expression, and may be information that can be converted into a digital signal such as text, voice, video, sound, and image.

The user status information includes at least one of the user's body information and the environment information, and is preferably status information when the user generates chat information. The user's body information may be heart rate, blood pressure, oxygen saturation, sweat, eye movement, pupil size, etc., and the surrounding information may be ambient noise, noise, position, time, temperature, humidity and illumination.

Next, the median expression analyzing apparatus analyzes external information according to the type of external information (step S620). The median expression analyzing apparatus includes a step S621 for determining whether the user information is user state information for analyzing the external information acquired in the step S610, a step S622 for analyzing the user state information, and a step S623 for analyzing the chat information .

If the external information obtained in step S610 is determined to be the user state information in step S621, the median expression analysis device performs step of analyzing the user state information in step S622. At this time, the median expression analyzing device may analyze at least one of the user's body information and the surrounding environment information acquired at step S610, generate at least one of the analyzed user state information, and set the state of the current user. At this time, the user's state may be an emotional state of the current user as an example.

On the other hand, if the external information is not the user state information in step S621, the implicit expression analysis apparatus determines the external information as the chat information, and analyzes the chat information in step S623. At this time, the apparatus for analyzing the Chinese characters analyzes the chat information to determine the presence of the Chinese word representation. For example, using the method of analyzing the sentence using the morpheme represented by n grams, And can generate the meditative expression information.

Next, the median expression analyzing apparatus judges the meaning of the chronological expression of the chat information by using the external information analysis result and selects it (step S630). The median expression analyzing apparatus obtains the user state information and the median expression information analyzed respectively from the steps S622 and S633, and selects the meaning of the median expression using the obtained medication expression information. The meaning of the median expression can be selected by combining the meditative expression information and the state of the user obtained at the moment when the user inputs the chat information.

For example, when the chat information is 'night', the apparatus for analyzing the Chinese characters confirms the state of the moment when the user inputs the chat information, and when the time is before the sunset, the 'night' By the judgment, the chronological expression analyzing apparatus can select a representation that matches the user's intention.

Finally, the median expression analyzing apparatus includes a step S640 of generating and providing response information to be provided to the user using the selected meaning. The median expression analyzing apparatus can generate response information corresponding to the user's chat information using the median expression selected in step S630 and transmit the generated response information to the user.

Meanwhile, the median expression analysis method considering the user state according to an embodiment of the present invention may further include a step of learning the accuracy to increase the reliability of the median expression analysis result. Referring to FIG. 8, the method of analyzing a spontaneous expression according to an embodiment of the present invention includes the steps of determining whether the apparatus for analyzing a spoken language has acquired user-added chat information after performing step S650 (S810) Step S820 of determining whether the information is positive, step S830 of accumulating the accuracy of the result of determination of the idiomatic expression, and step S840 of providing response information of another idiomatic expression.

First, if the meditative expression analyzing apparatus provides the response information to the user through step S650, it is determined whether the additional user's chat information is acquired (step S810). At this time, if the chat information acquired by the apparatus for analyzing the Chinese phrase is information related to the response information, the apparatus for analyzing the Chinese characters performs step S820. The ambiguity analyzing apparatus judges whether the additional chat information is positive information if the additional chat information is information related to the response information (step S820). At this time, if the additional chat information is positive information, the median expression analyzing device performs step S830 of accumulating the accuracy of the medial expression determination result for generating the response information of step S650. In addition, the median expression analyzing apparatus may accumulate the accuracy by performing step S830 if the additional chat information is not acquired in step S810 or the additional chat information is not related to the response information.

On the other hand, if it is determined in step S820 that the additional expression analyzing apparatus is not positive, the implicit expression analyzing apparatus generates response information using another ambiguous expression and provides the generated response information to the user (S840) The accuracy of the analysis of the median expression can be improved.

Meanwhile, the apparatus and method for analyzing medial expression in consideration of a user state according to an embodiment of the present invention may acquire and analyze user activity data to generate response information and provide it to a user.

FIG. 9 is a block diagram including a keyword processing unit, which is a further configuration for generating response information among the apparatus for analyzing medial expression in consideration of a user state according to an embodiment of the present invention. In more detail. Hereinafter, an additional configuration for generating response information using FIG. 2, FIG. 9, and FIG. 10 will be described in more detail.

Referring to FIG. 9, the heuristic expression analysis module 231 according to the embodiment of the present invention may further include a keyword processing unit 910. The keyword processing unit 910 extracts and classifies keywords. At this time, the keyword processing unit 910 may extract the keyword from the interaction information, and classify the extracted keyword according to a predetermined criterion. For this, the median expression analysis apparatus 130 considering the user state according to an embodiment of the present invention further includes an interaction information acquisition module (not shown) in the external information acquisition module 210, From the user terminal 110 to the user's interaction or information stored in the internal storage device or the cloud of the user terminal 150 as the interaction information.

In other words, the interaction information may be activity information that a user performs using a communication network with another user, or storage information that a user inputs to the user terminal 150 and stores the information. In addition, the interaction information may be in a form that can be converted into a digital signal such as text, voice, photograph or image.

On the other hand, the keyword processing unit 910 receives the interaction information obtained from the interaction information acquisition module (not shown), extracts keywords from the interaction information, and classifies the extracted keywords according to a predetermined criterion. At this time, the keyword processing unit 910 may include a keyword extracting unit 1010 and a keyword sorting unit 1030 shown in FIG. 10 for extracting and classifying keywords.

The keyword extracting unit 1010 extracts keywords from the interaction information. The keyword extracting unit 1010 may extract the keyword included in the interaction information by analyzing the interaction information received from the external information acquiring module 210. [ At this time, the keyword may be formed to include the meaning of the interaction information.

For example, when the interaction information is an image, the keyword extracting unit 1010 can extract a keyword of an entire image, a keyword of an object in an image, a keyword of a person in an image, a keyword of an image atmosphere, and the like.

The keyword extracting unit 1010 can grasp the meaning of the image in order to acquire the keyword of the entire image. For example, when an image photographed inside the restaurant is obtained as the interaction information, the keyword extracting unit 1010 can extract a restaurant from keywords of the entire image in the interaction information.

In addition, the keyword extracting unit 1010 may extract a keyword of an image from an object in the image. If a chair is included in the image obtained by the interaction information, the keyword extracting unit 1010 can extract the chair included in the image as one of the keywords of the image. At this time, the keyword of the object in the image may be obtained from all things that can be identified in the image as much as the setting.

Also, the keyword extracting unit 1010 may extract a keyword of an image from a person in the image. When a person is included in the image obtained by the interaction information, the keyword extracting unit 1010 analyzes the emotion through a facial expression of the person included in the image, and extracts the analyzed emotion as one of the keywords of the image. At this time, the keyword of the person in the image may be obtained from all the persons who are identifiable in the image as much as the setting.

In addition, the keyword extracting unit 1010 may extract keywords related to the atmosphere of the image. The keyword extracting unit 1010 analyzes color, saturation, brightness, etc. of the image acquired by the interaction information. Accordingly, the keyword extracting unit 1010 can determine which atmosphere the image has, and at this time, the atmosphere of the image can be extracted as a keyword for the atmosphere of the image.

On the other hand, if the text information is included in the interaction information, the keyword extracting unit 1010 may extract the keyword from the text by analyzing the text. At this time, the keyword extracting unit 1010 can extract the keywords for the text by analyzing the tag information, contents, etc. from the text.

For example, when the text obtained by the interaction information includes tag information, the keyword extracting unit 1010 may extract each tag information as a keyword, and may extract keywords by analyzing the contents of the text. As an example, if the content of the text is " Zapagateji also at the weekend, " the keyword extraction unit 1010 may extract 'weekend' and 'chapaeguti' as keywords of the text.

The keyword classification unit 1030 classifies the keywords extracted by the keyword extraction unit 1010. The keyword classification unit 1030 receives the keyword from the keyword extraction unit 1010 and classifies the keyword according to a predetermined classification criterion.

As an example of such a keyword classification, the keyword classification unit 1030 may use a part-of-speech classification. In this case, the keyword classification unit 1030 can classify the keywords extracted by the keyword extraction unit 1010 by using five to nine parts words represented by nouns, verbs, adjectives, adverbs, have.

Meanwhile, the chat information analysis unit 530 analyzes the chat information input by the user. The user inputs chat information including the intention of the user through the user terminal 150. At this time, the chat information including the intention of the user may be information that can be converted into a digital signal such as text, voice, video, sound, and image. The chat information analysis unit 530 further analyzes the chat information using a predetermined rule in order to grasp the intention of the user.

Here, an example of the predetermined rule may be the same as that performed by the keyword classifier 1030. This is because a keyword corresponding to a user's intention is acquired more easily by forming the same classification criteria as the classification criteria of the chat information. However, the predetermined rule of the present invention is not limited to this, and other analysis methods for grasping the intention of the user can be used.

The response information generation module 233 further confirms and provides the response information requested by the user by using the classified keywords and the analyzed chat information respectively acquired by the keyword processing unit 910 and the chat information analysis unit 530. [

The response information generation module 233 acquires the chronological expression selected by the chronological expression selection unit 550 and the chat information analyzed by the chat information analysis unit 530 to more accurately confirm the intention of the user. Thereafter, the response information generation module 233 extracts a keyword that is determined to be included in the user's intention, and generates response information to be provided to the user using the classified keyword.

At this time, the response information generation module 233 may use a method of selecting the interaction information having the highest number or ratio of keywords included in the user's intention, for example, as a method of generating response information.

FIG. 11 is a flowchart illustrating a step of analyzing chat information of a median expression analysis method in consideration of a user state according to an exemplary embodiment of the present invention.

Referring to FIG. 11, in step S623 of analyzing chat information among the methods for analyzing the jargon expression according to an exemplary embodiment of the present invention, it is further determined whether the acquired external information is user chat information (S1110) (S1121) extracting a keyword, classifying extracted keywords (S1131), analyzing user chat information (S1123), and selecting a keyword corresponding to the analyzed chat information (S1133).

First, if the external information obtained in step S610 is not the user state information in step S621, the apparatus for analyzing ambiguity further determines whether external information is user chat information (step S1110). At this time, the user chat information may be the information inputted by the user to store in the internal storage device, that is, not the interaction information, but the chat information inputted by the user for the chat with the chatbot, for example, the user intentionally input to the terminal.

That is, if the external information acquired in step S610 is not the user chat information, the meditative expression analyzing apparatus determines that the external information is the interaction information and extracts the keyword from the interaction information (step S1121).

In step S1121, the median expression analyzing apparatus analyzes the interaction information to extract the keywords included in the interaction information. At this time, the keyword may be formed to include the meaning of the interaction information.

In the case where the interaction information analyzed in step S1121 is an image, the barometric expression analyzing apparatus may include a keyword of an entire image, a keyword of an object in an image, a keyword of a person in an image, Can be extracted.

At this time, the apparatus for analyzing the Chinese characters can grasp the meaning of the image in order to acquire the keyword of the entire image. For example, when an image photographed inside a restaurant is acquired as the interaction information, the meditative expression analyzing device can extract the restaurant from the interaction information by keywords of the entire image.

In addition, the median expression analyzing apparatus may extract keywords of an image from an object in an image. When a chair is included in the image obtained by the interaction information, the apparatus for analyzing the Chinese characters can extract the chair included in the image as one of the keywords of the image. At this time, the keyword of the object in the image may be obtained from all things that can be identified in the image as much as the setting.

Further, the apparatus for analyzing the Chinese characters may extract a keyword of an image from a person in the image. When a person is included in the image obtained by the interaction information, the meditative expression analyzing apparatus analyzes the emotion through the expression of the person included in the image, and extracts the analyzed emotion as one of the keywords of the image. At this time, the keyword of the person in the image may be obtained from all the persons who are identifiable in the image as much as the setting.

In addition, the apparatus for analyzing the Chinese characters may extract keywords related to the atmosphere of the image. The median expression analyzer analyzes the color, saturation, and brightness of the image acquired by the interaction information. In this way, the apparatus for analyzing the Chinese characters can determine the atmosphere of the image, and the atmosphere of the image can be extracted as a keyword of the atmosphere of the image.

On the other hand, if the text information is included in the interaction information, the median expression analyzing apparatus may extract the keyword from the text by analyzing the text. At this time, the apparatus for analyzing the Chinese characters can extract keywords for the text by analyzing the tag information and contents from the text.

For example, when the tag information is included in the text acquired by the interaction information, the clumsy expression analysis apparatus can extract each tag information as a keyword, and extract the keyword by analyzing the contents of the text. As an example, if the content of the text is "Zapagateji" at the weekend, the Chinese character analysis device can extract "weekend" and "chiapagatei" as keywords of the text.

Next, the target expression analyzing apparatus that has extracted the keywords classifies the extracted keywords (step S1131).

The median expression analyzing apparatus can classify the keywords according to a predetermined classification criterion. As an example of such keyword classification, the phrase classification can be used. The median expression analyzing device can classify extracted keywords by using 5 words and 9 phrases expressed as nouns, verbs, adjectives, and adverbs as criteria for classifying keywords.

On the other hand, if the meditative expression analyzing apparatus confirms the external information as the user chat information in step S1110, the medullary expression analyzing apparatus grasps the intention of the user in the acquired user's chat information (step S1123).

At this time, the chat information including the intention of the user may be information that can be converted into a digital signal such as text, voice, video, sound, and image. The median expression analysis device analyzes the chat information using a predetermined rule in order to grasp the intention of the user.

One example of the predetermined rule may be the same as the method performed in step S1131. This is because a keyword corresponding to a user's intention is acquired more easily by forming the same classification criteria as the classification criteria of the chat information. However, the predetermined rule of the present invention is not limited to this, and other analysis methods for grasping the intention of the user can be used.

Next, the meditative expression analyzing apparatus selects a keyword corresponding to the analyzed chat information (step S1133). The median expression analyzing apparatus can select a keyword corresponding to the analyzed chat information using the result of analysis of the user chat information obtained in step S1123 and the result of classification of the extracted keyword obtained in step S1131.

Using the detailed operation sequence of step S623 shown in FIG. 11, in step S640 of FIG. 7, the median expression analyzing apparatus generates the response information using the determined meaning by using the keyword of the user chat information and the interaction information It is possible to provide the user with the response information required by the user with higher accuracy.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.

100: Chattbot System 110: Interaction Offering Set
130: Chatbot, a meditative expression analyzer that takes into account the user's state
150: user terminal 210: external information acquisition module
230: Median Expression Processing Module 231: Median Expression Analysis Module
233: response information generation module 310: chat information acquisition module
330: User state determination module 410:
430: measurement information analysis unit 450: user state setting unit
510: User state storage unit 530: Chat information analysis unit
550: negative selection part 570: accuracy learning part
910: Keyword processing unit 1010: Keyword extraction unit
1030: Keyword classification section

Claims (18)

An external information acquisition module for acquiring status information of the user and chat information of the user as external information; And
And a median expression processing module for generating a median expression analysis result using the external information and generating response information to be provided to the user using the median expression analysis result,
Wherein the external information acquisition module further acquires the interaction information of the user and the interaction information includes at least one of activity information performed by the user using the communication network with another user and information previously stored in the internal storage device by the user An apparatus for analyzing mediocre expressions that considers one user state. The method according to claim 1,
Wherein the external information acquisition module comprises:
A chat information acquiring module for acquiring the chat information input by the user; And
And a user state determination module for obtaining at least one state of the user's body information and surrounding environment information to determine the state of the user. 3. The method of claim 2,
Wherein the user state determination module comprises:
A measuring unit for obtaining at least one of information on the user's body and surrounding environment;
A measurement information analyzer for analyzing the external information acquired by the measurement unit and generating at least one analysis result of the user's state information; And
And a user state setting unit for setting the state of the user using at least one of the information analysis results. The method according to claim 1,
The median expression processing module comprises:
A middle expression analyzing module for generating an analysis result of the ambiguous expression included in the chat information; And
And a response information generation module for generating the response information to be provided to the user using the result of the heuristic expression analysis. 5. The method of claim 4,
Wherein the heuristic expression analysis module comprises:
A user state storage unit for storing the state of the user among the external information acquired by the external information acquisition module;
A chat information analyzer for analyzing chat information of the user among the external information acquired by the external information acquisition module; And
A user state, and chat information to determine the meaning of the ambiguous expression, and to select a judgment result. 6. The method of claim 5,
Wherein the heuristic expression analysis module comprises:
And an accuracy learning unit that further acquires the response information and the chat information of the user with respect to the response information, and accumulates the accuracy of the result of determination of the ambiguous expression, based on the user state Median expression analysis device. delete The method according to claim 1,
Wherein the heuristic expression analysis module comprises:
And a keyword processing module for extracting a keyword from the interaction information and classifying the extracted keyword according to a predetermined criterion, wherein the response information to be provided to the user is further generated using the keyword and the chat information A device for analyzing mediocre expression considering user state. 9. The method of claim 8,
The keyword processing module includes:
A keyword extracting unit for analyzing the interaction information and extracting a keyword including the meaning of the interaction information; And
And a keyword classifier for classifying the keywords extracted by the keyword extractor according to a predetermined rule. A method of using a median expression analyzing device,
Obtaining status information of the user and chat information of the user as external information;
Analyzing the external information according to the type of external information;
Determining and selecting a meaning of the chronological expression of the chat information using the external information analysis result; And
And generating response information to be provided to the user using the determined meaning,
The step of generating the response information may further include accumulating the accuracy of the semantic determination result of the ambiguous expression using the additional chat information of the user after generating the response information and providing the response information to the user A method for analyzing mediating expressions considering user state. 11. The method of claim 10,
Wherein the external information includes at least one of status information of the user's body information and surrounding environment information, and user information that is chat information input by the user. 12. The method of claim 11,
Wherein the step of acquiring the external information includes setting a state of the user using at least one state information of the user's body information and the surrounding environment information, . 11. The method of claim 10,
Wherein the step of generating the response information includes generating a response information corresponding to the user's intention using the meaning of the user state and the heuristic expression. delete 11. The method of claim 10,
If the additional chat information of the user is positive, accumulating the accuracy and providing response information using another substitute expression if the additional chat information is negative or non-existent. A method for analyzing ambiguous expressions. 11. The method of claim 10,
Wherein the external information further includes interaction information of the user and the interaction information includes at least one of activity information that the user performs using the communication network with another user and information that the user has previously stored in the internal storage device, A method of analysis of mediocre expression considering state. 17. The method of claim 16,
Wherein the analyzing the external information comprises:
Extracting a keyword from the interaction information when the external information is the interaction information, and classifying the extracted keyword according to a predetermined criterion. 18. The method of claim 17,
Wherein the step of generating the response information comprises:
And generating the response information to be provided to the user by further using the keyword and the chat information.

Images